PDA

View Full Version : Multi-Cards (8Bit/16Bit)



dr.zeissler
January 8th, 2016, 12:29 AM
Hi,

due to the lack of enough ISA-Slots I always wondered, why nobody offers new Multi-Cards.
There are project for XT-IDE, VGA, Network-Cards etc. but nobody has build a plattform-system
that is really flexible. Is it a hardware-problem that such things are not possible, or to expensive?

e.g.:

I own a T1000 RL/HD. It has a XT-IDE in the only 8Bit-ISA Slot.
Therefore a combination of an XT-IDE with a VGA or a SB or a MPU401 or a real in/out LPT-Port would be very nice.

dOC

3pcedev
January 8th, 2016, 04:05 AM
It's more of a supply/demand problem.

Sure it can be done (I have an EGA / parallel port / serial port combo card) but there just isn't enough interest to justify the production time/costs. IDE & serial / parallel ports are easy enough to implement, making a good VGA adaptor is harder but still possible. What eventually happens though is space runs out at the rear of the card to fit all of the connectors (i.e. DB15 for VGA, DB9 for serial, DB25 for parallel etc) and you need to branch them off with headers. The headers then usually connect to a ribbon cable which connects to a couple of connectors on a rear plate (taking up another 'slot space' on the computer).

Also you have the problem of person X wants CGA,Serial and XT-IDE, person B wants VGA, parallel and XT-IDE and person C wants EGA, no serial or parallel and XT-IDE. You could easily disable the serial / parallel ports but having multiple video adaptors on one card would be tricky at best. Also your effectively paying for functionality you probably dont want or need.

Personally I think it would be cool to see a new combo card produced; but given that vintage computing is primarily just a hobby I think it wont happen.

pearce_jj
January 10th, 2016, 01:09 PM
Also we quickly run out of board space whilst keeping through-hole components, meaning it would likely end up being a full length card. And those are VERY expensive to produce in small numbers (not to mention the software licensing).

mrcity
January 11th, 2016, 10:09 AM
Also we quickly run out of board space whilst keeping through-hole components

Any qualms about using SMD components? I know it won't keep that vintage feel, but I think it'd be way more practical, and opens us up to way more types of available chips that'll really shrink down the board footprint. As an exercise, one of the first things I'd like to do with the old wire-wrap ISA proto board I just found in an old warehouse is use some of the old SIMM memory lying around in my house as expansion for my 5150 to bring it up to 640K from 256K. One tiny lil' SIMM oughtta leave plenty of space for a whole lot else... :D

Chuck(G)
January 11th, 2016, 10:35 AM
Yes--I've had a couple of VCF projects shot down because it involved SMT components.

SpidersWeb
January 11th, 2016, 10:45 AM
I've found that squirting a line of solder paste + chisel tip soldering iron + slide down the pins= can do chips in seconds! It can be fun in it's own right once you've done a few projects. But I do hate fiddling with tiny fiddly little resistors etc.

For me, through hole components for 2-3 pin stuff, and SMD for IC's is actually the preference.

3pcedev
January 11th, 2016, 01:27 PM
I've found that squirting a line of solder paste + chisel tip soldering iron + slide down the pins= can do chips in seconds! It can be fun in it's own right once you've done a few projects. But I do hate fiddling with tiny fiddly little resistors etc..

That's fine when working with SMD that has a reasonable pin width/pitch. For example I have soldered a lo-tech EMS card using the method you mentioned above. That being said I also solder SMD at work using paste and a rework station which is 1000x easier and leaves a much neater finish.

When you get to SMD with a high pin count in a small package the soldering iron doesn't work. The mask is too thin and the pins just bridge out instantly. You can clean it up using some solder wick but it's very easy to end up with dry joints.

mrcity
January 11th, 2016, 07:46 PM
Both solid approaches for soldering SMD. Usually I use a blob of solder on the iron's tip rather than solder paste. The rework/reflow setup becomes necessary for QFN parts, which are real "fun" to get lined up by hand, especially if anything goes wrong after the first attempt. :sarcasm: I've had bad luck with reworking stuff though, since usually something ends up burnt or cracked. With practice comes refinement & elegance...

Anyway, it might help to figure out footprints for the various types of adapter circuitry (e.g. chips required & physical space) and then find some way to jam them all together onto a PCB on-demand to the user's request, or at least pre-define several key combinations. Probably even better is to simply make the PCB support as much stuff as possible but only populate components for that which the user desires. PCB cost is usually not high compared to your components, especially when you minimize variations on the PCB. Otherwise you're looking at a lot of art fees, setup fees, minimum order sizes per design, etc; just for a few dozen 5"x7" panels of one design, I've had it close to $1K to get PCBs made (not including parts or assembly) from a local manufacturer with decent turn time (which you'll want for at least the first few go-rounds given the superior quality & customer service). The Voltera (http://voltera.io/) could inspire a lot of folks to do small runs of this (plus all sorts of other) hardware, but it could still be a lot of money just to buy all the one-off parts you'd need to make it work... and then, does it ever work right on the first revision? ;) With all that, you still have make sure you aren't violating anyone's IP, and obtaining (let alone reverse engineering) any necessary firmware to run the adapter could be challenging, as pearce_jj suggested.

Hopefully, if a lot of people are averse to SMD, they can at least tolerate DIP-socket ATmega's or PIC chips handling most of the task. Looks like my HxC2001 SD/floppy emulator runs with a PIC chip, and it works great!

MarsMan2020
January 11th, 2016, 08:50 PM
Yes--I've had a couple of VCF projects shot down because it involved SMT components.

I think if you used GALs or a CPLD for things like address decode instead of 7400 logic, a lot of space could be saved and it might be possible to create a useable full-length ISA multicard with quite a few features packed in.

Ports are a harder problem. Maybe do something like a higher-density connector and an external breakout box to the actual connectors?

Chuck(G)
January 11th, 2016, 09:42 PM
Sure--and I do that on occasion. But folks need to program the GALs or CPLDs--and 5V parts are getting scarce. I recently did a project using an XC95108 (84 pin PLCC). I'd never dream of making that a project for the masses--just too hard to find parts, much less programming for them.

Face it--we live in a 3.3V (or less) SMT world. Remember when digital ICs were in TO-100 and flatpack?

MarsMan2020
January 11th, 2016, 10:20 PM
I don't think I was born in the TO-100 and flatpack days ;)

Atmel still has 5V SPLDs and CPLDs that are in production, and there are still Altera 5V FPGAs floating around... But I agree the 5V programmable logic seems to be going away.

I don't think I'd completely rule out a project with them, but it would probably need to a PCB + programmed chip deal to mitigate the supply and programming issues.

Chuck(G)
January 11th, 2016, 11:10 PM
I don't think that any of the Atmel parts are through-hole however. PLCC was a good compromise because you could use through-hole sockets or mount the package directly SMT-style. Well, you get the idea.

dr.zeissler
January 11th, 2016, 11:30 PM
It's more of a supply/demand problem.
Sure it can be done (I have an EGA / parallel port / serial port combo card)

You talk about the very rare Juko G7-B :)
https://th99.bl4ckb0x.de/v/I-L/50442.htm

pearce_jj
January 12th, 2016, 01:46 AM
Well, I do have about 50 XC9572s awaiting a use (5V parts).

MarsMan2020
January 12th, 2016, 08:35 AM
http://www.atmel.com/products/programmable-logic/spld-cpld/default.aspx:
ATF16V8/ATF22V10 - DIP/PLCC GAL replacements (~350 gates)
ATF750 - 750 gate CPLD in a 24-pin DIP or 28-pin PLCC
ATF2500C - 2500 gate CPLD in a 40-pin DIP or a 44-pin PLCC
ATF15xx series - 5V CPLD, 32-128 macrocells, available in 84-pin PLCC

The one big disadvantage I see with these parts is that the development tools provided are not as nice as provided by the other vendors.

John M. used the ATF1508 on the new S-100 386 board - http://www.s100computers.com/My%20System%20Pages/80386%20Board-II/80386%20CPU%20Board_II.htm

Gerardcjat
January 12th, 2016, 09:05 AM
Interesting subject !!
I am currently working on something similar.
I try to install ( solder ) an ALTERA TQFP 100 ( EPF7128, pitch : 0.5 mm ) on an square adapter with 0.1" pitch.
Idea beeing : Having a "standard mount CPLD" available for different printed circuit / card designs.
I first looked at the 84- pin PLCC but infortunatly I was short by 6 / 8 pins for the first design :-(( ...
so I had to go TQFP 100.
My first three attemps were not good, but were improving each time. ;-)
Next attempt will be with : paste / stencil / regulated hot air gun.
Waiting for the stencils to arrive ...

I did not mention, but of course, it is a 5V logic design ;-)

pearce_jj
January 12th, 2016, 09:19 AM
Try some FL-22 flux and solder wick for any excess, they come up pretty good very easily :)

Gerardcjat
January 12th, 2016, 09:31 AM
Try some FL-22 flux and solder wick for any excess, they come up pretty good very easily :)

I DID !! Last try : Fresh flux, fresh wick, carefully cleaned board ( gold plated !! ) .... but basically I had an aligment problem, mostly on one side ???.
You see, what needs to be more fresh is .... the operator ;-)

Chuck(G)
January 12th, 2016, 10:47 AM
You're not alone in fumbling with TQFPs. I've got a method that appears to work for me, even with a plain old TC soldering iron.

I first apply a bit of clear (although it probably doesn't matter--hot pink would work as well) fingernail polish to the underside of the package, and then use an inspection microscope to position the package accurately and allow the nail polish to cure (a little gentle heat speeds the process). I then "tack" each of the corner contacts in place with solder, then solder the remainder of the contacts and then clean up. For me, this seems to be foolproof.

Gerardcjat
January 12th, 2016, 11:13 AM
Ha Chuck !!!
Great idea, the fingernail polish and waiting it cures !!!
I will try it as soon as I can.
Are you using a classic optical microscope or an "USB" device like I do ??? ( a bit sensitive :-( must have a day without coffee BEFORE using it !! )
BTW, I am still wondering how I could have got misaligment ( pins almost between tracks ) MOSTLY on ONE side and the three other side correct ??

vwestlife
January 12th, 2016, 11:21 AM
I'd love to have a single card for my IBM 5150 PC which can do floppy controller + serial/parallel ports + clock chip + RAM expansion (to get the full 640K) all on one card. I've looked high and low but have never been able to find a single card which can do all this. So I have to choose between I/O + clock + RAM (like an AST SixPak) and a separate floppy controller card, or floppy + I/O + clock (like my Chaplet 7Plus) and a separate RAM card.

pietja
January 12th, 2016, 12:48 PM
The combination of floppy controller + XTIDE + serial/parallel ports + clock chip + RAM expansion should be possible with only through-hole components.
If you can settle with only one serial and one parallel port then there are ready made brackets (http://www.keyelco.com/product.cfm/Brackets-with-Mutiple-D-Sub-Cutouts/9200-17/product_id/2491) for those.

As for the component count, it is not that hard to use something like a GAL for address decoding which are through-hole, period correct and programmable with a basic EPROM programmer everyone should have to burn BIOS chips anyway ;)

mrcity
January 12th, 2016, 07:31 PM
I DID !! Last try : Fresh flux, fresh wick, carefully cleaned board ( gold plated !! ) .... but basically I had an aligment problem, mostly on one side ???.
You see, what needs to be more fresh is .... the operator ;-)

I dunno, you're probably still doing better than some particular contract manufacturers I used once; even with their "pick & place", we still had to straighten out dozens of TQFP32's (and remove probably hundreds of solder bridges) by hand after the fact... :curse:

Usually if there's a solder bridge, just apply a very small amount of flux to it and drag the soldering iron tip outward along the pins away from the casing. This way, it breaks the surface tension of the solder & the bridge comes apart. Also, if The Big Lebowski is on while you're doing this sort of rework, you must proclaim that you've found Jeff Bridges when you've spotted such a bridge. I'll be listening for it. ;)

I'm partial to regular ol' microcontollers like the ATmega328 which come in DIP flavor plus all sorts of SMD varieties, and also happily takes 5V. Maybe you could replace even a couple CPLDs & your ROM chip with an MCU with some nice ICSP headers? Not sure, haven't studied the exact circuitry involved with adapters just yet. However, it's not easy making exactly 4.77MHz from an output pin that maxes out at 8MHz. ;)

wesleyfurr
January 15th, 2016, 05:05 AM
Also we quickly run out of board space whilst keeping through-hole components, meaning it would likely end up being a full length card. And those are VERY expensive to produce in small numbers (not to mention the software licensing).

Software licensing? Different design software required or something? Probably a stupid question, I just don't know anything about PCB creation...

Thanks,

Wesley

wesleyfurr
January 15th, 2016, 05:11 AM
With regards to SMT solder...I asked for and received for Christmas, an el-cheapo hot air rework station. If anyone is interested, I can try to dig up a thread talking about them and possible pitfalls, but they sound decent for hobbyist use, and the price is right. Anyway, I then started looking to see about solder paste, but then I'm reading about it having an expiration date due to things changing over time (things bonding or sticking together on a very small level), even just from it sitting. Looks like a year or two is all a tube of solder paste is good for. Can anyone comment on that one way or the other? No more such soldering than I do, I would probably only use it once or twice before it was time to throw it out...in which case, this nice new tool I have many not be viable for me due to solder cost...

Thanks,

Wesley

Chuck(G)
January 15th, 2016, 08:33 AM
I've wondered about that myself, since I don't currently use the reflow solder paste--I wonder if simply keeping unused stuff in the freezer will extend the life.

pearce_jj
January 15th, 2016, 11:33 AM
Paste is a whole subject in itself. I'm yet to get any useable results from my reflow oven despite fitting a much more flexible controller module; I think the issue is the paste mainly. I find a simple iron, flux and multi core is fine - although slow.

Re software, personally I've settled on CadSoft Eagle for design, which has size limitations by license version. To go over 160x100mm it's about 1k; recouping that would be difficult IMO. There are alternatives, but re-learning the software is not something I have time to do unfortunately.

pearce_jj
January 15th, 2016, 12:16 PM
Just a thought... if someone that knows/has an unrestricted PCB layout tool could put together an ISA card 10x30cm that has on it simply 3x 2x32 pin headers spaced evenly across it near the bottom, plus three sets of holes higher up, we would potentially fit between one and three daughter boards to provide the user-required functionality.

These could be adapted easily from existing designs (in my case), just replacing the card-edge pads with 62-pins and keeping each within say 95x100mm.

Cost wise the 10x30cm board I guess might be 15 then the required daughter boards and components of top of course. Could use Keystone 9202 for the slot bracket.

pietja
January 15th, 2016, 12:42 PM
For designing bigger pcb's i could use Mentor Graphics Xpedition (https://www.mentor.com/pcb/xpedition/layout/) i use at work, but that is is not compatible with anything that is reusable at a budget level.

Solder paste come in many different types, think about the kind of flux and the size and shape of the powder/balls.
The most imported thing is keeping the lit on it if you have it in a jar so the flux wont evaporate.
Also minimize the amount of paste you but back in the jar because the past you used to used to cover the pcb has a lot less flux due to evaporation and you entire jar will dry out very quickly.
Its therefore best to store it in the refrigerator and let it warm-up before you use it.

Maybe its an idea to check what features everyone would like on a Multi-Card and design one big card for it.

pearce_jj
January 15th, 2016, 12:47 PM
I think the issue is that there will be a hundred different combinations proposed.

3pcedev
January 15th, 2016, 12:57 PM
KiCAD can handle pretty much anything. Its free/open source plus cross platform so you can run it on just about anything.

pietja
January 15th, 2016, 01:12 PM
What would you want in an "modern" XT system? i would think of the following list:
Floppy, IDE, serial, parallel, soundblaster, NE2000 NIC, RTC, USB storage and extended memory.

Almost everyone already has a graphics card in their system so that is not something you have to put on it.

All this is possible with of the shelf parts, some parts are only in SMD available so if you cant solder that you would miss those features (soundblaster, NIC and USB).
Also if you have more things on one card you have more slots left to use for other stuff. (scsi and midi)

If you count everything i listed you end up with 4 cards, that leaves one slot left on a 5150 and you would already have a beast of a system if it can handle all of the hardware resources needed.

pearce_jj
January 15th, 2016, 01:21 PM
One of the most common requests I get relates to machines with less slots - some only had one. Of course they typically also had more stuff onboard, like video, serial, parallel etc. But for those machines, some combination of memory/storage/network/sound would likely be useful.

wesleyfurr
January 15th, 2016, 04:10 PM
I've wondered about that myself, since I don't currently use the reflow solder paste--I wonder if simply keeping unused stuff in the freezer will extend the life.

From what I gather, keeping in the fridge at least (not sure about the freezer) will increase its shelf life, but it still expires. I was hoping someone here had some real-life experience...such as does it last much longer than that for our hobbyist type uses?

Thanks,

Wesley

mrcity
January 15th, 2016, 06:07 PM
From what I gather, keeping in the fridge at least (not sure about the freezer) will increase its shelf life, but it still expires.

I have a jar of Kester R520A RoHS solder paste that expired 8 years ago; I used it last on about a dozen panels' worth of PCBs this time last year with great success. It does help to keep it in the fridge, but our fridges at home are for food and beer (and the local Makerspace feels the same way). :D Each time I need to use this jar, I scrape all the dried paste off the top and then mix in a bit of solder flux to reconstitute the rest of the paste.

As for PCB editors, I personally prefer EagleCAD, but it looks like it'd cost at least $625 for a version that'll support a full-sized ISA card. You could try Fritzing; last time I messed around it was in 2014, and since I already had experience with a "real" CAD program, I found it fairly frustrating to use, as it's aimed for beginners.

wesleyfurr
January 16th, 2016, 07:52 AM
I have a jar of Kester R520A RoHS solder paste that expired 8 years ago; I used it last on about a dozen panels' worth of PCBs this time last year with great success. It does help to keep it in the fridge, but our fridges at home are for food and beer (and the local Makerspace feels the same way). :D Each time I need to use this jar, I scrape all the dried paste off the top and then mix in a bit of solder flux to reconstitute the rest of the paste.

That's good to hear. I was thinking of one of the squeeze tubes such that I could just squeeze it out of a fine tip onto the pads as shown some places online...that wouldn't work so well to mix some flux back in though. I may just take a shot at some and see what happens...and let you all know in a bunch of years!

LOL about fridges. :-) One of my other hobbies is old cameras, mainly Argus branded ones...though I don't shoot a lot. Whenever I find a good deal on film, I toss a little more in the freezer. It stops whatever slow decline in quality is going on and makes film last more or less indefinitely. So there are about two small boxes in our deep freeze...a little solder paste probably wouldn't make much difference to add to it. :-) That is assuming the freezer is a good thing...or better than the fridge.

Wesley