PDA

View Full Version : Anyone Selling XT-IDE Cards?



Pages : [1] 2

tempest
April 29th, 2016, 06:54 AM
Is anyone making and selling XT-IDE cards? Making one myself is beyond my skills, and the tiny 20MB hard drive in my 5150 isn't nearly big enough. I see a lot of people discussing the card and I've found parts lists, but I don't see anyone actually selling them (perhaps there's a reason for this?).

glitch
April 29th, 2016, 07:25 AM
I've got a few left on hand, I'll PM you about it.

Posted this publicly so that future searchers for XT-IDE will find it! I have bare rev 2 boards on hand, and can assemble + test as well.

acadiel
April 29th, 2016, 09:50 AM
Same here. I'm debating about getting one myself.

Half-Saint
April 29th, 2016, 11:20 AM
There's also a guy (CarlosTex) over at AmiBay making/selling them.

glitch
April 29th, 2016, 12:18 PM
There's also a guy (CarlosTex) over at AmiBay making/selling them.

Isn't he selling the XT-CF rather than the full XT-IDE?

GrizzlyAdams
April 29th, 2016, 05:04 PM
I'd love one, either bag of parts or fully assembled.


I've got a few left on hand, I'll PM you about it.

Posted this publicly so that future searchers for XT-IDE will find it! I have bare rev 2 boards on hand, and can assemble + test as well.

glitch
April 29th, 2016, 05:27 PM
OK, so I don't have enough for everyone who wants one, which means it's time to do another XT-IDE run! Now, as someone who's personally assembled probably 20-25 XT-IDE boards, including some from the original run, I have a few things I'd like to change and come up with a Rev 3 board. Some of them are minor technical details (resizing pin header holes to better accommodate the detent headers I use, for example), but some are function changes:

- Return to DIP switches, much nicer than jumpers in my opinion
- Removal of UART circuit (not one board I've assembled was requested to have it)
- Removal of 4-pin Molex power connector, everyone is using CF or industrial Flash modules
- Permanent "Chuck mod"
- Switch to 74xx373 latches (more common in hobbyist parts bins, IMO)

Does anyone see a problem with any of these mods? Obviously the Rev 2 design could still be run, and I can run it as well if people really want the Rev 2 unmodified.

Additionally, I'll be optimizing the board for size, which should get the cost down a little per-board. I should still be able to do hard gold plating on the edge connectors. I'll try and put as much documentation on the silkscreen as possible. I'd like to stick with the "All TTL" design to make it easier for future builders to get an XT-IDE up and going without having to worry about programming GALs and such. I see no reason to switch to surface mount parts since the ISA card edge and 40-pin connector constrain the size of the board anyway.

Moderators, does someone want to rename this thread "XT-IDE Version 3"?

tempest
April 29th, 2016, 05:43 PM
What's the Chuck Mod?

EDIT: Looks like I can't change the thread title after all.

glitch
April 29th, 2016, 05:47 PM
What's the Chuck Mod?

ChuckG figured out that two of the ports could be put on consecutive I/O addresses and accessed with a single 16-bit instruction rather than two 8-bit instructions. This makes it much faster.

jharre
April 29th, 2016, 08:03 PM
FYI, there's still a bunch of the blank boards available from Todd Goodman. According to the board inventory at https://www.retrobrewcomputers.org/doku.php?id=boardinventory (scroll down to "Other (formerly known as "Mini") Boards"), he has 24 on hand. They are $12 each plus S&H.

<*> Jim

glitch
April 29th, 2016, 08:39 PM
Huh, didn't know about that list. I'm still going to go ahead with XT-IDE rev 3 -- in looking through the schematic, I found what I think are some pretty good improvements to make. It'll be much easier to configure!

I noticed the upper six I/O address bits aren't decoded. From what I understand, this is "sort of" OK on old ISA systems, but has the potential to cause address conflicts due to the card repeating through address space. Seems like the card really should decode them for maximum compatibility and minimum headache. Probably end up doing it with diode-OR since I'm trying to stay away from a GAL. I'll start a separate thread about it.

Half-Saint
April 29th, 2016, 09:56 PM
Isn't he selling the XT-CF rather than the full XT-IDE?

Yes, he is. However, the board is using a standard IDE connector so any ATA hard drive should work.

Malc
April 29th, 2016, 11:45 PM
Yes, he is. However, the board is using a standard IDE connector so any ATA hard drive should work.

It looks like he's selling the Lo-Tech XT-CF V3, So it's suitable for CF but not 'Any ATA hard drive', Only ATA-2 compliant hard drives.

glitch
April 30th, 2016, 07:12 AM
Yeah, no latches for the upper byte. I suppose 8-bit IDE drives would also work, but don't expect usual 16-bit IDE drives to work unless, as Malc said, it specifically supports it. I didn't know it was in the ATA-2 spec.

glitch
April 30th, 2016, 11:48 AM
Does anyone know why D7 (pin 3) on the IDE bus is pulled down via 10K resistor to ground? None of the other data bus lines are pulled down, I can't find a requirement for it in the IDE spec...doesn't make sense to me. It's also pulled down on the XT-CF Lite board, according to its schematic.

Robin4
April 30th, 2016, 05:38 PM
iam also intrested in these XT-IDE controllers. Iam not intrested in the XT-CF versions.

smyke
May 2nd, 2016, 05:29 AM
I am also interested in 1 or 2 of them. Isn't it possible to utilize a CF card with an IDE adapter if you went with just the XT-IDE adapter?

glitch
May 2nd, 2016, 06:13 AM
Isn't it possible to utilize a CF card with an IDE adapter if you went with just the XT-IDE adapter?

Yes, I'd guess this or industrial Flash modules (DOMs) are the most popular uses for XT-IDEs.

Quick update, I've almost finished with the new layout. Probably order a prototype run this week.

glitch
May 2nd, 2016, 08:49 AM
I think the prototype layout is finished:

http://i.imgur.com/aPvmPxP.png

Changes:

- No UART
- Removed some extraneous pullup/pulldown resistors
- Sense is now ON == 1 for the config switches
- ROM and I/O port options all switch configured
- Decoding an extra bit on the I/O port, gives more possible port options
- Single LED/connector option
- Switch/Jumper descriptions added to silkscreen
- Jumper header hole size adjusted for detent pin headers
- Simplified settings for 28C64B/28C256 EEPROM selection
- Swapped status/control latch for buffer (latch was always enabled, it was being used as a buffer)
- Removed silkscreen around ISA connector and screw holes
- All component values placed inside part legends, values placed outside
- Capacitors optimized for axial bead type capacitors

I ended up leaving the jumpers for the Chuck Mod option rather than permanent Chuck Mod, since the V1 BIOS doesn't seem to support Chuck Mod. I also left the 74LS537 latches as apparently I ordered a quantity of them a while back, so I've got plenty for kits. The board is somewhat smaller than the rev 2 boards. I also left the I/O decoder alone w.r.t the upper 6 I/O address bits.

glitch
May 2nd, 2016, 09:08 AM
As per RetroHacker_'s request on #vc IRC, J1 has been moved so it's not possible to accidentally put the IDE connector in its position during assembly!

prime
May 2nd, 2016, 09:22 AM
As per RetroHacker_'s request on #vc IRC, J1 has been moved so it's not possible to accidentally put the IDE connector in its position during assembly!

Also prevents J1 being blocked if you decide to fit a boxed 40 pin connector for the IDE.

Cheers.

Phill.

glitch
May 2nd, 2016, 11:51 AM
Yup, much better to move it to the new position -- but that's one of the great things about an open dev process, this was caught before I even ordered prototypes.

I also caught a clearance issue with the card bracket and the 74LS138, so everything in that area got nudged left and rerouted:

http://i.imgur.com/moqQ2Ub.png

lyonadmiral
May 2nd, 2016, 03:06 PM
Absolutely amazing work. :)

prime
May 2nd, 2016, 10:13 PM
Would probably also be useful to have a jumper so you could have the LED on the front of the case.

Cheers.

Phill.

glitch
May 3rd, 2016, 04:35 AM
Would probably also be useful to have a jumper so you could have the LED on the front of the case.

The intent is to replace the LED with a jumper header when this is desired. I'll probably ship a Molex KK-100 2-pin connector in the parts kit so the assembler can choose.

glitch
May 3rd, 2016, 07:17 PM
Preorder form here:

http://www.glitchwrks.com/xt-ide/preorder

rosaage
May 4th, 2016, 03:48 AM
Is the kit parts and pcb or parts only? Do you have any idea what the price would be, somewhere around 10-30$ or more?

glitch
May 4th, 2016, 04:48 AM
Is the kit parts and pcb or parts only? Do you have any idea what the price would be, somewhere around 10-30$ or more?

Kit does not include PC board; however, if I find preorders with only a kit value filled in, I'll assume they wanted an equal number of boards too.

Bare board should cost between $10-15, parts kit will probably be $25-30. There's no financial commitment with the preorder, strictly for size estimating purposes.

hargle
May 4th, 2016, 05:36 AM
- Decoding an extra bit on the I/O port, gives more possible port options

I believe this change would require a modification to the BIOS and/or xtidecfg to allow for the additional IO options. Are the universal BIOS folks aware of this?

Love to see the XT-IDE project living on!

compu_85
May 4th, 2016, 07:23 AM
Sweet! I just submitted my preorder.

pearce_jj
May 4th, 2016, 01:57 PM
Re D7 PD, this is for device detection. CSEL should be tied directly to ground (sorry if this is already caught). Nice PCB layout (and colour!).

There is a Lo-tech full IDE PCB too btw.

Robin4
May 4th, 2016, 05:29 PM
Preorder form here:

http://www.glitchwrks.com/xt-ide/preorder


I have a few question about your XT-ide board..

1. Is it possible to get this PCB in a different color ( i would prefer green instead)

2. Is it possible to edit /move some of the tekst on the PCB.

3. Like to know if you can make the kit somewhat cheaper to only provide the parts that are hard to get (seems that the IC chips easilly could be bought by the buyer on ebay or so. (like the bracket)

4. Which ROM / bios are you going to use?

5. Why is interrupt 2 right behind interrupt 7 ? and not 2,3,5,7


I really have intrest to order 5 boards, but before that i really would like see a working example. Maybe i could buy a bigger lot from you.. (20 and or 30 pieces) but i really need think about that.

pearce_jj
May 5th, 2016, 01:21 AM
Interrupts are not used in the XTIDE Universal BIOS anyway.

glitch
May 5th, 2016, 05:12 AM
1. Is it possible to get this PCB in a different color ( i would prefer green instead)

The purple is just a prototype rendering from OSH Park. The final boards will be deep cyan blue. You are of course welcome to run them in a color of your choosing, but I'll be doing a single run of a single color to keep costs down.


2. Is it possible to edit /move some of the tekst on the PCB.

Yes, but I've placed the current text where it is to try and keep the board self-documenting. What do you think should be moved?


3. Like to know if you can make the kit somewhat cheaper to only provide the parts that are hard to get (seems that the IC chips easilly could be bought by the buyer on ebay or so. (like the bracket)

The short answer to this is, "probably no." I will likely pull parts for a number of kits and bag them up. Splitting out parts and figuring out what the remaining kit bits should cost is more work for me. Besides, I doubt the average person, buying quantities of 2-3 ICs at a time, will be able to beat my kit costs if purchasing from reputable suppliers (i.e. actual parts suppliers, not some random guy on eBay).


4. Which ROM / bios are you going to use?

Assembled boards will probably ship with the 1.x series Universal BIOS. I will be using Atmel 28C64B 8K EEPROMs unless there's a lot of demand for 28C256. The board can of course use any of the available BIOSes since I've left the jumpers for the Chuck Mod on.


5. Why is interrupt 2 right behind interrupt 7 ? and not 2,3,5,7

Because that was the easiest way to lay it out. If this actually bothers people I'm sure I can put them in order. As pearce_jj said, interrupts aren't even used on this board.


I really have intrest to order 5 boards, but before that i really would like see a working example. Maybe i could buy a bigger lot from you.. (20 and or 30 pieces) but i really need think about that.

And before any one individual decides they want 20 or 30, they'll need to put cash up front! I got stuck with a huge surplus of various expensive components (anyone need Eurobus connectors?) after a previous failed kit project, wherein many people said they wanted kits, but no one wanted to pay for them when I'd gotten all of the components together. Lesson learned.

glitch
May 5th, 2016, 05:15 AM
Re D7 PD, this is for device detection. CSEL should be tied directly to ground (sorry if this is already caught). Nice PCB layout (and colour!).

There is a Lo-tech full IDE PCB too btw.

Thanks for the clarification! I'd figured out CSEL needed to be tied to ground, but I'd left the pull-down on D7 just because I wasn't 100% sure about its function.

Robin4
May 5th, 2016, 06:25 AM
Isnt a production run on a green pcb not cheaper then deep cyan blue, because green was the basic color back then and was used for a lot of ISA card.. Deep cyan blue would be a more special color to me so it would be more expensiver.

At question 3: I presume that you going to sell a full kit then? So the buyer dont have to look after the right parts.

Why are there interrupt settings on the board if the universal bios dont use them at all? And does this adapter then works faster without these interrupt settings? (faster transfer rate)

pearce_jj
May 5th, 2016, 08:00 AM
The IRQs can be removed, all the BIOS does is request a transfer then poll the status bit, then rep INSW to transfer the sector (with chuck mod). Writes are slower with this MUX design, of course writes are less frequent in practice.

T-R-A
May 5th, 2016, 03:01 PM
Isnt a production run on a green pcb not cheaper then deep cyan blue...

Not necessarily. Individual board houses (mostly all in China now) base their raw PCB prices very close, irregardless of silk-screen color. Personally I'd prefer any blue board over a green one since soldering issues are far easier to inspect on a blue board than a green one.

Trixter
May 5th, 2016, 04:40 PM
then rep INSW to transfer the sector (with chuck mod)

It just dawned on me, this won't work on dodgy systems like the AT&T PC 6300. If the "Chuck Mod" is permanent, then this board has no chance of working in those systems, does it?

glitch
May 5th, 2016, 05:43 PM
Isnt a production run on a green pcb not cheaper then deep cyan blue, because green was the basic color back then and was used for a lot of ISA card.. Deep cyan blue would be a more special color to me so it would be more expensiver.

With the board house I use, solder mask costs the same regardless of color. The fact that the board is an ISA card has no bearing on its cost, other than the added expense of having the edge connector plated with hard gold.

Not to be flippant with anyone, but I do have extensive experience in managing small production runs of PC boards. It's part of what I do for a living. I'd like to think I'm decent at it!


At question 3: I presume that you going to sell a full kit then? So the buyer dont have to look after the right parts.

Yes, there will be a full kit, and I'll probably take the time to write an assembly manual, or at least do a writeup on my assembly of an example board.


Why are there interrupt settings on the board if the universal bios dont use them at all? And does this adapter then works faster without these interrupt settings? (faster transfer rate)

They exist because someone might want to use them in the future, and they're not hurting anything by being there but not being used. I see no reason to remove functionality and limit potential


Personally I'd prefer any blue board over a green one since soldering issues are far easier to inspect on a blue board than a green one.

I also find the contrast to be better with blue boards, as long as they're not too dark. The board house I'm working with has a very nice deep cyan which really makes the traces stand out. Plus I like to think it differentiates my projects a little bit :)


It just dawned on me, this won't work on dodgy systems like the AT&T PC 6300. If the "Chuck Mod" is permanent, then this board has no chance of working in those systems, does it?

This and BIOS compatibility is why I chose to leave the Chuck Mod as a jumperable option.

glitch
May 5th, 2016, 05:57 PM
In case anyone is wondering, this is the deep cyan I'm using:

http://i.imgur.com/g31Tioj.jpg

This is of course the prototype base board for the 20 mA current loop converter project.

compu_85
May 6th, 2016, 07:51 AM
Would it be possible to change the PCB so it's more conducive for use as a hard card? I'd be using this in a Compaq Portable and would prefer to mount the HDD in the expansion slot area, keeping both original floppies.

Perhaps the easiest way would be to have an option for hardcard buyers... make the PCB full length and add mounting holes in the long section?

-J

glitch
May 6th, 2016, 08:06 AM
Would it be possible to change the PCB so it's more conducive for use as a hard card? I'd be using this in a Compaq Portable and would prefer to mount the HDD in the expansion slot area, keeping both original floppies.

Perhaps the easiest way would be to have an option for hardcard buyers... make the PCB full length and add mounting holes in the long section?

See this writeup: http://www.glitchwrks.com/2010/12/16/xtide

Adding a bunch of blank space for mounting a hard drive would add a lot of cost to the board. Using an industrial Flash module keeps the board compact, draws significantly less power (important for systems like the IBM 5150 PC with its original low wattage supply), and then you don't have to worry about a mechanical hard disk, especially in a portable machine. Personally I use an XT-IDE with industrial Flash module in my Leading Edge Model D, which lets me keep both floppy drives and not worry about overloading its supply.

I'll be selling various capacities of industrial Flash modules as an option for these, too.

Krille
May 6th, 2016, 11:16 AM
It just dawned on me, this won't work on dodgy systems like the AT&T PC 6300. If the "Chuck Mod" is permanent, then this board has no chance of working in those systems, does it?

Correct. You will have to use it as a rev 1 card (i.e. compatible mode). However, it is possible to add yet another transfer mode to the BIOS but it wouldn't be much faster than using the existing compatible mode. Also, this additional transfer mode would really only be needed if you absolutely must use a normal harddrive since, when using CF media, you should be able to use other types of controllers such as the XT-CF or even a regular 16-bit IDE controller. And of course, with memory mapped I/O controllers such as the ADP50 there is no problem at all regardless of drive type.

glitch
May 8th, 2016, 05:28 PM
Final design, as submitted to the manufacturer:

http://i.imgur.com/74HieR6.png

Final design changes:

- Extra LED output using a spare inverter from the 74LS04
- Few component value tweaks
- Added layer for hard gold plating on edge connector

Should have prototype boards in hand in about a week.

bobba84
May 8th, 2016, 05:48 PM
Very exciting!!

prime
May 13th, 2016, 03:56 AM
Glitch

Are the design files for the board available from anywhere I'd like to take a look at them if possible.

Cheers.

Phill.

glitch
May 13th, 2016, 04:11 AM
Are the design files for the board available from anywhere I'd like to take a look at them if possible.

The full project will be hosted on GitHub, I just need to set up a repository and get the file structure cleaned up.

Prototype boards should be in today!

glitch
May 13th, 2016, 06:09 PM
Boards arrived today!

http://i.imgur.com/6K2jnpa.jpg

One slight problem -- KiCad apparently changed cathode designations on diodes and LEDs between the version that the rev 2 board schematic was created with, and the new KiCad libs. So the LED is backwards on the silkscreen. I also eliminated a discrete resistor by using a spare from one of the resistor packs. Other than that, it works!

As you can see from the picture, the board house I do prototype runs with wasn't able to do selective hard gold for the edge connector. I'm getting quotes from a few of the production shops I've used in the past, including PCBCart, which I believe is who did the rev 2 run. I'll probably remove the unused ISA fingers to save on plating costs.

compu_85
May 13th, 2016, 07:01 PM
Sweet! :thumbsup:

glitch
May 14th, 2016, 09:33 AM
GitHub repo created/current work pushed:

https://github.com/glitchwrks/xt_ide

Wouldn't recommend making boards from those Gerbers yet. There were no authors/contributors listed on the old KiCad project from the rev 2 board, which is what I started with. If you worked on it and would like your name added, please contact me, or send a pull request.

CarlosTex
May 15th, 2016, 02:56 PM
Isn't he selling the XT-CF rather than the full XT-IDE?


That is correct. I'm selling the XT-IDE CF rev 3 cards from lo tech.

I have however just 2 days ago built the full XT-IDE version from lo tech for myself:

3123531236

I can say that the PCB's are absolutely top notch. If someone is interested in the lo tech version of the full XT-IDE (https://www.lo-tech.co.uk/wiki/Lo-tech_8-bit_IDE_Adapter_rev.2) i can build a small batch of them. Price of course would be a little higher than XT-IDE CF due to the component count being almost triple and higher PCB cost too.

Not to too my own horn but unlike sellers on eBay (http://www.ebay.com/itm/ISA-Compact-Flash-8-Bit-IDE-Hard-Drive-CF-Card-XT-IDE-Vintage-PCs-Tandy-/172203023597?hash=item2818199ced:g:GikAAOSwfZhXM3y E) i don't cut any corners when making my cards. All functionality is to be included by installing even the optional parts. I believe that the functionality provided by the extra components (activity LED, slot 8, ROM address space selection) are really important. I like to use my XT-IDE cards on slot 8, as it allows me to save other slot for other stuff that won't work on slot 8.

glitch
May 15th, 2016, 03:46 PM
I like to use my XT-IDE cards on slot 8, as it allows me to save other slot for other stuff that won't work on slot 8.

That's probably something worthwhile to add to the XT-IDE rev 3. I'll see what that will take, tack it onto the current proto boards, and work it into the final rev 3 board.

Stone
May 15th, 2016, 04:04 PM
If someone is interested in the lo tech version of the full XT-IDE (https://www.lo-tech.co.uk/wiki/Lo-tech_8-bit_IDE_Adapter_rev.2) i can build a small batch of them. Price of course would be a little higher than XT-IDE CF due to the component count being almost triple and higher PCB cost too.Can you provide an estimate of the actual price for your finished card? I'm interested but I can't assemble one myself.

I asked glitch last week but he failed to respond.

lyonadmiral
May 15th, 2016, 04:48 PM
Can you provide an estimate of the actual price for your finished card? I'm interested but I can't assemble one myself.

I asked glitch last week but he failed to respond.

I do not want to speak out of turn, but in an e-mail he sent me (messages have been falling through the cracks, so I don't believe he is intentionally being ignorant of requests) that the estimated cost of a fully assembled card was approximately $100 U.S.

Stone
May 15th, 2016, 05:11 PM
I do not want to speak out of turn, but in an e-mail he sent me (messages have been falling through the cracks, so I don't believe he is intentionally being ignorant of requests) that the estimated cost of a fully assembled card was approximately $100 U.S.Thanks, Daniel.

BTW, I made the request in a thread he had posted in, not in an email or PM:

http://www.vcfed.org/forum/showthread.php?52487-IDE-CF-adapter&p=413088#post413088

Maybe he missed that post.

glitch
May 15th, 2016, 05:16 PM
I asked glitch last week but he failed to respond.

Sorry, been super busy with this project, as well as day-job work! I must've missed your question. Typically in the past I've sold the rev 2 boards for $100 assembled or $85 if the buyer didn't want the UART bits. I'm hoping to offer the rev 3 to *hobbyists* at $85 each, assembled, but it will depend on what kitting up actually costs. I'd previously been running from stocks of ICs that were basically surplus from other day-job billable work so I don't have a solid figure on what the parts kits will cost right this moment.

Assembled of course really means "assembled and tested."

Stone
May 15th, 2016, 05:27 PM
You say UART. Did this have a serial port as well?

glitch
May 15th, 2016, 05:32 PM
You say UART. Did this have a serial port as well?

The rev 2 board had a built-in UART for "bare metal" booting. I've had very few (none?) people request it be populated. As such, I removed it from the rev 3 board entirely. The UART was not usable as a COM port under DOS, strictly for booting. You could've used it with customized software, of course.

pearce_jj
May 15th, 2016, 11:17 PM
That's probably something worthwhile to add to the XT-IDE rev 3. I'll see what that will take, tack it onto the current proto boards, and work it into the final rev 3 board.

All this and the bus MUX timing has been solved already by my board which is available already.

The nice feature of the XTIDE R2 project is the serial boot, which can be used for file transfer between machines as well.

Krille
May 16th, 2016, 12:54 AM
The rev 2 board had a built-in UART for "bare metal" booting. I've had very few (none?) people request it be populated. As such, I removed it from the rev 3 board entirely. The UART was not usable as a COM port under DOS, strictly for booting. You could've used it with customized software, of course.

I'm not sure why you would need customized software? I know at least one person that used it with a serial mouse.

glitch
May 16th, 2016, 04:32 AM
EDIT Totally wrong about the lack of interrupt capability, rechecked the schematic and it does in fact have an interrupt jumper block.

I suppose you could make the UART appear at the proper I/O port for DOS COM port compatibility, but it has no interrupt capability, so it would only work with software doing polled I/O. I don't know what percentage of software does polled COM port I/O.

glitch
May 16th, 2016, 05:01 AM
All this and the bus MUX timing has been solved already by my board which is available already.

I seem to remember the Lo-Tech XT-IDE also uses a fixed I/O port address of 0x300, which for me is a pretty serious restriction.

I think it's great that people have multiple options for open-source hardware!

Great Hierophant
May 16th, 2016, 06:13 AM
I suppose you could make the UART appear at the proper I/O port for DOS COM port compatibility, but it has no interrupt capability, so it would only work with software doing polled I/O. I don't know what percentage of software does polled COM port I/O.

I would suggest that booting programs would have a very hard time with a polled-only UART, but few such programs support a mouse anyway. When running a DOS program, I would think it is a function of the mouse driver to provide its information, whether by polled I/O or via interrupts.

As far as the Serial UART goes, I believe most people find other ways to format a hard drive if they have no working floppy drive. I believe it is possible to format a CF drive, for example, and make it DOS bootable by a CF reader attached to a modern computer. It may be possible to do this even with DOSBox.

Krille
May 16th, 2016, 07:43 AM
For reference, the person using a serial mouse was Sziklai and the discussion starts here (http://www.vcfed.org/forum/showthread.php?19591-XTIDE-tech-support-thread&p=378665#post378665). He later confirmed it working here (http://www.vcfed.org/forum/showthread.php?19591-XTIDE-tech-support-thread&p=381612#post381612).

Personally, I don't understand why people don't want the serial port. Sure, the functionality is still available as long as there is a serial port in the system but having a high-speed port for a small additional cost is a no-brainer to me. I suppose it's one of those things people don't know they need/want until they have actually seen it in use. :)

glitch
May 16th, 2016, 08:18 AM
Oops, I was mistaken about the UART not having interrupt capability! I took a look at the rev 2 schematic and it does in fact have a jumper block for it. I was probably thinking of the nonstandard crystal throwing it off for DOS. Of course you can change that out, as mentioned.

I can't speak for everyone else, but I've never had a need for it. AFAIK the server-side software is a Windows application, so that sort of rules it out for me anyway (no Windows machines). If you solder down your ICs instead of using sockets everywhere, you're stuck with the UART if you solder down its components -- can't be disabled without cutting, even if you pull the UART the bus driver still responds on reads so you have to just map it onto an unused I/O address, consuming more I/O space.

I could see it being useful if you were stuck with a PC and no way to create floppies or write to the drive you were using with the XT-IDE, and your machine didn't have a serial port, but how many use cases does that cover? The board layout and routing are cleaner without it. I would think most people without a way to write floppies would need a USB -> RS232 converter anyway, so why not use a CF card + USB reader?

compu_85
May 16th, 2016, 08:53 AM
Having a 2nd serial port for a mouse would be something I'd like for my Portable, but I can easily live without it.

Krille
May 17th, 2016, 03:41 AM
AFAIK the server-side software is a Windows application, so that sort of rules it out for me anyway (no Windows machines).
There is also a Linux port available on GitHub. See this (http://www.insentricity.com/a.cl/244/adding-a-hard-drive-to-an-original-ibm-pc-using-a-raspberry-pi). Additional ports of the server-side software for other operating systems are more than welcome.


If you solder down your ICs instead of using sockets everywhere, you're stuck with the UART if you solder down its components -- can't be disabled without cutting, even if you pull the UART the bus driver still responds on reads so you have to just map it onto an unused I/O address, consuming more I/O space.

As I see it, everything should be socketed anyway on hobbyist boards like these. But don't get me wrong, I'm not arguing whether you should have the serial port ciruitry or not on this board. That's completely up to you. I'm just surprised that so few people find it useful, especially when there are lots of threads on here about problems booting these old machines due to all kinds of issues with floppies and floppy drives/controllers etc.

glitch
May 18th, 2016, 06:24 PM
Starting to order parts through my various IC supplier sources. I'm hoping I can get the parts kit in under $20!

pearce_jj
May 19th, 2016, 03:46 AM
I seem to remember the Lo-Tech XT-IDE also uses a fixed I/O port address of 0x300, which for me is a pretty serious restriction.

It is the first time this has been raised. There is probably just about enough board space to add a selection jumper to provide another option.

glitch
May 19th, 2016, 05:00 AM
It is the first time this has been raised. There is probably just about enough board space to add a selection jumper to provide another option.

I like having the full address select option that the original XT-IDE boards provide -- it lets you map around a system already full of hardware, when you might not be able to move anything else around. A fixed I/O port is probably not an issue for many people, especially if you're building up a system and would just rejumper other cards in the system.

Just to be clear, I think the Lo-Tech XT-IDE and XT-CF are great boards, especially since the size has been reduced to the bare minimum, but I also think there's merit in keeping the original XT-IDE in development. I can't imagine it *hurts* the hobbyist community to have more options available.

CarlosTex
May 23rd, 2016, 04:20 AM
Just to be clear, I think the Lo-Tech XT-IDE and XT-CF are great boards, especially since the size has been reduced to the bare minimum, but I also think there's merit in keeping the original XT-IDE in development. I can't imagine it *hurts* the hobbyist community to have more options available.

Of course, the more options the better. And yes Lo-tech stuff is awesome. I recently built the full XT-IDE and it works great. The performance is great too. There's also a big plus: James is a very helpful person. He will ALWAYS respond and try to help you if you find any problems. He is also very open to new ideas and his shop is getting bigger with lots of options. I'm very excited about a new line of products in his shop. 8-)

BTW, there is great stuff happening from Vogons forums members too. Besides the Inovation SSI-2001 replica that was an absolute success, there's also the Covox Sound Master and ARGUS (GUS Pnp enhanced clone). I believe Tronix from here is also working on a Creative Game Blaster Clone. Then there's also a favorite of mine, the Tandy Sound adapter.

glitch
May 23rd, 2016, 09:52 AM
Parts are starting to arrive! The majority are from US suppliers, but the AT28C64B EEPROMs are coming from Taiwan as they were significantly cheaper.

Looks like there will be no XT Slot 8 support -- I pulled out my XT over the weekend, and there's insufficient clearance with the shortest Flash module I have. Extending the board a little bit to allow the addition of another IC would probably make it too big for XT Slot 8 anwyay.

lyonadmiral
May 23rd, 2016, 10:44 AM
Parts are starting to arrive! The majority are from US suppliers, but the AT28C64B EEPROMs are coming from Taiwan as they were significantly cheaper.

Looks like there will be no XT Slot 8 support -- I pulled out my XT over the weekend, and there's insufficient clearance with the shortest Flash module I have. Extending the board a little bit to allow the addition of another IC would probably make it too big for XT Slot 8 anwyay.

What if you used an IDE cable to flex back around and then connect the CF to the cable?

glitch
May 23rd, 2016, 10:49 AM
What if you used an IDE cable to flex back around and then connect the CF to the cable?

That would work with the current board length, but adding a little length to accommodate Slot 8 support would probably make the board too long even for a cable. There's not a whole lot of room there, which the vertical connector of the Lo-Tech boards get around. I guess if you absolutely positively must have slot 8 support and want to use a XT-IDE rev 3, you can "dead bug" the circuit on top. I can't imagine there's a ton of demand for it, or situations where the Lo-Tech board wouldn't work.

jharre
May 24th, 2016, 02:47 PM
Playing catch-up reading here after vacation. Your prototype card looks nice.

I could be wrong, but didn't Andrew request that people stop using his ham call-sign "N8VEM" on boards?

<*> Jim

glitch
May 24th, 2016, 03:01 PM
I could be wrong, but didn't Andrew request that people stop using his ham call-sign "N8VEM" on boards?

You are indeed correct, and it has been removed in the current layout (master on GitHub). I was unaware of this, but Mike Brutman pointed it out to me in IRC when I posted a picture of the prototype board. The prototype run did not have gold plated edge connectors, so they won't be used in any of the kits or assembled units.

lyonadmiral
May 24th, 2016, 03:38 PM
Playing catch-up reading here after vacation. Your prototype card looks nice.

I could be wrong, but didn't Andrew request that people stop using his ham call-sign "N8VEM" on boards?

<*> Jim

I've been on Mars the last few weeks, can someone fill me in on the significance on N8VEM?

glitch
May 24th, 2016, 03:54 PM
I've been on Mars the last few weeks, can someone fill me in on the significance on N8VEM?

Andrew Lynch, N8VEM, did a lot of organizational work with the first XT-IDE community run. He's helped on a lot of other open source community hardware projects in the past. IIRC the thought about putting his callsign on boards was that it would let people find their source even if websites went down, et c. He's no longer active with these projects.

glitch
May 25th, 2016, 04:30 PM
Noticed a spacing error on the board -- hole spacing for the card bracket was off! Looks like it was off on rev 1 and rev 2 as well -- rev 2 compensated with larger holes :)

I added a test point for the /CARDSEL (bus pin B8 ) signal in case anyone wants to dead bug XT Slot 8 support on there. The manual will probably include instructions on doing so.

Removed unused edge connector pads to reduce the cost of hard gold plating. Submitted the Gerber files to Advanced Circuits to see if they can come close to the Chinese. I'd like to have the board made in the US if possible, but I also recognize that keeping costs low is kinda important for hobbyist projects!

compu_85
May 25th, 2016, 08:07 PM
If the cost difference isn't too much I'd pay more for a USA made card :)

glitch
May 27th, 2016, 09:18 AM
I've been talking with the Advanced Circuits sales rep for my area, it appears they can do $15/board + shipping and e-test, but that's with a four-week lead time. PCBCart will be more like $10/board shipped, and 8 day lead time, plus 2 or 3 days to ship DHL from China.

I'm not sure the "Made in USA" factor is worth that to me personally -- especially the month-long turnaround! What's everyone else think?

ibmapc
May 27th, 2016, 10:55 AM
I'd be fine with the Chinese boards if they are good quality.

glitch
May 27th, 2016, 11:09 AM
I'd be fine with the Chinese boards if they are good quality.

Same board house that did previous rev 2 boards and many of John Monahan's/s100computers.com boards, so the quality is solid and completely adequate for a through-hole board.

Ordered all shipping supplies for this run. Bare boards will ship in a bubble mailer to save on postage, kits and assembled boards will ship in a USPS Small Flat Rate Box to save on hassle. Antistatic bags for parts kits/assembled boards.

compu_85
May 27th, 2016, 11:20 AM
Ya that lead time is a bit killer.

glitch
May 27th, 2016, 11:51 AM
Ya that lead time is a bit killer.

The more I think about it, the more I think it sort of kills being able to do the board printing in the US. Add a month on, then the time it'll take me to get the orders processed...plus it's more expensive anyway, which will quickly add up for the people who wanted multiple boards.

We'll run it in China this time. Perhaps the next run(s), if planned for enough in advance, can be done in the US.

glitch
May 27th, 2016, 12:09 PM
Order placed with PCBCart, same manufacturers as the last run of rev 2 boards (at least, the last run I purchased from).

glitch
May 27th, 2016, 07:00 PM
Closed the preorder through glitchwrks.com, you can still confirm if you have a confirmation link. I'll email everyone who confirmed when orders are ready.

compu_85
June 3rd, 2016, 11:40 AM
Looking forward to this. I've got my Portable ready... full height 5.25 drives swapped out for a half height 3.5 and 5.25, and a HDD mounted in the now empty bay!

https://lh3.googleusercontent.com/-4Enf0ZkF_v8/V1DUB_AnNeI/AAAAAAAAKVM/Mme5RNz09NMuDJ0grSVNnGKpYeZ_51f9wCCo/s800/IMG_20160602_200637.jpg

-J

lyonadmiral
June 4th, 2016, 03:32 AM
If the cost difference isn't too much I'd pay more for a USA made card :)

Would you settle for "Assembled in the United States of America with Foreign & Domestic Components"? :)

glitch
June 4th, 2016, 06:33 AM
Would you settle for "Assembled in the United States of America with Foreign & Domestic Components"? :)

Heh, marketing-speak :) That *is* another point though, even if the board is made in the US, most (all?) of the ICs are not.

Board house says they should be shipping out 11 JUN. I'd imagine they will have been delivered by 15 JUN (DHL shipping is usually 2-3 days from China -> here)

glitch
June 9th, 2016, 06:32 PM
XT-IDE REV 3 BOARDS HAVE SHIPPED! Got the email a few moments ago, should be here in 2-3 days, as that's typical for DHL China -> Here.

CarlosTex
June 13th, 2016, 03:01 AM
If anyone is interested on the Lo-tech version check and you don't want to assemble them yourselves check here (http://www.amibay.com/showthread.php?83610-Fully-assembled-Lo-tech-8-bit-IDE-Adapter-rev-2-(Full-XT-IDE)).

glitch
June 13th, 2016, 05:14 AM
I'm out of town today, but it looks like the cards were delivered from China this morning!

Keep in mind the XT-IDE rev 3 will be available as just a board, board + parts kit, or fully assembled + tested. Kit includes all components, hardware, and brackets. I'll assemble one and post pictures probably tomorrow.

glitch
June 14th, 2016, 09:25 AM
Got home last night, tore into the package with the rev 3 boards this morning! They look great, typical high quality PCB Cart work. I assembled one using the parts that will be included in kits/assembled units:

http://i.imgur.com/WhNaoRn.jpg

http://i.imgur.com/biywgZ8.jpg

Tested it in my IBM 5150 PC, worked no problem!

Should be getting kits together this week or next, if you preordered you'll get an email with instructions on how to complete the transaction.

Agent Orange
June 14th, 2016, 09:54 AM
That's just downright purdy! Excellent solder work too. :thumbsup:

Stone
June 14th, 2016, 10:05 AM
What? No SMT? Geez, I could probably put that together. I didn't realize it would be done without SMT involved.

So, what's the final price for a kit, board plus components?

glitch
June 14th, 2016, 10:22 AM
What? No SMT? Geez, I could probably put that together. I didn't realize it would be done without SMT involved.

Correct, hobbyists seem to be pretty against SMT, so I stuck with all through-hole parts. Part of the point in the rev 3 board was to make it easier on hobbyists with little/no soldering experience to put the board together -- increased clearances, larger traces, oblong pads, et c.


So, what's the final price for a kit, board plus components?

I need to sit down and go through my ordering spreadsheets, but bare boards should be around $8 each and parts kits will probably be $20-25. Expect final numbers in the next week or so, as my day job allows time to finish this up :)

DDS
June 14th, 2016, 10:51 AM
Looks great!

Can't wait to get my hands on one & try it out. Or maybe two if any survive the first round of takers. ;-)

Chuck(G)
June 14th, 2016, 11:01 AM
Nice work, Glitch! Too bad that there's so little demand that it can't be farmed out to a Chinese fab house to reduce costs.

Malc
June 14th, 2016, 11:15 AM
They look great, typical high quality PCB Cart work.

Very nice :-) Wasn't it PCB Cart that fab'd the original R1 cards ?, I don't currently need any otherwise i would have put my name down, My R1 cards are still going strong.

glitch
June 14th, 2016, 11:40 AM
Thanks for the support, everyone! Yes, PCB Cart did the original run, and many of the production run s100computer.com boards. There should be plenty left after the preorders are fulfilled, I ordered about 40% more than was preordered. I plan on keeping a stock of these boards, and possibly kits, so that they're available to the community as a regular purchase item.

Chuck, I didn't look into having the boards populated in China, but I would guess that it would be more economical to design a SMD version if I were going that route. I can't imagine, with the limited number of "I want finished product" preorders I got, that it would be cost effective to have the boards assembled overseas.

T-R-A
June 14th, 2016, 02:24 PM
...I didn't look into having the boards populated in China, but I would guess that it would be more economical to design a SMD version if I were going that route. I can't imagine, with the limited number of "I want finished product" preorders I got, that it would be cost effective to have the boards assembled overseas.

You're correct, it wouldn't be (cost-effective). As with the PWB manufacturing, the cost of SMT assembly would be about the same for 5 boards as it would be for 500. It's even less cost-effective to have it done in the US (where SMT-manufacturing has almost now completely disappeared). You have costs for screen/jet-printers, assembly-line time, and inspection (possibly even test) that just can't be done cheaply, especially for small-batch runs (this comes from 20+ years in SMT-manufacturing experience).

compu_85
June 16th, 2016, 01:13 PM
$25 for a ready to assemble kit seems like a great deal to me! Eagerly waiting on the confirmation email!

compu_85
June 18th, 2016, 07:25 PM
Is the Google Code Page (https://code.google.com/archive/p/xtideuniversalbios/) still the best place to get the firmware for this?

-J

pearce_jj
June 18th, 2016, 11:51 PM
This is work in progress, but:

http://www.xtideuniversalbios.org
http://www.xtideuniversalbios.org/binaries

Trixter
June 19th, 2016, 07:54 AM
http://www.xtideuniversalbios.org/binaries/ needs to be a prominant link on the main page somewhere. Otherwise, if you consider that a work in progress, I cower in fear at the thought of what the finished product is -- it's already brilliant.

Malc
June 19th, 2016, 09:35 AM
-- it's already brilliant.

I agree, An excellent job, Better than the old google code site, Good idea on the 'Pre-built binaries download center', Handy for those who don't want to build from source and yet have access to the latest revision, And the source is available as it used to be.

alecv
June 19th, 2016, 01:10 PM
There are still two uncovered issues with xtideuniversalbios:

1) It hangs on the drive benchmark in the Norton Utilities (NU SYSINFO menu-driven version).
http://www.vcfed.org/forum/showthread.php?19591-XTIDE-tech-support-thread&p=364313#post364313
It seems SYSINFO plays dirty tricks with interrupt timer trying to be "multitask" and BIOS hangs.

2) Bad coexistance with 8-bit MFM controllers and their BIOSes while FULL_MODE and boot menu are disabled.
Most of 8-kb "lite" controllers use short BIOS version and affected. Simptoms: can't boot nor MFM
nor xtideuniversalbios-driven drive, "C" and "D" letters does not work. Both controllers are working
"solo". No resource overlapped.

pearce_jj
June 19th, 2016, 02:00 PM
Thanks for the comments. The binaries thing does indeed need a link, but anyway it's driven from a post-commit hook so as soon as new code is pushed in, the binaries will be right there.

Thanks for the issue reports. Something not yet working is self service registration that will allow you to create an issue ticket. But please bear in mind all of this is incredibly time consuming so please be patient :-)

glitch
June 19th, 2016, 02:02 PM
TBut please bear in mind all of this is incredibly time consuming so please be patient :-)

Important to remember with any open-source project, especially those where the work isn't anyone's day-job!

Trixter
June 20th, 2016, 09:07 AM
1) It hangs on the drive benchmark in the Norton Utilities (NU SYSINFO menu-driven version).
http://www.vcfed.org/forum/showthread.php?19591-XTIDE-tech-support-thread&p=364313#post364313
It seems SYSINFO plays dirty tricks with interrupt timer trying to be "multitask" and BIOS hangs.


According to http://www.vcfed.org/forum/showthread.php?19591-XTIDE-tech-support-thread&p=368707#post368707 it is a bug in SYSINFO: "The BIOS actually returns from the call to INT 13h AH=10h (AH10h_HandlerForCheckDriveReady as you noted) with a Timeout error (AH=80h) but SI ignores that and keeps calling the same function in its own timeout loop which keeps spinning for 65535 times.".

That sounds like the problem is with SYSINFO, not the XUB.

alecv
June 20th, 2016, 10:52 AM
I suppose SYSINFO triggers internal error in the XUB. I seems a stack corruption or race condition or reenterancy failary. I did a trace, it hangs on the Drive Ready waiting. SYSINFO does a very quick poll a BIOS to determine transfer speed. On some fast or caching controllers SYSINFO can't determine speed and writes message about "advanced controller".

pearce_jj
June 20th, 2016, 11:04 AM
Feel free to debug....

Trixter
June 20th, 2016, 11:14 AM
I suppose SYSINFO triggers internal error in the XUB.

I don't think the XUB is at fault here. I think you're ignoring the part where sysinfo "keeps calling the same function in its own timeout loop which keeps spinning for 65535 times". That's a terrible design choice. A proper busywait loop would set a time quantum of (for example) 5 seconds, and stop retrying after that time period had elapsed. The designers of SYSINFO assumed all int 13h,10h calls return almost immediately, but that was never their assumption to make. IBM's own PS2 and PC BIOS Interface Technical Reference makes no mention of how long that status call should or shouldn't take.

I don't consider the SYSINFO hang to be a result of the XUB, but rather poor coding. The XUB Int 13h,10h call returns successfully after a few seconds.

alecv
June 20th, 2016, 12:30 PM
Agreed, code style is ugly. However NU SYSINFO was extensively used in 90'th and there were no reports "hangs on transfer speed test" for many many systems. As shown in trace, this code

call IdeIO_InputStatusRegisterToAL
test al, FLG_STATUS_BSY ; Controller busy?
jnz SHORT .UpdateTimeout ; If so, jump to timeout update always return result "controller is in BUSY state". It's very interesting which call sequence renders controller to this state even for uber-fast Compact Flash?

There was some reports the early 1.1.x XUB does not hangs. SVN bisect may help us to find critical change.

It's interesting also to build test-bench with identical hardware i.e. 286 with 16-bit ISA (to exclude 8-to-16 mux effects) and debug native BIOS and XUB.

compu_85
June 23rd, 2016, 04:44 AM
glitch,

After posting all the teaser pictures inquiring minds want to know... how are the kits coming along? :D

-J

glitch
June 23rd, 2016, 04:25 PM
Just got back from a week-long work trip, it looks like all of the remaining kit parts have come in. I'm not digging into it tonight (9 hours of driving! woot!) but payment request emails should be going out this weekend or Monday/Tuesday next week.

Gerardcjat
June 25th, 2016, 01:34 AM
Hello Glitch,
Please, remember ( from my old P.M. ) that althrough I have not preordered, I am interested in a full assembled unit OR "board + part kit" depending on price.
Please P.M. me if one "full kit" is left and if it is conveniant for you to send it with the first batch.

glitch
June 26th, 2016, 05:33 PM
OK! So I finally had time to sit down and finish out the formulas in my big spreadsheet 'o parts, and I've got totals for everything as follows:

Bare Board $8
Parts Kit (KIT ONLY, NO BOARD) $20
Assembled unit $85

Shipping in the US as follows:

Bare Board (up to 3) $5
Parts Kit (up to 3) $7 (USPS Flat Rate Box)
Assembled Unit (up to 2) $7 (USPS Flat Rate Box)

Greater quantities and international shipping will be quoted per-order.

Please note! The parts kit does not include full sockets for everything, only the EEPROM. It's been my experience that sockets for all ICs, unless you're using high quality (read: expensive) sockets for every location, add more trouble than they're worth. All ICs have been sourced from reputable sources with supply chain documentation, so there should be a very low probability of dead ICs out of the package.

The EEPROM socket is a high quality machine pin socket with gold inserts. If people *really* want a socket kit -- again, it's been my experience that full socketing is more trouble than it's worth -- I can probably come up with an add-on socket kit.

If you preordered bare boards only (I think that was only one person) then order request emails have been sent out. I'm finishing up the script+template to email out the requests for parts kit orders.

bobba84
June 26th, 2016, 06:00 PM
Very exciting! What do you think the lead-time is for assembled boards? (No pressure at all, just curious :))

compu_85
June 26th, 2016, 08:42 PM
So $35 shipped for one complete kit + board, $62 for two? Great! Let us know how to pay :cool:

-J

glitch
June 27th, 2016, 04:35 PM
All preorder emails have been sent! First time using the new mailer system, so sorry if anyone gets a duplicate! If you preordered but don't see an email shortly, check your junk folder -- I haven't set up DKIM for the mailer yet, so some really strict spam filters might trip on it.

For those who didn't preorder, we've got plenty of boards and parts kits, but we're working on filling all of the preorders first. I'll post information on ordering once the preorders are out the door.

compu_85
June 27th, 2016, 05:23 PM
Got the email at my Gmail account. Payment sent!

glitch
June 27th, 2016, 07:05 PM
If you got your payment in before this post (2016-06-28 0300 UTC) then your package will ship tomorrow around noon!

glitch
June 27th, 2016, 07:12 PM
Thought maybe people would like to see the kits:

http://i.imgur.com/fBOIs80.jpg

That's all the bits and pieces. Amy and I picked 15 kits tonight. Here they are waiting to be boxed up for shipping:

http://i.imgur.com/9Ls2Xko.jpg

ibmapc
June 27th, 2016, 07:44 PM
I sent paypal payment to the Glitchworks email. the other one in your email message (chapmajs@vt.edu) was rejected by paypal

compu_85
June 27th, 2016, 08:39 PM
Must be too many people sending $$ at once. We've had similar problems at a car club I'm in.

-J

bobba84
June 27th, 2016, 10:42 PM
Mine to chapmajs@vt.edu worked! :)

glitch
June 28th, 2016, 03:59 AM
I sent paypal payment to the Glitchworks email.
DO NOT send PayPal payments to any @glitchwrks.com address -- none of those are associated with my PayPal account, and will not be associated with PayPal any time in the future.


the other one in your email message (chapmajs@vt.edu) was rejected by paypal

That is the correct PayPal address for me. Good thing I only use it for PayPal -- I'm sure it'll be getting a ton of spam now :)

glitch
June 28th, 2016, 09:49 AM
Started assembling boards for the confirmed/paid for orders:

http://i.imgur.com/O4zLJ3j.jpg

That picture was taken immediately after washing -- I assemble with an organic core solder, the flux is water soluble so it comes off with warm water and an old toothbrush. They've been burning in for a few hours in various PCs, probably going to head out in an hour or so!

ibmapc
June 28th, 2016, 04:14 PM
I sent paypal payment to the Glitchworks email. the other one in your email message (chapmajs@vt.edu) was rejected by paypal
I figured out what happened. I cut and pasted your email address and must have included a space in front so PayPal rejected it. Got rid of the extra space and it went through. DOH!!

bobba84
June 28th, 2016, 05:29 PM
Started assembling boards for the confirmed/paid for orders:

Exciting! They look great!

Just checking - they do some with the BIOS ROM don't they? What about the metal bracket?

Forgive me if this has already been answered :) Either way, I can't wait to get mine and try it!!

Bobby.

glitch
June 28th, 2016, 06:04 PM
Parts kits come with all necessary bits to build the boards -- bracket, screws, and EEPROM included! You provide solder and tools. The EEPROMs in the parts kits are blank, but of course the XT-IDE board can program the EEPROM in-board using the XUB config tools.

The assembled units come with the EEPROM pre-loaded with the latest version of the XT-IDE Universal BIOS, 1.1.x series (IIRC the 2.x series is still beta, so I haven't been shipping it with assembled boards).

RizThomas
June 29th, 2016, 03:51 AM
I know this is a silly question but I'll go ahead anyway...can I use a pre-programmed EPROM (27C256) instead of the EEPROM?

glitch
June 29th, 2016, 04:56 AM
I know this is a silly question but I'll go ahead anyway...can I use a pre-programmed EPROM (27C256) instead of the EEPROM?

Sure! Just set the switches appropriately -- enable 32K mode, and disable ROM Write.

compu_85
June 30th, 2016, 08:36 PM
Got my board assembled :D

Is there a listing of what the dip switches do? I found some docs on prior versions but want to be sure I'm correct with this R3 card.

-J

modem7
July 1st, 2016, 05:11 AM
Got my board assembled :D
Is there a listing of what the dip switches do? I found some docs on prior versions but want to be sure I'm correct with this R3 card.
I have drafted some switch setting diagrams:

[1. Basic IDE functionality (http://www.minuszerodegrees.net/xtide/rev_3/XT-IDE%20Rev%203%20-%20jumper%20configuration%20-%20IDE%20functionality.jpg)]
[2. Boot ROM (http://www.minuszerodegrees.net/xtide/rev_3/XT-IDE%20Rev%203%20-%20jumper%20configuration%20-%20optional%20boot%20rom.jpg)]

Glitch should be able to verify their content.

glitch
July 1st, 2016, 05:28 AM
I'll see if I can't write them up today, but the default of 0x300 I/O, 0xD000 ROM would be:

SW1-1 ON
SW1-2 OFF
SW1-3 OFF
SW1-4 OFF
SW1-5 ON (ON for 8K PROM)
SW1-6 ON (ON for 8K PROM)
SW1-7 ON (Enables EEPROM)
SW1-8 ON (Controls EEPROM write)

SW2-1 ON
SW2-2 ON
SW2-3 thru SW2-8 OFF

glitch
July 1st, 2016, 11:42 AM
Just to let everyone know, we'll be out of town for the 4th of July weekend, and I won't be back in until Tuesday afternoon/evening, so the next round of shipments will happen Wednesday. If you've already paid as of 2016-07-01 19:30:00 UTC then your order is already in the mail!

modem7
July 1st, 2016, 03:27 PM
I'll see if I can't write them up today,...
If you like, you can utilize the ones that I linked to in post #138.
Free for use/distribution.
No point in reinventing the wheel.

glitch
July 2nd, 2016, 06:59 AM
Oh, super awesome! I didn't see your post as it wasn't there when I hit reply yesterday! I'll add your diagrams to the GitHub repo, and include them in my writeup. They are correct, but you might add the external IDE activity LED jumper, if you don't think that's self-explanatory enough.

modem7
July 2nd, 2016, 04:05 PM
They are correct, but you might add the external IDE activity LED jumper, if you don't think that's self-explanatory enough.
I have added it.

lyonadmiral
July 2nd, 2016, 04:30 PM
Oh, super awesome! I didn't see your post as it wasn't there when I hit reply yesterday! I'll add your diagrams to the GitHub repo, and include them in my writeup. They are correct, but you might add the external IDE activity LED jumper, if you don't think that's self-explanatory enough.

I think it would be a wonderful add on product if there isn't one already; both a half height or full height faceplate kit to go with a card such as this?

glitch
July 3rd, 2016, 06:34 AM
I think it would be a wonderful add on product if there isn't one already; both a half height or full height faceplate kit to go with a card such as this?

You mean to go in an otherwise empty drive bay? If so, I'm kinda working on something in that vein at the moment... :D

DDS
July 3rd, 2016, 11:13 AM
You mean to go in an otherwise empty drive bay? If so, I'm kinda working on something in that vein at the moment... :D

I'm headed in that general direction with my 5160. I pulled the full height 360K drive out of the center drive bay and replaced it with a black bezel TEAC 360K 5.25" half height on the bottom and a generic black bezel 720K 3.5" in a half height adapter on top. Well, it's really a 1.44M but if I don't tell the controller maybe it won't notice. ;-) A little black electrical tape over the diskette density hole works wonders. Then in the right bay where the full height 20M IBM Japan MFM drive used to be I'm going to put two more 5.25" -> 3.5" adapters with IDE to CF card units for the C and D "drives". I figure it will be handy to swap out a CF card to change OS or data and for easily removable backup media.

Something like this:

http://www.ebay.com/itm/Startech-3-5in-Drive-Bay-IDE-to-Single-CF-SSD-Adapter-Card-Reader-/301661181996?hash=item463c68582c:g:WecAAOSwxj5XQBo G

Be advised that the eject buttons on these have received some bad press for being easy to break.

glitch
July 4th, 2016, 05:21 PM
Be advised that the eject buttons on these have received some bad press for being easy to break.

I can confirm that the super cheap versions of these types of devices do break easily. I prefer the kind with no eject button, or one from a reputable industrial supplier for that reason.

Chuck(G)
July 4th, 2016, 05:43 PM
The CF card has a big weakness--that's the connector. I've messed up more than one socket by pushing in a CF card with too much force when it wasn't quite in straight. Maybe we can do a version that uses SD card or USB flash. Heaven knows, they're fast enough now for vintage computer work.

pearce_jj
July 4th, 2016, 10:20 PM
Chuck, I have already an ISA USB card.

archeocomp
July 5th, 2016, 10:48 AM
Yeah, but last time I asked about it, nobody confirmed that it works.

glitch
July 5th, 2016, 10:58 AM
Maybe we can do a version that uses SD card or USB flash. Heaven knows, they're fast enough now for vintage computer work.

I don't know about USB -> IDE bridges (at least, not the kind that go the way *we* need :) ) but SD -> IDE bridges do exist. I know I've seen them with the SD slot accessible through a PCI/ISA slot cover, but I don't know that I've seen them in 3.5" form factor.

Chuck(G)
July 5th, 2016, 11:06 AM
Just thinking aloud--CF cards, I think, are sunsetting, as are PATA devices in general. Interfacing to an SD card using SPI is very simple. That Orange Pi PC that I'm using has a 16GB Micro SD Class 10 card in it and it's nearly as fast as a SATA hard disk. The biggest problem that I see with Micro SD is losing the blasted tiny things.

glitch
July 5th, 2016, 11:53 AM
That and SMD soldering -- seems a lot of hobbyists are allergic to it. I suppose leaving footprints for e.g. an Adafruit breakout board, or just providing boards with the SMD part(s) soldered down would be OK.

DDS
July 5th, 2016, 02:36 PM
"That and SMD soldering -- seems a lot of hobbyists are allergic to it."

Unsteady hands and dimming eyesight? Nah, that couldn't be it!

glitch
July 5th, 2016, 02:43 PM
Unsteady hands and dimming eyesight? Nah, that couldn't be it!

Even younger hobbyists seem reluctant to try it out! The larger pitch stuff requires no special equipment, I started doing SOIC PICs with a 15 Watt Rat Shack pencil iron and whatever sorta-small solder they had on the shelf, on home etched boards. With a proper solder masked board, assuming you're not using paste+stencil, it's easy to just drag solder it with a squirt of liquid flux and clean up with solder wick.

But, regardless of the reason for not wanting to do it, SMD seems to be out for most hobbyist kits. I'm set up to do small runs of SMD boards here, I'm sure I could expand enough to handle hobby level order quantities. I figure assembled (or "already contains the SMD bits" assembled) will be the way any SMD projects go for hobbyist stuff.

compu_85
July 5th, 2016, 02:57 PM
Assembled one of my kits last weekend. After a little configuration so the address didn't conflict with the NIC the interface is working fine!

https://lh3.googleusercontent.com/-Xh71v--SOVA/V3ZeovFIWwI/AAAAAAAAKpQ/TePZ8bIe_nIUEl3vnwBFOO0bVj_dC7LnQCCo/s640/IMG_20160701_073109.jpg

https://lh3.googleusercontent.com/-Yj2D5Dofac0/V3w4760vkSI/AAAAAAAAKtI/Do1vfGsSOFcdEK1B-6IgW7SZW0PLfJHkwCCo/s850/IMG_20160705_182931.jpg

With a ST-157a I'm getting quite respectable speeds... 588k/s and 30ms seek times!

https://lh3.googleusercontent.com/-8mYJxBVRTNg/V3w6_1O0SgI/AAAAAAAAKtU/cLWIzyyUThEI76lI5MfhM8c8L58ZjiLCwCCo/s584/Compaq-SI.GIF

-J

Chuck(G)
July 5th, 2016, 03:20 PM
My eyesight's as dim and my hands unsteadied by age as much as anyone's on this thread and I still managed to put together Jame's XTCF-V2 without incident (100% SMT). You just have to be careful.

glitch
July 5th, 2016, 05:18 PM
Added modem7's diagrams to the Git repo, and linked to his site in the README.

Trixter
July 5th, 2016, 05:28 PM
it's easy to just drag solder it with a squirt of liquid flux and clean up with solder wick.

I doubted this method until Jim Brain demonstrated it to me in person recently. I'm still a ways away from trying it, but it did indeed work and didn't seem that bad. I thought the desoldering braid would remove all of the solder, but it only removed the solder from the top of the connections; the solder underneath remained.

Chuck(G)
July 5th, 2016, 05:32 PM
Yup, works a treat--it's particularly interesting in that you don't use a fine tip on the iron--a nice fat chisel tip works extremely well.

I do check my work with an inspection microscope after I'm done, just to be sure.

pearce_jj
July 5th, 2016, 11:07 PM
That's the technique I tend to use too. I tried a reflow oven but I didn't get on with that.

rosaage
July 6th, 2016, 02:49 AM
Got my kit today. What would be the best to use, leadfree or 60/40 solder? And what about flux and cleaning? This is the first time I am soldering something besides repair jobs and small kits at school.

pearce_jj
July 6th, 2016, 02:56 AM
Traditional multi core 60/40 is by far the easiest. Clean if you want to with toothbrush and warm water or isopropyl alcohol depending on the exact type. If using lead free you need higher bit temperature - and it smells bad.

glitch
July 6th, 2016, 04:11 AM
Wow, that got to Norway pretty quick! I use 63/37 lead/tin solder with an organic acid core, which is water soluble, for the assembled units. I recommend 63/37 rosin core if you're not going to wash it or don't want to deal with organic core (which *MUST* be washed, or it causes issues). Personally I prefer Kester "44 Core" if I'm using rosin core.

I don't use lead free unless a customer requests it. That happens so infrequently that I don't think I've even got a roll of it on the shelf. It does require more heat. The board I shipped was finished in leaded HASL (Hot Air Solder Leveling) for the pads, and I'm betting that most of the components shipped in the kits are tinned with leaded solder too (most of them are pre-ROHS).

smyke
July 6th, 2016, 04:48 AM
Added modem7's diagrams to the Git repo, and linked to his site in the README.

What is the link to the Git repo? I can't seem to find it.

glitch
July 6th, 2016, 05:00 AM
What is the link to the Git repo? I can't seem to find it.

https://github.com/glitchwrks/xt_ide/

Robin4
July 6th, 2016, 11:32 AM
Where can we buy this glitch XT IDE controller card? Iam intrested in one for now, like to use it as an overview example for now..

If people want to buy the pcb only, isnt it better to make a small kit with pcb + bracket + screws (because maybe this bracket could be hard to get, if it is included it makes things a lot easier)

I only dont like the color of the pcb, but for an first try-out example i dont care.. Hopefully it would be possible when ordering some bare pcb`s the color can be changed or text on the pcb can be editted.

If someone can give me the link i will purchase one. I only hope that the shipping costs would not max-out to much.

rosaage
July 6th, 2016, 11:56 AM
I'm having trouble getting it to work, It won't boot from hdd and fdisk reports "no fixed disks" Is there some kind of menu I can enter to test if the card is working?

Robin4
July 6th, 2016, 12:09 PM
Boards arrived today!

http://i.imgur.com/6K2jnpa.jpg

One slight problem -- KiCad apparently changed cathode designations on diodes and LEDs between the version that the rev 2 board schematic was created with, and the new KiCad libs. So the LED is backwards on the silkscreen. I also eliminated a discrete resistor by using a spare from one of the resistor packs. Other than that, it works!

As you can see from the picture, the board house I do prototype runs with wasn't able to do selective hard gold for the edge connector. I'm getting quotes from a few of the production shops I've used in the past, including PCBCart, which I believe is who did the rev 2 run. I'll probably remove the unused ISA fingers to save on plating costs.

You dont use gold on the ISA interface fingers? ( i hope this isnt tin only)


Sorry, been super busy with this project, as well as day-job work! I must've missed your question. Typically in the past I've sold the rev 2 boards for $100 assembled or $85 if the buyer didn't want the UART bits. I'm hoping to offer the rev 3 to *hobbyists* at $85 each, assembled, but it will depend on what kitting up actually costs. I'd previously been running from stocks of ICs that were basically surplus from other day-job billable work so I don't have a solid figure on what the parts kits will cost right this moment.

Assembled of course really means "assembled and tested."

You said that you are trying to get the full kit to a price of $85 when it is assembled, how much would the price be if i need to assembled it my self?

Chuck(G)
July 6th, 2016, 12:10 PM
Do you see the XTIDE BIOS signon message?

glitch
July 6th, 2016, 01:13 PM
*grumble* Please go back and read the rest of the development thread -- that's a prototype board, which had HASL fingers, the production boards have true hard gold plated edge fingers. The color's blue because, for me at least, it gives excellent trace contrast. Plus I like it. As this is an open source project, the Gerber files for the current in-production revision are available on GitHub. You're free to run the boards in any color you wish, but a single board with hard gold edge connector is probably going to cost more than $100 USD. ENIG finish is not sufficient.

The Keystone bracket is available through the normal channels, the brackets included in the kits are from Mouser. They had plenty on hand.

$85 assembled is the assembled price, the blank boards are $8 each, and the parts kits are $20 each. Unless you just happen to have all of the parts for the XT-IDE in your shop, my price on the parts kit is, IIRC, cheaper than just the ICs from Mouser. Not exactly making bank on the parts kit, and the board is basically at cost + shipping. I'm hesitant to start offering partial kits because then I either have to pick the kits as each order comes in, or put together (and store) a bunch of different kit configurations.

And do be mindful that this is a *hobbyist* run and the design is covered by the GNU GPL under which I've licensed it. That means if you buy one, use it in your $10 million CNC machine, and then your CNC machine dies, it's not my problem.

glitch
July 6th, 2016, 01:14 PM
I'm having trouble getting it to work, It won't boot from hdd and fdisk reports "no fixed disks" Is there some kind of menu I can enter to test if the card is working?

If you get the BIOS message but don't get any drives identified, make sure you chose the correct port parameters for your switch settings. Also the XT-IDE Universal BIOS config tool for version 1.1.5 (the one I ship in the assembled units) will incorrectly tell you to add 0x200 to the base I/O address for the second address -- this is wrong, you add 8 to the base address.

SpidersWeb
July 6th, 2016, 01:17 PM
You said that you are trying to get the full kit to a price of $85 when it is assembled, how much would the price be if i need to assembled it my self?

Just for the sake of a quick answer I'll requote glitch, it was at the bottom of page 12.


Bare Board $8
Parts Kit (KIT ONLY, NO BOARD) $20
Assembled unit $85

Shipping in the US as follows:

Bare Board (up to 3) $5
Parts Kit (up to 3) $7 (USPS Flat Rate Box)
Assembled Unit (up to 2) $7 (USPS Flat Rate Box

Edit: I posted this without seeing glitch had already replied, but oh well, there it is again.

rosaage
July 6th, 2016, 01:22 PM
I have a bad fealing now. It does ram count then tries floppy. After this it starts basic. No new messages relating to the xt-ide.

glitch
July 6th, 2016, 01:27 PM
I have a bad fealing now. It does ram count then tries floppy. After this it starts basic. No new messages relating to the xt-ide.

Did you choose a ROM address that conflicts with something else in your system? Is the ROM enabled (SW1 ENA should be ON)

rosaage
July 6th, 2016, 01:45 PM
Did you choose a ROM address that conflicts with something else in your system? Is the ROM enabled (SW1 ENA should be ON)

I have no idea. I used the default values (i think), sw2 set to 1100000 and sw1 10001111

It is an IBM Portable Personal Computer (5155) with IBM cga card in slot 1, ibm floppy controller in slot 3 and the XT-ide in slot 4. Currently 1 floppy drive connected.

glitch
July 6th, 2016, 01:55 PM
Hmm, doesn't sound like there should be any conflicts. Those switch settings should be the default of 0xD000 for ROM and 0x300 I/O base. You said the EEPROM programmed fine using the DOS utility?

rosaage
July 6th, 2016, 02:00 PM
Hmm, doesn't sound like there should be any conflicts. Those switch settings should be the default of 0xD000 for ROM and 0x300 I/O base. You said the EEPROM programmed fine using the DOS utility?

Yes, no errors after selecting 64 page size and flashing. This was on a pentium II system running 98se with a pci ethernet card, pci soundblaster and a ati rage II on agp. Could I have flashed something else?

glitch
July 6th, 2016, 02:15 PM
Yes, no errors after selecting 64 page size and flashing. This was on a pentium II system running 98se with a pci ethernet card, pci soundblaster and a ati rage II on agp. Could I have flashed something else?

It's unlikely. Try booting the XT-IDE in the Pentium 2 system, it should come up with the XT-IDE Universal BIOS menu.

compu_85
July 6th, 2016, 07:23 PM
I'm using the latest bios in mine... so far so good.

I did initially have an issue with address conflicts. This caused random bits to end up in the eeprom when I tried to flash it with the software tool.

-J

rosaage
July 7th, 2016, 03:43 AM
I've come one step closer to something now. I copied the xtidecfg.com and the bin file to a empty floppy and tried to do it on my 5155. Auto-configure: No controllers found, but when I flashed it this time the bios works. I get the XT-IDE menu where it scans for drives.

Now over to the bad part. Connected a IDE drive with ms-dos 6.22, No drives found. Tried starting the config program again after flashing and selecting auto-configure. Still "no controllers found" message.

The Pentium system Didn't work either before. I'll try with the win98 drive and this card now and see if it works.

EDIT: Pentium II system reported same. Moved the IDE cable from mobo header to XT-IDE. The card booted the bios fine, but reported no drive on master and no drive on slave.

glitch
July 7th, 2016, 05:02 AM
If you're getting the ROM menu but no drives detected, that's often a sign that you've got something misconfigured -- since you've already provided your switch settings, double-check that the BIOS is set up for the right base address, and that your high-speed/Chuck Mod jumpers are in the right position. Unfortunately I can't really help with the 2.x BIOS configuration, I've never personally used it. I still ship the 1.1.5 XUB with assembled cards, since the 2.x series is still listed as beta.

Malc
July 7th, 2016, 05:13 AM
Or possibly a bad solder joint / Chip, Have a good look at all solder joints and reflow any that looks suspect.

pearce_jj
July 7th, 2016, 05:24 AM
Also try a different device.

glitch
July 7th, 2016, 07:10 AM
Writeup on the rev 3 design process:

http://www.glitchwrks.com/2016/07/06/xt-ide-rev3

I'll be doing a writeup, or maybe a video, on how I assemble them as well, sort of a "this is the easiest order to stuff components" thing.

I'd like to wrap up the preorders by some time next week. Currently I'm waiting on payments, kits and assembled units are sitting on the shelf, ready to ship. I want to get the preorders squared away so that the people who missed the preorder can place their orders. If you preorderd and I haven't heard something from you by next week, I will likely cancel your preorder and sell the board(s)/kits(s) I'd reserved for you. If you've already messaged me about paying later for whatever reason, then you're already taken care of and your preorder will be held!

jltursan
July 7th, 2016, 08:30 AM
Are there PCBs still in stock?; if so, I'll happily get one of them (PCB+parts if both are available) :)

glitch
July 7th, 2016, 09:41 AM
Are there PCBs still in stock?; if so, I'll happily get one of them (PCB+parts if both are available) :)

Yep! I ordered about 40% more than was preordered. I will post information on placing orders once preorders are squared away.

Stone
July 7th, 2016, 02:57 PM
If you preorderd and I haven't heard something from you by next week, I will likely cancel your preorder and sell the board(s)/kits(s) I'd reserved for you.Not for anything but I hope you don't find that the number of deadbeats here corresponds with the numbers observed in general, elsewhere. Too many losers will guarantee you that it's a done deal when it's anything but that. Asking for a measly $10 deposit will weed the a**holes right out.

bobba84
July 7th, 2016, 03:13 PM
Sorry if this has been answered before, but what are the limitations of this board?

- Can it do over 8GB?

- Can it be used with an IDE > SATA Adapter, the kind that plugs in the back of a SATA drive?

Thanks!

Keenly awaiting my order's arrival (Australia is so far away haha)

Stone
July 7th, 2016, 03:20 PM
(Australia is so far away haha)I speak with my mate in Sydney every other day for nearly an hour via SKYPE so it doesn't seem all that far. :-)

glitch
July 7th, 2016, 03:27 PM
Not for anything but I hope you don't find that the number of deadbeats here corresponds with the numbers observed in general, elsewhere. Too many losers will guarantee you that it's a done deal when it's anything but that. Asking for a measly $10 deposit will weed the a**holes right out.

A formal preorder process seems to have helped greatly with it, so far I'm at about 10% of confirmed preorders who haven't made any sort of contact with me, which is within my expected tolerances. Plus, anyone who fails to pay on this preorder will be logged and refused preorders on any future projects.


- Can it do over 8GB?

Sure, but your OS will also need to support hard drives > 8GB to make use of it.


- Can it be used with an IDE > SATA Adapter, the kind that plugs in the back of a SATA drive?

I would assume so, as long as the SATA bridge is well behaved. I haven't personally tested this configuration.

ibmapc
July 8th, 2016, 10:39 AM
Success,

Got my board working last night. So far no problems running a spinning hard disk. I have a DOM coming in a week or two and look forward to using that. However, I could use some advice. There are a bunch of different versions of the XTIDE Universal BIOS and I can't figure out Which one would be best for my situation. I started with XTIDE_Universal_BIOS_v2.0.0_beta3 from HERE (https://code.google.com/archive/p/xtideuniversalbios/downloads). I know James Pearce has more recent versions on his web site. Should I use one of those?

My current set up is as follows:

XT/286 5162.
XTIDE rev 3 with a 3MB Fujitsu IDE HD but will be changing to a 1 GB DOM.
Original 20 MB MFM HD with Original Controller. (I read somewhere that the newer version of XTIDE Universal BIOS didn't get along well with this drive, but so far so good with XTIDE_Universal_BIOS_v2.0.0_beta3 )

Also, I plan to move the board over to my 5155 later on.

All suggestions welcome.

What are the pro's and cons of the different versions?

glitch
July 8th, 2016, 11:46 AM
I ship the XT-IDE Universal BIOS 1.1.5 with assembled boards, since the 2.x series is still listed as beta.

ibmapc
July 8th, 2016, 12:10 PM
I ship the XT-IDE Universal BIOS 1.1.5 with assembled boards, since the 2.x series is still listed as beta.
Yep, I was wondering why there are three BETA versions and the most recent non-beta is 6 years old. I just stared with the most recent version I could find quickly, knowing that there could be issues.

pearce_jj
July 8th, 2016, 12:25 PM
You can download the latest code from here (http://www.xtideuniversalbios.org/binaries/). You will need to configure it for your card as you program the EEPROM.

Krille
July 8th, 2016, 01:41 PM
I know James Pearce has more recent versions on his web site. Should I use one of those?
Yes, always use the latest version. I will post in the v2 beta testing thread if/when I or someone else commits code unsuitable for use. Don't worry about the v2.x versions being 'beta' just because they explicitly say that in the version string. As I see it, they are all beta versions, more or less.

Keep in mind, if everyone only uses old versions then development will stop due to lack of testing and feedback. Yes, it sucks to be a guinea pig but if I were to do all the testing myself, then it would be extremely time consuming, and expensive (to acquire all the hardware). Yes, I'm a cheap bastard. ;)


What are the pro's and cons of the different versions?
The only 'pro' I can think of with regard to v1.1.5 (and earlier, I suppose) is that the harddisk benchmark in Norton Utilities SI.EXE works. With some builds of v2.x it will (appear to) hang the machine.

Pros for v2.x;
* Support for more hardware (different types of harddrive controllers and of course the virtual serial drive feature).
* It's _required_ for the XT-IDE v2 and v3 "High speed mode".
* More features in general (power saving for example).
* It's compatible with (can run from the ROM socket on) 3Com 3C503 cards unlike older versions.
* Higher performance.
* It's modular which is great if you want to make your own custom builds from source using only the features you want.

Note that I'm talking about the latest versions here, not the official releases available from Google Code.

I'm sure other people can add more pro's and cons to the list (for old and new versions). I would be interested to hear if anyone is using an older version (v1.x) for some specific reason.

glitch
July 8th, 2016, 01:53 PM
It's _required_ for the XT-IDE v2 and v3 "High speed mode".

Chuck's modified version of Hargle's BIOS also works with the high speed mode/Chuck Mod. That's what's running on my Chuck Modded rev 1 card.

rosaage
July 8th, 2016, 02:07 PM
I have no idea any longer. When I came home today and tried the flashing utility it suddenly said "Found 1 controller" on the auto-configure option. It wasn't kidding either, Master at 80h: WD something...
The problem now is that I have no idea how I can make this bootable now. The partitioning I tried on another pc didn't work. I tried using windows formatting to clear the drive (unformatted) and used fdisk from PC-Dos 3.3 to create a primary C Partition. So far so good, then comes the next big issue. I type in "C:" it changes fine. I try "mkdir DOS", then "Error reading drive C: " Is there some special formatting I have to do in order to use it?

Krille
July 8th, 2016, 02:13 PM
Chuck's modified version of Hargle's BIOS also works with the high speed mode/Chuck Mod. That's what's running on my Chuck Modded rev 1 card.

Ah yes, that's true. I had forgotten about that. Are you using that version for a specific reason?

Stone
July 8th, 2016, 02:15 PM
The problem now is that I have no idea how I can make this bootable now. The partitioning I tried on another pc didn't work. I tried using windows formatting to clear the drive (unformatted) and used fdisk from PC-Dos 3.3 to create a primary C Partition. So far so good, then comes the next big issue. I type in "C:" it changes fine. I try "mkdir DOS", then "Error reading drive C: " Is there some special formatting I have to do in order to use it?It would help to know what drive you are using.

rosaage
July 8th, 2016, 02:18 PM
it would help to know what drive you are using.

wd2500jb 250gb

glitch
July 8th, 2016, 02:32 PM
Ah yes, that's true. I had forgotten about that. Are you using that version for a specific reason?

Not really, I burned it into an EPROM back when Chuck published his modification. It works fine, so I haven't seen a need to replace it.

Stone
July 8th, 2016, 02:42 PM
wd2500jb 250gbYou don't really expect to be able to use a 250GB drive with DOS 3.3 do you?

rosaage
July 8th, 2016, 02:48 PM
You don't really expect to be able to use a 250GB drive with DOS 3.3 do you?

I thought any size was fine, but dos would only see some of it.

Stone
July 8th, 2016, 02:54 PM
I thought any size was fine, but dos would only see some of it.That ~may~ be correct but DOS 3.3 can *only* see 32MB partitions.

rosaage
July 8th, 2016, 03:01 PM
That ~may~ be correct but DOS 3.3 can *only* see 32MB partitions.

How can I make a 32MB partition? If I try fdisk it doesn't work. fdisk option 4 reports the created partition as: "Start: 0, End: 3, Syl: 4. Total sylinders: 1024"

Stone
July 8th, 2016, 03:23 PM
Does FDISK give you the option to... Delete non-DOS Partition ... in its menu?

Aside from that a much better alternative would be to use DOS 6.xx or 7.x.

pearce_jj
July 8th, 2016, 03:40 PM
Re BIOS version, the main reason to use v2 is that Tomi, Krille and Gregg have poured hundreds of hours into this code to make it as fast and as compatible as it can be.. It works, it works well, and none of these projects would be possible without their efforts.

pearce_jj
July 8th, 2016, 03:48 PM
How can I make a 32MB partition? If I try fdisk it doesn't work. fdisk option 4 reports the created partition as: "Start: 0, End: 3, Syl: 4. Total sylinders: 1024"

Incidentally FAT12 can put-perform FAT16 for random IO because the resultant FAT fits in the DOS buffers; digging out some DOS2 disks can be worthwhile.

glitch
July 8th, 2016, 04:00 PM
Does FDISK give you the option to... Delete non-DOS Partition ... in its menu?

You'll probably need to use DOS 5 or 6.22 to delete the partition. Start with no partitions at all on the drive, and do all partitioning/formatting with DOS 3.3, if that's what you're going to use.

And yes, you can use a 250 GB drive, you'll just waste almost all of it. Part of the reason I recommend people use smaller industrial Flash modules, or CompactFlash Cards.

Stone
July 8th, 2016, 04:04 PM
And yes, you can use a 250 GB drive, you'll just waste almost all of it.You could get about 24 32MB partitions. That would use almost 1GB. :-)

lyonadmiral
July 8th, 2016, 04:24 PM
You could get about 24 32MB partitions. That would use almost 1GB. :-)

There have to be some small drives on eBay somewhere....

rosaage
July 8th, 2016, 04:44 PM
You'll probably need to use DOS 5 or 6.22 to delete the partition. Start with no partitions at all on the drive, and do all partitioning/formatting with DOS 3.3, if that's what you're going to use.

And yes, you can use a 250 GB drive, you'll just waste almost all of it. Part of the reason I recommend people use smaller industrial Flash modules, or CompactFlash Cards.

I have a 128MB DOM coming my way, It entered the country a day ago so I will probably have it some time next week.

I feel a bit stupid now. I forgot to type format C: after fdisk (Not used to dos yet :( ). DOS 5.0 managed to format it working, but PC-DOS 3.3 is not working. "format C:" reports "Insert a new disk into drive C:" I can't find any info about switches to force hard drive format. Is there something big I'm missing?

Chuck(G)
July 8th, 2016, 04:52 PM
Try partitioning and setting active under PC-DOS 3.3. There were some serious changes between 3.3 and 5.0. Chances are that 3.3 just doesn't recognize a 5.0 partition table entry.

lyonadmiral
July 8th, 2016, 04:53 PM
I have a 128MB DOM coming my way, It entered the country a day ago so I will probably have it some time next week.

I feel a bit stupid now. I forgot to type format C: after fdisk (Not used to dos yet :( ). DOS 5.0 managed to format it working, but PC-DOS 3.3 is not working. "format C:" reports "Insert a new disk into drive C:" I can't find any info about switches to force hard drive format. Is there something big I'm missing?

Run FDISK from your DOS 3.3 disk and see if it recognizes the drive.

rosaage
July 8th, 2016, 05:05 PM
Try partitioning and setting active under PC-DOS 3.3. There were some serious changes between 3.3 and 5.0. Chances are that 3.3 just doesn't recognize a 5.0 partition table entry.

That's what I did, deleted all partitions from fdisk DOS 5.0 then booted PC-DOS 3.3. fdisk 4 reports no partitions. Then I choose option 1 and set it to use max. Everything goes well until I try formatting the partition. I tried an english version of PC-DOs 3.3 too, but same issue there. It wants me to put in a floppy to the hard drive. If i press enter it just stay still for a few seconds (no light on the XT-IDE card) and report format failed.

lyonadmiral
July 8th, 2016, 05:12 PM
That's what I did, deleted all partitions from fdisk DOS 5.0 then booted PC-DOS 3.3. fdisk 4 reports no partitions. Then I choose option 1 and set it to use max. Everything goes well until I try formatting the partition. I tried an english version of PC-DOs 3.3 too, but same issue there. It wants me to put in a floppy to the hard drive. If i press enter it just stay still for a few seconds (no light on the XT-IDE card) and report format failed.

Don't have it do the MAX, manually set the partition size.

rosaage
July 8th, 2016, 05:21 PM
Don't have it do the MAX, manually set the partition size.

I tried now by manually selecting 4 cylinders and 1 cylinder, mark as active and reboot. same result both times, still thinks C: is a floppy drive.

Chuck(G)
July 8th, 2016, 05:27 PM
What happens if you do a FORMAT D: or FORMAT E: or FORMAT F:? Perhaps DOS thinks you've got 4 floppy drives.

Stone
July 8th, 2016, 05:30 PM
Try MB instead of cylinders.

Stone
July 8th, 2016, 05:34 PM
There have to be some small drives on eBay somewhere....How small?

I have an 80MB and a 170MB.

rosaage
July 8th, 2016, 05:39 PM
What happens if you do a FORMAT D: or FORMAT E: or FORMAT F:? Perhaps DOS thinks you've got 4 floppy drives.

It worked. format d: was the magic command. I looked at the dip switch on the motherboard. It was set to 3 floppy drives.
EDIT: Setting it back to 2 floppies fixed booting. I think it was set to 3 floppies when I experimented with a 1.44 and external cable.

glitch
July 8th, 2016, 05:59 PM
Glad you figured it out!

ibmapc
July 9th, 2016, 04:32 PM
I ship the XT-IDE Universal BIOS 1.1.5 with assembled boards, since the 2.x series is still listed as beta.
Glitch,
So You are using "Compatibility" mode? The "Chuck Mod" isn't supported 'til 2.x. Do you think it's more stable without the "Chuck Mod"?

glitch
July 9th, 2016, 05:05 PM
Glitch,
So You are using "Compatibility" mode? The "Chuck Mod" isn't supported 'til 2.x. Do you think it's more stable without the "Chuck Mod"?

I don't think that makes a bit of difference for fully IBM PC compatible users. See previous entries in this thread for why it's not a universal solution (that's why I kept the jumpers). AFAIK it's a works/doesn't work matter with the incompatible units.

As mentioned, I'm running a modified Hargle BIOS on my Chuck Mod rev 1 card. That seems to be stable. The only reason I'm still shipping 1.1.5 is simply because 2.x is listed as beta. I figure, if the devs aren't willing to call it out-of-beta, then I'm not willing to ship to to people who expect a card to "just work" when they plug it in. I will probably switch to the 2.x series as soon as someone's willing to call it out-of-beta. I doubt there's any real issues with it, a lot of folks seem to be using it, but remember *I* get the angry emails when someone pays for an assembled board and it doesn't work :)

Chuck(G)
July 9th, 2016, 05:31 PM
I'll add a bit here. The "Chuck mod" generally works when the host CPU is an 8088 or V20 (i.e. a CPU with an 8-bit bus interface).
Its operation depends upon the functioning of the 8088/V20 BIU. If you've got an 8086/V30/80286 CPU, then it's a roll of the dice--it all depends on how carefully the designers of your board decided to emulate the 8088 BIU.

Most have it wrong--the Olivetti M24/ATT 6300, for example, does split a 16-bit read/write into two 8 bit ones, but gets the order wrong. I don't know about the Tandy 1000 286 systems, but I suspect they've also got it wrong.

ibmapc
July 9th, 2016, 07:30 PM
I'll add a bit here. The "Chuck mod" generally works when the host CPU is an 8088 or V20 (i.e. a CPU with an 8-bit bus interface).
Its operation depends upon the functioning of the 8088/V20 BIU. If you've got an 8086/V30/80286 CPU, then it's a roll of the dice--it all depends on how carefully the designers of your board decided to emulate the 8088 BIU.

Most have it wrong--the Olivetti M24/ATT 6300, for example, does split a 16-bit read/write into two 8 bit ones, but gets the order wrong. I don't know about the Tandy 1000 286 systems, but I suspect they've also got it wrong.
So, as mentioned earlier, I currently am using it in an IBM XT/286 5162. Do you think the "Chuck Mod" will work properly in this system? Eventually, it's going in my 5155. Pretty sure it won't be an issue there. I just want to make sure it's rock solid. I never did get my rev 1 board to work reliably.

Chuck(G)
July 9th, 2016, 07:48 PM
I can't say, as I don't have my hands on a 5162, so you'll just have to try it. Really, why anyone would want to use an XTIDE on a machine with 16-bit slots is a bit of a mystery to me. Is it just for the BIOS? A 16-bit 0WS IDE controller should be able to outrun an XTIDE easily.

ibmapc
July 9th, 2016, 08:18 PM
...Really, why anyone would want to use an XTIDE on a machine with 16-bit slots is a bit of a mystery to me. Is it just for the BIOS? A 16-bit 0WS IDE controller should be able to outrun an XTIDE easily.
The 5162 only has 5 16 bit slots. -1 has video card -2 has NIC -3 has Memory Board -4 has Sound blaster 16 -5 has original MFM HDD / FDD controller. That leaves only 8 bit slots. Besides, I'm only playing with it in the 5162. Eventually it'll go in my 5155. I guess I should put it in there and then look for the best BIOS setup.

Chuck(G)
July 9th, 2016, 08:33 PM
There were IDE/Floppy cards to replace the original HDD/FDD card.

ibmapc
July 9th, 2016, 09:43 PM
Yes, but I find it strangely gratifying to use the original 20MB MFM disk to boot. It has zero defects according to spinrite and the sound it makes is really cool. Brings me back to my 20's, helping my Mom figure out Her 5160. This 5162 is a recent purchase that Trixter pointed me to. It is really an amazing find. Apart from the dead battery, it is in perfect condition and for some reason, I can't stand the idea of De-comissioning the original HDD and Controller.

pearce_jj
July 9th, 2016, 11:50 PM
Most have it wrong--the Olivetti M24/ATT 6300, for example, does split a 16-bit read/write into two 8 bit ones, but gets the order wrong. I don't know about the Tandy 1000 286 systems, but I suspect they've also got it wrong.

AFAIK the '6300 is the only machine confirmed to have this issue.

Great Hierophant
July 10th, 2016, 11:12 PM
I'll add a bit here. The "Chuck mod" generally works when the host CPU is an 8088 or V20 (i.e. a CPU with an 8-bit bus interface).
Its operation depends upon the functioning of the 8088/V20 BIU. If you've got an 8086/V30/80286 CPU, then it's a roll of the dice--it all depends on how carefully the designers of your board decided to emulate the 8088 BIU.

Most have it wrong--the Olivetti M24/ATT 6300, for example, does split a 16-bit read/write into two 8 bit ones, but gets the order wrong. I don't know about the Tandy 1000 286 systems, but I suspect they've also got it wrong.

Some Tandy 1000s came with 8086 CPUs (SL, SL/2, RL), some with 80286 CPUs (TL, TL/2, TL/3, RLX) and all except the RSX come with an 8-bit slot. I do not recall anyone complaining about any issues with the Chuck mod.

glitch
July 12th, 2016, 12:18 PM
FINAL NOTICE: I've sent out emails to the last four people who haven't gotten in touch with me about their preorder. If I haven't heard from you by close of business tomorrow (2016-07-13 1900 UTC) your preorder will be cancelled.

ibmapc
July 12th, 2016, 12:30 PM
If any of those go unclaimed, I'm considering buying one more kit. Are those Kits or assembled units that are unclaimed?

glitch
July 12th, 2016, 01:06 PM
If any of those go unclaimed, I'm considering buying one more kit. Are those Kits or assembled units that are unclaimed?

I have *plenty* of boards and parts kits. I have some assembled units, but I just assemble more as needed. This is going to be a "stock item" for me -- I'm planning on keeping boards and/or parts kits available for several projects.

General availability will start on Thursday morning. I'll post more information on how to place your orders then. I may not have my online order system finished by then, so it may be, "send an email to this address with what you want and I'll get you a total." Trying to avoid putting them on eBay, as I'll have to charge the eBay fees into the item's cost.

alecv
July 12th, 2016, 01:24 PM
I've got another batch from the glitch's gerbers by Sitopway. Thank you, glitch! Great design!

http://wiki.sensi.org/images/xtide3s.jpg (http://wiki.sensi.org/images/xtide3.jpg)

It's HASL not ENIG, just a reflection. These boards are not for sale, just for me and my friends...
It costs me about 6$ each.

I ordered also a bunch of brackets at local laser metal cutting service.
http://wiki.sensi.org/images/bracket1s.jpg (http://wiki.sensi.org/images/bracket1.jpg) http://wiki.sensi.org/images/bracket2s.jpg (http://wiki.sensi.org/images/bracket2.jpg)
(for XT-IDE V3, XT-IDE V2 and CF-Lite)
It costs me about 0.4$ each w/o coating and screw-thread.

glitch
July 12th, 2016, 01:30 PM
Very nice! Looks like Sitopway does a fine job making boards! Are the edge connectors gold plated, or is that also part of the reflection?

Robin4
July 12th, 2016, 02:56 PM
Those green pcb`s looks very nice!! And i hope the brackets having the same thickness like the original stuff and are really stiff..

Yes those ISA fingers are gold plated as well, that very good for the endurance on longer time.. Tin fingers will worn out much easily.

Will does brackets been pre-bend before selling / shipping?

I also would recommend to get off those sharp corners on the ISA connector .. So the board would pop-in in an ISA slot much easier.

How much would a bracket cost to produce with coating? (this coating is this some kind of paint, or a protection layer against rust?)

But it will be good if those brackets are protected against rust for longer durance.. This would add more quality to this nice product.

I hope someone can get this same kind of stuff on the same factory..

glitch
July 12th, 2016, 03:25 PM
But it will be good if those brackets are protected against rust for longer durance.. This would add more quality to this nice product.

You did read the bit about how they're for him and his friends only, and not for sale, right?

alecv
July 13th, 2016, 03:24 AM
glitch These board has tin plated connctor (HASL). My goal was "price optimization".
However gold plated PCB (ENIG) will cost about 8..9$, not a big difference.

Brackets have no coating, just a sanded seel plate (to protect from laser reflection) standard for this factory.
Rust would be a problem, yes, ater 20 years ;) I ordered 0.8mm and 1.0mm steel. The 0.8mm seems
too flexible especially for bracket with holes. 1.0mm is too rigid and hard to bent. 0.9mm would be good.

glitch, I've found some cosmetic problems:
- OSHW logo has low thikness
- ICs, Caps and R numbering is chaotic. U1 should be in the upper left corner.
- Block-jumper and resostor holes have different pad diameter
- J2 and J3 holes are small

For "price optimization" I've ordered parth from the Aliexpress. Most of ICs
are "second hand", pins are bent but still sufficient for this hobby project.

8P DIP Switch:
http://www.aliexpress.com/item/20pcs-8P-DIP-switch-level-DIP-switch-8P-2-54mm-pitch/32451991261.html

10K A10-103G Resistor network
http://www.aliexpress.com/item/100-Free-Shipping-exclusion-10p-10K-A10-103G-line-resistance-A103G-10-feet-A103J/32633287686.html

SN74F245N
http://www.aliexpress.com/item/20pcs-lot-SN74F245N-DIP-20/32476086046.html

SN74F573N
http://www.aliexpress.com/item/20pcs-lot-SN74F573N-74F573-DIP-20/32401673569.html

74HCT688N
http://www.aliexpress.com/item/10PCS-lot-74HCT688N-in-stock-can-pay/32624624325.html

Axial capacitors
http://www.aliexpress.com/item/Free-Shippping-one-lot-CT42-Axial-multilayer-ceramic-capacitor-104-50VM-100PCS/1625773814.html
rather expensive, standard capacitors are cheaper:
http://www.aliexpress.com/item/500pcs-0-1uF-100NF-50V-104-Multilayer-Monolithic-Ceramic-Capacitor/32434931188.html

glitch
July 13th, 2016, 04:53 AM
glitch These board has tin plated connctor (HASL). My goal was "price optimization".
However gold plated PCB (ENIG) will cost about 8..9$, not a big difference.

Ah, OK, I couldn't tell from the picture if it was HASL or what. The hard gold plating on the edge connector was of course the big expense with the run I did -- boards would have probably been $4-5 instead of $8 if I'd used HASL for the edge connector.


glitch, I've found some cosmetic problems:
- OSHW logo has low thikness

This looks fine to me on the PCB Cart run, but it did look way too thin on the prototypes I ordered. The shop I use for the prototypes uses an inkjet silkscreen process though, which I've never considered production quality.


- ICs, Caps and R numbering is chaotic. U1 should be in the upper left corner.

This is based off of the ordering in the schematic annotation. That's typically how I number parts, so I didn't think about people finding it confusing. I'll change it if it's a problem for people!


- Block-jumper and resostor holes have different pad diameter
- J2 and J3 holes are smal

That's intentional -- I use press-fit headers with a detent in the pin, the holes are sized so that they snap into the board and don't fall out when flipped over.l


For "price optimization" I've ordered parth from the Aliexpress. Most of ICs
are "second hand", pins are bent but still sufficient for this hobby project.

I use a lot of desoldered ICs for my personal projects, usually no problem there. All of the parts included in the kits and assembled boards for the XT-IDE rev 3 were new/NOS though, so that did add a little expense there. Still came out cheaper for the whole parts kit than just the ICs from Mouser!

Excellent work, I'd like to see pictures of the finished boards!

compu_85
July 13th, 2016, 05:22 AM
Still came out cheaper for the whole parts kit than just the ICs from Mouser!



Yes, I found the cost of the kits to be very reasonable. I just bought an Adafruit TV BeGone kit, and it cost more than the XT-IDE kit even though it's a smaller board with fewer parts!

-J

glitch
July 13th, 2016, 05:32 AM
Yes, I found the cost of the kits to be very reasonable. I just bought an Adafruit TV BeGone kit, and it cost more than the XT-IDE kit even though it's a smaller board with fewer parts!

Thanks, I tried really hard to get the price as low as possible without sacrificing quality :) Of course, I have a "ordered it for my day job" supply to tap in to -- resistors, caps, headers, screws, et c. usually get ordered by the 1K or 10K quantity. Like the HP/Agilent green LEDs used on the boards, those were part of a 10K purchase *years* ago, and they ended up being a few cents apiece because of that.

jltursan
July 14th, 2016, 10:40 PM
I'm going to purchase one of these v3 boards; but I'm still confused about it being totally compatible with Amstrad XTs (PC-1512/PC500, PC-1640/PC6400).
Has anyone any successful experience working with XT-IDE and these XT machines?

Robin4
July 15th, 2016, 04:45 PM
You did read the bit about how they're for him and his friends only, and not for sale, right?

Sorry i miss red on that. Question: Are you going order these parts with the same factory, or would that even possible?

glitch
July 15th, 2016, 06:49 PM
Question: Are you going order these parts with the same factory, or would that even possible?

Not sure I understand the question. You could buy all components from anyone who stocks Keystone's full line (that's who makes the ISA bracket), if you're referring to the components necessary to build an XT-IDE.

Robin4
July 22nd, 2016, 01:11 PM
Do you know the ID number or part number for that keystone ISA bracket that will fitt this Rev 3 version board? And do you also have any date when kits / or boards are available for sale?

rosaage
July 22nd, 2016, 01:24 PM
Do you know the ID number or part number for that keystone ISA bracket that will fitt this Rev 3 version board? And do you also have any date when kits / or boards are available for sale?

If you take a look inn the sales forum he has a thread there with sine prices and an email address.
http://www.vcfed.org/forum/showthread.php?53238-FS-XT-IDE-rev-3-Boards-Kits-and-Assembled-Units

pearce_jj
July 22nd, 2016, 01:45 PM
Keystone 9202: http://www.mouser.co.uk/ProductDetail/Keystone-Electronics/9202/?qs=g5fiFmky%2fl4bFM8ICVGfOQ%3d%3d