Eurorack modular 8085 computer

[MK14]

Just a few comments.

If (rather than "wasting" 5 inverters) you used 2 of the spare inverters, to (optionally) invert A14/15, you could fully map the memory board on the entire 64K.
I.e. You use on board jumpers/switches to choose which of the 4 16K memory blocks (by choosing the non-inverted OR inverted signals), activate the ram board.
But that is fine, if you are happy to leave it in that 16K segment.

Is 4K enough, there seems to be room for 4 x 2K = 8K. But 4K is fine, if that is what you want ?

The write line, probably needs a pull up (or something). Since when the switch/jumper/front-panel-switch is open (to disable writing), it seems to leave the write line, floating. Which could cause it to  sometimes corrupt itself, while writing is disabled.

[ChristofferB]

That's a really good idea! The more flexibility, the better!
Pullup'ing the /WR line is a good idea. I'd imagined using a switch between /WR and just +5v, thus pulling up that way, but that may not be logical when I find the schematic in a dusty attic in 20 years.
(or next week, for that matter)

Here's the revision:

--Christoffer

[ChristofferB]

And yeah, 4K might be a little on the small side, but adding in 4K more (even up to 16K) would be pretty easy, even on the same board.

The reason I'm using the 6116 2K ram chips is because I have a million of them, and would like to see them used. When I get to ROM, I'll go for 8K (2764) chips, but on a very similar board.

--Christoffer

[MK14]

And yeah, 4K might be a little on the small side, but adding in 4K more (even up to 16K) would be pretty easy, even on the same board.

The reason I'm using the 6116 2K ram chips is because I have a million of them, and would like to see them used. When I get to ROM, I'll go for 8K (2764) chips, but on a very similar board.

--Christoffer

That makes sense (2K ram chips). At the end of the day, you had to use something specific. Whatever you ended up using, could have still been criticized.
E.g. 1K ==> Too small
64K ==> Non-authentic for the era

You can't really win. I guess you just go with something, and that is just fine.

I think there was one early computer, that only had something like a total of 3 bits (approx), of RAM.
It is an amazing piece of technology.
From my memory, how it works is that there are three or four paper tapes.
They are processed simultaneously, 2 are inputs, one is the program, and the final one is the output/answers.
Hence it only needs a few internal bits of ram.
tl;dr
It processes the information, bit by bit, byte by byte, so it does NOT need to store the information inside the cpu.
I.e. The adder (or whatever the operation), receives the numbers to add up, directly from the paper tape bits. The output of the adder, goes straight to the output tape.
The program tape (loop) or whatever method it used (patch or something), can't remember, tells the cpu what the instruction is, such as add, subtract etc.

[ChristofferB]

Yeah, and I'd rather go with same-size chunks of memory. Decoding for  multiple sizes of memory on the same board would be a logical nightmare, I'd imagine.

You can't really win. I guess you just go with something, and that is just fine.

I'd prefer thinking that I can't really lose 

I think there was one early computer, that only had something like a total of 3 bits (approx), of RAM.
It is an amazing piece of technology.
From my memory, how it works is that there are three or four paper tapes.
They are processed simultaneously, 2 are inputs, one is the program, and the final one is the output/answers.
Hence it only needs a few internal bits of ram.
tl;dr
It processes the information, bit by bit, byte by byte, so it does NOT need to store the information inside the cpu.
I.e. The adder (or whatever the operation), receives the numbers to add up, directly from the paper tape bits. The output of the adder, goes straight to the output tape.
The program tape (loop) or whatever method it used (patch or something), can't remember, tells the cpu what the instruction is, such as add, subtract etc.

That sounds strange by todays standards! but there sure were many amazing architectures in the early prototype computers.

--Christoffer

[rrinker]

It is indeed a big pain, there are no commonly, reasonably priced hex displays and/or driver chips.

A long while ago, I managed to bag, a modest pile of these:



at reasonable cost. As I think they look REALLY cool, and are very quick and convenient, for such projects.

Additionally, I got some similar displays, but they are actually alphanumeric (I also managed to bag a modest quantity of those, as well, if I remember correctly, if I'm mixed up, then they are seven segment and/or hex, rather than alphanumeric).

I think your idea of slowing or even stopping the clock (as long as compatibility issues and any (if applicable, you could be all sram) Dram refresh issues, are resolvable), is a good one.

 Are those the HP displays? People in the RCA 1802 crowd will beat a path to your door to buy them from you, a pair of those are what was used on the original PopTronics Elf. My later iteration just uses regular 7 segment LEDs and some 9368's for hex display. Fun part is, a pair of those HP displays, or a pair of 9368s uses more power than the entire rest of the computer. The 1802 is all static CMOS, so it can be clocked to 0 and then restarted at any time. And draws microwatts.

 

[MK14]

It is indeed a big pain, there are no commonly, reasonably priced hex displays and/or driver chips.

A long while ago, I managed to bag, a modest pile of these:



at reasonable cost. As I think they look REALLY cool, and are very quick and convenient, for such projects.

Additionally, I got some similar displays, but they are actually alphanumeric (I also managed to bag a modest quantity of those, as well, if I remember correctly, if I'm mixed up, then they are seven segment and/or hex, rather than alphanumeric).

I think your idea of slowing or even stopping the clock (as long as compatibility issues and any (if applicable, you could be all sram) Dram refresh issues, are resolvable), is a good one.

 Are those the HP displays? People in the RCA 1802 crowd will beat a path to your door to buy them from you, a pair of those are what was used on the original PopTronics Elf. My later iteration just uses regular 7 segment LEDs and some 9368's for hex display. Fun part is, a pair of those HP displays, or a pair of 9368s uses more power than the entire rest of the computer. The 1802 is all static CMOS, so it can be clocked to 0 and then restarted at any time. And draws microwatts.

The cosmac (ELF, RCA)1802 is a real classic microprocessor. The 1802 + LEDs/Switches/Led-Hex displays in some projects and articles, make a tempting vintage/retro build. In a similar fashion to the OPs endeavors.

That photograph (of the display) is NOT mine, it was from the internet.
But yes, they are the HP ones, that I have got. (TIL311's I believe).

I REALLY, REALLY love those displays, and have no intention of selling them.

I can't quiet put my finger on it. But there is something about those displays, that is just so, so nice!
A kind of magic or secret source.

But there are many nice types of display.

Nixie tubes and VFDs (especially the multi-lined alphanumeric ones), are nice as well.

You were saying about it being CRAZY that the displays use more current than the rest of the system.

Well apparently the original Sinclair miniature TV, complete with tiny/miniature CRT screen. Used LESS electricity for the whole TV, than "just" the speaker and amplifier, of the time. (If I remember correctly, if wrong, it is a big proportion of the total current).

[ChristofferB]

The TIL-311's are really nice! I love those tiny early 70's LED dies. There's a local electronics store in Denmark selling them, at about 8£ each. That becomes a bit pricey- especially for 16 bit stuff!

The cosmac ELF is in my opinion the archetypal hobby computer. I managed to find the CDP1802 processor, in a regular electronics store, just like any other component. I thought at the time it mustv'e been one of the very last store sold 1802's, but it turns out they're still being made, for space /satellite stuff.

I've finished design of the 8K rom module, almost the same as the RAM card, but I'll post it here for future reference:

 

[MK14]

They are available for £2.80 each including worldwide shipping (apparently), on ebay, link below.
They may or may not be TIL-311 substitutes, I'm not sure.

http://www.ebay.co.uk/itm/NEW-Til311-DIS1417-Texas-Alphanumeric-Hexadecimal-Display-Logic-UltraCart-128/141924967748?_trksid=p2047675.c100005.m1851&_trkparms=aid%3D222007%26algo%3DSIC.MBE%26ao%3D1%26asc%3D20140122125356%26meid%3D270c7facf4974afcae28535ea6150335%26pid%3D100005%26rk%3D1%26rkt%3D2%26sd%3D281809099154

One would have thought that some Chinese clone/fake manufacturer, would be making them by now. I would have thought the technology in them is relatively really simple, by today's standard.
You could even use a really cheap, bare die pic (or something) and a pile of tiny LEDs.

[ChristofferB]

So they are! Nice find, definitely bookmarked!

If one can live with the 7-seg, I think making a small driver/lookup table in a PROM/ROM/PAL
would be a good alternative - probably not cheaper than £2.8, though.

--Christoffer

[ChristofferB]

Okay this one is a bit more tricky. I've completed design for a serial interface, using the 8250, but I'm not completely certain I got it right, as I've never used the chip before.
It still has the 74138 decoder with jumpers (or DIP switch), and else is pretty much just hooked up to the bus and line drivers.
I wont bother with a baudrate gen. chip, as the 8250 has a built in programmable divider.
I'm using MC1488/MC1489 as RS232 drivers, though I may change that to MAX232's, even though that's a bit anachronistic.
Any thoughts or design critique would be much appreciated!

--Christoffer

[MK14]

The Wiki seems to think the original 8250 had some bad bugs, and seems to be recommending the debugged version. The 8250B, instead.

https://en.wikipedia.org/wiki/8250_UART

The chip designations carry suffix letters for later versions of the same chip series. For example, the original 8250 was soon followed by the 8250A and 8250B versions that corrected some bugs. In particular, the original 8250 could repeat transmission of a character if the CTS line was asserted asynchronously during the first transmission attempt.

The +5 and GND power pins are apparently open circuit.

I'd be tempted to connect the /RI pin up to the RS232, using the (spare) available gates, for completeness.

Transferring long serial files, can take a boringly/annoyingly long time. Especially when you need to do it LOTS of times. Therefore:

The 3.072 MHz will work perfectly fine, up to about 38 Kbaud.

BUT if you went with the 8 MHz, it will still do all (or just about all), the baud rates up to 38 Kbaud.
But then, would also do much higher (such as about 3x faster), rates. Depending on what max baud rate you can sync up with, as regards the other end (PC ?).

tl;dr
Seriously consider an 8 MHz crystal (unless you need/want software compatibility with 3.072 MHz clock divider bit settings). Since it apparently will give you just about all the rates the 3.072 MHz would have given you, AND the options on much faster ones, which go to at least slightly more than three times faster.

For quick test development work, it can be more pleasant for the download/upload etc, to take 1 minute, rather than 3 minutes, unless you like drinking coffee, every 5..10 minutes (Joke).

Another idea
If you are planning on developing the code for the ROMS, yourself. I.e. you are going to keep on iteratively changing it, while writing the software. You may want to create a "ROM emulator/simulator" card.
Which uses CMOS battery backed up RAM, with write protect, and read(?)/write access via external serial interface.

This is to avoid having to continually program/erase (in a UV eraser) and change the ZIP socketed 2764's. On a regular basis.

Alternatively you could (temporarily) use flash devices, until your software (for the ROMS), is finalized.

But don't worry. Using actual EPROMS is not the end of the world, and perfectly viable. One gets use to keeping a pile of 2764's (in your case) in the UV eraser, and get use to swapping them between the erase, burner and development board(s).

[ChristofferB]

Thanks for all the feedback! Yeah, the power lines are obviously an oversight, it's getting late.
I actually only have the 8250, but they're pretty easy to find. I'll get the B version, or even spring for a 16550 (pin compatible).
You're right, when I have the spare gates, might as well add the /RI.  It just weren't connected in the app note I looked up, and it's not so widely used, I believe. Might be wrong, though.

Actually, going for 8 MHz clock would be a good idea, even though I might not have the CPU to utilize it now. If I add an 8088 processor card, it would be nice. And it makes no difference softwarewise/componentwise.

But don't worry. Using actual EPROMS is not the end of the world, and perfectly viable. One gets use to keeping a pile of 2764's (in your case) in the UV eraser, and get use to swapping them between the erase, burner and development board(s).

That's exactly what I usually do! The most important step, though is to label EVERY EPROM you've burned to keep track of versions.

An EPROM emulator card would be really handy, but I didn't really plan on using the serial port for loading large programs. More just talking to the board via a dumb terminal.

If you are planning on developing the code for the ROMS, yourself.

In the development phase, that's what I'll do. But then I'd like to be able to run other people's software (CP/M, for example).

The next card I plan is a centronics printer interface. It'd be useful for debug (same use as serial) but waay easier in hardware and software. I'm using a typewriter as printer, so it basically just needs 8 bit ascii (just a latch) and then a strobe (74121?).

--Christoffer

[MK14]

EPROMS are quite a nice way of developing things, in some ways.

E.g. If you create a "test" EPROM, which flashes all LEDs/IO on and off, and continually tests ram etc. You can always put it in at any time, if you are not sure if it is the hardware or software, which has suddenly stopped working correctly.

Or keep a previous/stable/known-working version of your software, in case you suddenly want to show someone your computer working and/or re-test the hardware out, to see if the latest problems are software or flakey hardware.

EPROMS don't seem to work very well (they are problematic), if you DON'T put a light (absorbing, opaque) label on them. Some cheap paper labels, don't protect enough against light problems (if I remember correctly). When fully enclosed in a light tight case, it does not matter so much.

When EPROMS were common place, they were theories and arguments, as to if unlabeled EPROMS, if left in/near direct sunlight, could be erased after a couple of weeks, or so.

There were also suggestions, that if you could NOT afford a proper UV eraser. You could leave EPROMS outside in direct sunshine, for a couple of days (approx), to erase them for free. I never did hear (or try/test) if that rumor/theory was true or not. The UK is NOT the best of countries (weather wise) to try that theory out, anyway.

[ChristofferB]

Well most EPROM's you find in gear has these metalized, it looks, labels.
I have never had any trouble with accidental erasure, but I'm a bit cautious about photographing a board with flash that has unprotected EPROM's. Might be nonsense, dunno.

I've by the way recently aquired nearly 200 intel 1702 EPROMS (the first commercial eprom, if I recall correctly), but they're 256 bytes and impossible to program, requiring 0V, +5V, -9V,+47V and +56V for programming. 

[MK14]

Yes, metalised labels are  the way to go.

Aren't 200x 1702 EPROMS worth a small fortune ?

The gold look ones, seem to sell for a fair bit each on ebay, and the plastic ones a lot less, but at 200x, that would be a fair amount of money.

[ChristofferB]

They might be, I don't know. But they aren't the white/gold ones, unfortunately. They're mid-70's and purple ceramics. I think I'm holding on to them, though, as I'm trying to buy a vintage Intel development kit/system that might be able to use them.

[MK14]

They might be, I don't know. But they aren't the white/gold ones, unfortunately. They're mid-70's and purple ceramics. I think I'm holding on to them, though, as I'm trying to buy a vintage Intel development kit/system that might be able to use them.

Yes, you can make an interesting vintage computer(s), out of them, and stuff.

I guess at 256 bytes each, you would need a hefty quantity, to make a somewhat decent amount of memory.

[ChristofferB]

Yeah, but being able to make a set of back-up ROM's for early systems would be fantastic. For me, that's always step #1 before even turning on any very vintage gear I find: back up the rom.

Just realized it could be interesting to have an ISA interface card for the system.
Shouldn't be impossible, most ISA signals are just regular cpu bus stuff. With an interrupt controller, it'd work fine, I think. They made one for the S-100 bus system, at least.

This would allow fairly easy interfacing of a VGA/CGA graphics adapter for getting a CRT monitor running with it, along with other things.

--Christoffer

[MK14]

I agree, if you lost the original ROMS, on the old vintage system. That is the one bit of it, which could be irreplaceable. So backing them up, makes sense.
Although you could simply back them up (using an eprom programmer, or something), onto a computer file. Then you can make/remake any ROMS as necessary.

i.e. You don't need to go straight onto 256 byte eproms, straight away. But you could do that if you want, I guess.

Getting ISA card interface working, would open up a huge number of hardware options. So great, if you can do that.

[ChristofferB]

No you're right. If you have a PC programmer, no reason for making physical backups that might also die. But if you have a non-pc programmer, that only loads from an old EPROM/PROM, and it's an EPROM like the 1702 or 2708, it may be the only alternative. (Some very early eprom burners burned directly from punched tape, I've read)

The ISA thing would definitely be handy if I go for an 8088 CPU card at some point. The more complex things like floppy drive controllers or hard drives could be very easily interfaced, you you could find an old ISA card doing the trick for you.

[MK14]

No you're right. If you have a PC programmer, no reason for making physical backups that might also die. But if you have a non-pc programmer, that only loads from an old EPROM/PROM, and it's an EPROM like the 1702 or 2708, it may be the only alternative. (Some very early eprom burners burned directly from punched tape, I've read)

The ISA thing would definitely be handy if I go for an 8088 CPU card at some point. The more complex things like floppy drive controllers or hard drives could be very easily interfaced, you you could find an old ISA card doing the trick for you.

I hadn't thought of that. An ancient EPROM programmer, with no modern interfaces or PC capabilities.

But I would have thought that either a somewhat modern EPROM programmer would read them and/or you could hack it by sniffing the address and data lines, to pick up the information.

I have heard of some EPROMS which use NON-STANDARD stuff. Such as some HP ones, for some of the HP computers (Very OLD, pre-PC era).

But I'm beginning to understand now. So if it is incompatible to vaguely modern EPROM programmers and what programmers you can get hold of, use ancient techniques, such as paper tape. Then short of hacking/sniffing out the data transfers, or something. I can see where you are coming from, now.

It's a bit like restoring an ancient, home/business computer. Which has floppies, but they are NOT PC standard ones. So if you lose the boot disks (floppies) and/or software which comes with it (e.g. games), they are potentially lost forever, or would be a major headache to get copies of them again.

So you might be forced to use (ONLY) the original machine(s), to make copies of the floppies. Which would be starting to deteriorate, by now, potentially.

[ChristofferB]

Yeah, exactly. Of course, there'll always be workarounds. I'm sure there's someone out there who's gotten an Arduino to read/write bipolar PROM's, but I just hate being stuck in that "do a project to be able to do a project to be able to.." so on loop.

Also, it's not as fun. The salvation is serial communications. A lot of old programmer gear will have a serial port. My bipolar PROM programmer/reader has one, and apparently were designed to talk to a PDP-11, so if you can utilize that port, a lot of solutions becomes possible.

But then you need to figure out the format, maybe write some custom hex file transmit/receive routines, and you're doing a project to be able to do a project again.

[ChristofferB]

I did some research on other people's ECB projects. Not that many it seems, but this guy actually made an 8 bit ISA interface, with minimal circuitry:
http://www.elektrik-trick.de/archiv.htm

Apparently, there isn't much to it at all. Makes sense, I just thought it'd require at least some IRQ controller.

[TonyD]

They are available for £2.80 each including worldwide shipping (apparently), on ebay, link below.
They may or may not be TIL-311 substitutes, I'm not sure.

http://www.ebay.co.uk/itm/NEW-Til311-DIS1417-Texas-Alphanumeric-Hexadecimal-Display-Logic-UltraCart-128/141924967748?_trksid=p2047675.c100005.m1851&_trkparms=aid%3D222007%26algo%3DSIC.MBE%26ao%3D1%26asc%3D20140122125356%26meid%3D270c7facf4974afcae28535ea6150335%26pid%3D100005%26rk%3D1%26rkt%3D2%26sd%3D281809099154

great find (sorry if that's a bit late)