PDA

View Full Version : Modern 64K RAM for Northstar Horizon?



smp
March 2nd, 2018, 07:10 AM
Hi all,

I am (hopefully) going to receive a new-to-me Northstar Horizon system sometime soon. It is fully populated with NS ZPB-A processor, MDS-A-D disk controller, and NS 64K HRAM memory. I've always wanted this particular system for some reason, and now I may have a chance to get my wish.

The memory board has been flakey in the recent past, and I have little experience with dynamic RAM in S-100 systems.

I'm looking to possibly cobble together a modern static RAM board for this system. The NS system needs to reserve $E800 - $EFFF for its boot ROM. Other than that, all of RAM should be available.

In the past, I've owned a CompuPro RAM 17 board which would fit the bill perfectly. Of course, that board is now long gone, and none seem to be available anywhere that I've been looking. If a good working RAM 17 were to come available, I would expect it to be priced exorbitantly.

Does anyone have any good ideas to share? My design days are long behind me, and my eyes and dexterity aren't anywhere near what they used to be, so I'm kind of nervous about embarking on a path that I may not be able to navigate. Besides, doing this on my own may be an expensive option. What are your thoughts and advice?

Thanks for listening...

smp

glitch
March 2nd, 2018, 07:20 AM
The HRAM series of DRAM boards actually work quite well in the Horizons with a North Star CPU board. It's one of the few DRAM boards I don't hate :) If yours is flakey, I have others I can trade you for. It's probably just a bad 4116, they do have a somewhat higher-than-average failure rate. I've got other DRAM and static RAM boards, too. The important thing is, the board needs to be able to open holes in the upper part of memory to make way for the floppy controller, since the North Star controllers don't assert *PHANTOM.

Once my S-100 FeRAM/SRAM board is finished, it'd be the perfect thing for a modern replacement:

http://www.glitchwrks.com/2016/03/29/ferroelectric-ram-part-1

Unfortunately it's behind a number of other projects, including the Altair 680 RAM board!

I actually just got a Horizon up and going over the weekend (mouse mess cleanup) so if you need any of your boards tested, let me know!

smp
March 2nd, 2018, 10:13 AM
Hi Jonathan, and thanks for responding.

Yes sir, it appears that maybe I'll be lining up for one of your S-100 memory boards - as well as the Altair 680 memory board. :D

I'll keep you in mind if I get too wrapped around my axle with this system.

In the meantime, I dug out an old unfinished project. I have this bare board from S100computers.com:

http://www.s100computers.com/My%20System%20Pages/PROM%20Board/PROM%20Board.htm

It appears that one could configure this board for a bunch of RAM, and probably make it so that the board only responds to $0000 - $E000. The addressing scheme is quite complicated, given all the possibilities that were designed into it. I'm studying it now. Maybe...

smp

deramp5113
March 3rd, 2018, 05:02 AM
In the past, I've owned a CompuPro RAM 17 board which would fit the bill perfectly. Of course, that board is now long gone, and none seem to be available anywhere that I've been looking. If a good working RAM 17 were to come available, I would expect it to be priced exorbitantly.


I’d definitely look for a RAM 17 or equivalent board. There are several 64K RAM boards that use the 2K x 8 static RAM chips like the RAM 17. They do come up on eBay now and then and tend to go for a reasonable price - though not necessarily “cheap.” I’ve picked up most of mine for $100-$120 I recall.

These boards are low power, reliable, most are easy to make holes in the address space as required for floppy controllers and video boards, typically support phantom, and can typically accept a 2716 or 2732 EPROM in a socket or two without any problem. This saves adding a separate PROM board. Very handy in a small chassis like a Sol-20 or Poly-88.

While on the later side of S-100 history and not necessarily the correct vintage for a Poly-88 or Sol-20, they are boards that an owner of such a machine might have upgraded to while still activiley using one of these machines.

By the way, while several of the boards have some limitation on how many and where in address space 2K block(s) can be disabled, most can be easily “tricked” into more flexible 2K disable arrangements.

Mike

glitch
March 3rd, 2018, 06:13 AM
I've also built up some "hole maker" boards, which assert phantom for a given address range, in order to get boards that don't support phantom working with memory that does. I use one with my Dajen SCI most of the time. If you're interested, I can write up how to do it with an IO-2 board, since most of the work is already done on the IO-2.

smp
March 3rd, 2018, 07:15 AM
I’d definitely look for a RAM 17 or equivalent board...

Hi Mike,

Thanks very much for your thoughts. Perhaps I'll get lucky on eBay and something will appear sometime soon.

smp

smp
March 3rd, 2018, 07:17 AM
I've also built up some "hole maker" boards, which assert phantom for a given address range, in order to get boards that don't support phantom working with memory that does. I use one with my Dajen SCI most of the time. If you're interested, I can write up how to do it with an IO-2 board, since most of the work is already done on the IO-2.

Hi Jonathan,

Thanks a million! Yes, I'd love to see your How-To with your IO-2 board. Do you have any IO-2 boards available?

Actually, do you think I might fit both the "hole-maker" circuit and also something like a Alliance Memory, Inc. AS6C1008-55PCN on it?

https://www.digikey.com/product-detail/en/alliance-memory-inc/AS6C1008-55PCN/1450-1017-ND/4234576

Thanks!

smp

glitch
March 3rd, 2018, 07:19 AM
I sure do! I recently had one scanned and reproduced. They're super handy little boards, I'd been scavenging vintage ones and reusing them, but now there's a ready supply of new ones. I'll draw up a circuit this week and put it together in the later half of the week -- have a business trip to deal with during the first part.

Also I'm pretty sure I've got a spare RAM17 or similar kicking around. I've got a number of 6116 2K SRAM boards in the "to be tested" drawer, as Mike said they're pretty much the best thing going for vintage SRAM boards.

smp
March 3rd, 2018, 07:21 AM
I sure do! I recently had one scanned and reproduced. They're super handy little boards, I'd been scavenging vintage ones and reusing them, but now there's a ready supply of new ones. I'll draw up a circuit this week and put it together in the later half of the week -- have a business trip to deal with during the first part.

Excellent!!!

I was editing in another question for you when you posted - please see my question above about also putting my RAM on the IO-2 board.

Thanks!

smp

smp
March 3rd, 2018, 07:29 AM
Hey Jonathan,

I'm looking at your shop on Tindie...

I'd like to purchase both the IO-2 and the parts kit, but Tindie shows that there'll be a separate shipping charge for each of the packages.
Is there some way that you could combine shipping and save me $5?

Thanks!

smp

glitch
March 3rd, 2018, 07:37 AM
IIRC Tindie does combine them as long as you add both to your cart at once. If not, I can refund the difference.

You should definitely be able to fit a SRAM on there -- I've built one up as a combo 1702A ROM/2101 SRAM board and had plenty of room for that:

http://www.glitchwrks.com/2012/01/30/io2-rom-ram

You can use the Intel 8212s as the data bus buffers, you won't need to buffer the address bus since a single CMOS SRAM won't present much load (probably less than the old TTL buffers I used on my ROM/RAM board!). You'll have to either waste half the SRAM or set up a banking system to use the full space.

smp
March 3rd, 2018, 08:29 AM
IIRC Tindie does combine them as long as you add both to your cart at once. If not, I can refund the difference.

OK! My order is placed. I tried it a couple of times, but I got charged shipping for each item, $8 for the board and $5 for the parts kit. I asked in the order for a refund of $5 and please combine the two items into one package, but I do not have anything that shows that in my receipt. Thanks very much for combining the shipping, if you please.

I'll be waiting for the details on the "hole-maker" circuit, when you have time to provide it.

Thanks!

glitch
March 3rd, 2018, 08:43 AM
I'll refund whatever the difference ends up being -- it'll likely be somewhat more than $8 to ship since it's got to go in a 12x9x3" box.

smp
March 3rd, 2018, 11:34 AM
Hi Jonathan,

I've been studying the manual and schematic for the IO-2 board.

I could connect a 128K x 8 bit RAM chip for use as 64K RAM in the S-100 system. Of course, I would only use half of it. I could connect the address lines A0-A15 directly to the S-100 bus, and as you already suggested use the two 8212s as data bus buffers. It appears to me that I might be able to use the existing on-board circuitry to un-select the RAM starting at address $E800 and up. If I set the address selection switches for $E8 (that's A15, A14, A13 and A11) then pin 6 of U1 will go high for all addresses $E800 and up. As well, if I pull the SM pad high, then pin 11 of U5 will go low for all addresses $E800 and up. That provides me both a positive and a negative chip select signal to possibly use on my RAM chip.

On the data bus side, I can tie D0-D7 together on the side of the 8212s not connected to the S-100 bus, and bring D0-D7 to the RAM chip.

Please check me on all this. If I'm on the right track here, then all I would need is some suggestions on how to properly connect the DS1/, DS2 MD, STB, and CLR/ pins on the 8212s for proper operation.

That would provide a very simple 58K RAM board for any S-100 system with a Northstar floppy disk controller installed. Given that the Horizon has nothing in ROM except for the bootstrap at $E800-$EFFF, that may be a good alternative for a system with a sometimes flakey 64K HRAM. And, it would still allow the top most 4K of memory to be available for a system monitor in some sort of non-volatile memory.

Thanks very much for listening. Please let me know if I'm going off the rails.

smp

glitch
March 3rd, 2018, 12:02 PM
Yep, that's pretty much it. Furthermore, if you do like I did with the ROM/RAM build-up and run the chip select outputs from the 74LS42 into a 74LS30 8-input NAND gate, you can switch on/off each of the 256 byte block-out pages in a 2K segment, giving you finer control over how much RAM is disabled. You could possibly leave some few pages enabled between 0xE800 and the ROM monitor for use a monitor scratchpad/stack area.

smp
March 3rd, 2018, 01:32 PM
Yep, that's pretty much it...

Thanks very much for your response, Jonathan.

I have a couple questions about the IO-2 board:

Looking at the board diagram and the schematic: I see the SI and SO pads on the board diagram, but I cannot find them on the schematic. Can you tell me where the SI and SO signals come from? I read your description about tying SM high, tying SI to SMEMR (47), and tying SO to MWRT (68) in order to generate the proper INP STB and OUT STB signals for reading and writing memory. (At first I thought that SI and SO may be SINP and SOUT on the schematic, but those are also S-100 signals, so you wouldn't be tying two S-100 signals together like that.)

Another thing: Can you confirm that I should tie MD and CLR/ high on the 8212s? It looks like the tri-stating will be properly accomplished by the BOARD ENABLE signal?

Thanks very much for all your attention.

smp

glitch
March 3rd, 2018, 01:53 PM
Figure 4 (PDF page 7) has SI/SO and SINP/SOUT. SI and SO are just pads that connect to pins 46 and 45 of the S-100 bus, respectively. My instructions meant to connect SMEMR to the point where SI would've connected, and MWRT to the point where SO would've connected -- in the decode circuit, not the two pads near the edge connector! Here's a high res pic of the board, for clarity:

http://i.imgur.com/aLG2LbUh.jpg (https://imgur.com/aLG2LbU)

And here's a huge gerbv rendering of the Gerber files, which lets you see the top and bottom traces at once:

http://i.imgur.com/bmKAiZT.png (https://imgur.com/bmKAiZT)

(you can click on the images for full size)

MD is tied high on one of the 8212s and low on the other -- it determines data direction, so you need to set it properly for which way the 8212 will be moving data. This is show in fig 5 (PDF page 8) of the user manual. *CLR can be tied high, it isn't used when you want to use the 8212s as bus drivers (it can be hooked up to PRESET or POC when you're using it as an I/O port).

Tristating is indeed done by the board select signal. In your case, you'll only want to assert board select during memory access that isn't being blocked out.

smp
March 3rd, 2018, 03:25 PM
Figure 4 (PDF page 7) has SI/SO and SINP/SOUT. SI and SO are just pads that connect to pins 46 and 45 of the S-100 bus, respectively. My instructions meant to connect SMEMR to the point where SI would've connected, and MWRT to the point where SO would've connected -- in the decode circuit, not the two pads near the edge connector!

<...snip...>

And here's a huge gerbv rendering of the Gerber files, which lets you see the top and bottom traces at once:

<...snip...>

MD is tied high on one of the 8212s and low on the other -- it determines data direction, so you need to set it properly for which way the 8212 will be moving data. This is show in fig 5 (PDF page 8) of the user manual. *CLR can be tied high, it isn't used when you want to use the 8212s as bus drivers (it can be hooked up to PRESET or POC when you're using it as an I/O port).

Tristating is indeed done by the board select signal. In your case, you'll only want to assert board select during memory access that isn't being blocked out.

Excellent! Thanks very much for the clarification, Jonathan. It makes sense now. Also thanks for correcting me on the 8212 configuration. Greatly appreciated.

smp

smp
March 10th, 2018, 10:17 AM
I received my replica Solid State Music IO-2 board and parts kit today. The board is gorgeous! As I'm looking at it, I'm feeling that I should just save it away, rather than going ahead with my project. I'm sure the feeling will pass. ;)

smp

glitch
March 10th, 2018, 10:58 AM
I received my replica Solid State Music IO-2 board and parts kit today. The board is gorgeous! As I'm looking at it, I'm feeling that I should just save it away, rather than going ahead with my project. I'm sure the feeling will pass. ;)

smp

That's the opposite of what you should do! :) The supply is effectively unlimited now -- I'll order more as there's demand for them. I find them to be extremely useful boards so I plan to always have some on hand.