PDA

View Full Version : Mark-8 Clone project



falter
October 13th, 2016, 10:19 AM
Figured I'd make a post about this like I did with my TVT. I kind of have a long term plan where I do projects of increasing complexity. I started with that little ASCII encoder, the main part of my TVT, an ELF, and eventually a Mark-8.

So this is years down the road probably, although now that I have the original circuit boards to compare I can try out my plan to make copies of the boards using the plans, just for entertainment.

So here are initial questions I have:

1) Are there any good sources for 1101 RAMs, particularly of 75 or earlier vintage? For some reason ebay seems to choke on searches for those.. inconsistent at best because a lot of unrelated stuff has 1101 in their part numbers. I note in many pictures you see Mark-8 RAM boards and the chips themselves are gloss black with just '1101' printed on top. I'm assuming those aren't intel -- anyone know who made those ones? Might make searching easier. What I've found so far is $17USD per chip for 1977 vintage.

2) Were there any other machines/devices that used the Signetics 8263s? I have 8267s of 1974 vintage so I'm good there, but for the 8263s 1977 seems to be a brick wall. I thought maybe I could search and scrounge circuit boards. The same question might apply to the 1101 RAMs.. not sure what other systems used those.

3) What is the diff with the 8008 and 8008-1 ?

4) Is this just a rebranded 8008? Would a hobbyist in the day have had access to these? http://www.ebay.com/itm/Microsystems-International-MF8008R-Gold-Intel-8008-Rare-Old-Cpu-4004-/322280581115?hash=item4b096b67fb:g:47gAAOSwdIFXyYl Q

So far the rest looks good.. got my question on Molex strips answered (thanks!).

Thanks again, looking forward to stumbling my way through this.

smp
October 13th, 2016, 11:18 AM
I would dearly love to obtain a set of copied boards.
Please let me know if you plan on doing a set of documentation so that copies can be made.
Will you be signing folks up for their own set once you have made the copies?

Thanks,
smp

mwillegal
October 13th, 2016, 01:08 PM
Keep in mind that if you want a full 4K system, you need 128, 1101 chips. That will be hundreds of dollars, just for the SRAM. Even if you scale back to a single 1K board, you might spend up to a couple hundred dollars. Here are some 1101 with a 1981 date code, but they have plenty in stock. Price seems to be slowly ciimbing on these, as I think I paid less than $3.00 each a few years ago.
http://store.americanmicrosemiconductor.com/mm1101a-a1n.html

8008-1 is faster, the -1 part can run to 1Mhz

The auction says that particular 8008 is a clone, hard to say if they were generally available. Price seems wicked high to me, I don't really know why 8008s command such a premium price on ebay, as they were made into the 80's.

regards,
Mike Willegal


Figured I'd make a post about this like I did with my TVT. I kind of have a long term plan where I do projects of increasing complexity. I started with that little ASCII encoder, the main part of my TVT, an ELF, and eventually a Mark-8.

So this is years down the road probably, although now that I have the original circuit boards to compare I can try out my plan to make copies of the boards using the plans, just for entertainment.

So here are initial questions I have:

1) Are there any good sources for 1101 RAMs, particularly of 75 or earlier vintage? For some reason ebay seems to choke on searches for those.. inconsistent at best because a lot of unrelated stuff has 1101 in their part numbers. I note in many pictures you see Mark-8 RAM boards and the chips themselves are gloss black with just '1101' printed on top. I'm assuming those aren't intel -- anyone know who made those ones? Might make searching easier. What I've found so far is $17USD per chip for 1977 vintage.


2) Were there any other machines/devices that used the Signetics 8263s? I have 8267s of 1974 vintage so I'm good there, but for the 8263s 1977 seems to be a brick wall. I thought maybe I could search and scrounge circuit boards. The same question might apply to the 1101 RAMs.. not sure what other systems used those.

3) What is the diff with the 8008 and 8008-1 ?

4) Is this just a rebranded 8008? Would a hobbyist in the day have had access to these? http://www.ebay.com/itm/Microsystems-International-MF8008R-Gold-Intel-8008-Rare-Old-Cpu-4004-/322280581115?hash=item4b096b67fb:g:47gAAOSwdIFXyYl Q

So far the rest looks good.. got my question on Molex strips answered (thanks!).

Thanks again, looking forward to stumbling my way through this.

falter
October 13th, 2016, 02:05 PM
I would dearly love to obtain a set of copied boards.
Please let me know if you plan on doing a set of documentation so that copies can be made.
Will you be signing folks up for their own set once you have made the copies?

Thanks,
smp

I'm happy to make more sets once I get it nailed down. I was planning to try off the plans using the originals as a size guide. Some have suggested scanning the originals professionally.. that is an option too but longer term due to cost and my paranoia about losing them. Already we had a near miss with my usually excellent mail forwarder.

I enjoy making PCBs.. I just hate that there isn't more vintage looking substrate out there.

falter
October 13th, 2016, 02:07 PM
Keep in mind that if you want a full 4K system, you need 128, 1101 chips. That will be hundreds of dollars, just for the SRAM. Even if you scale back to a single 1K board, you might spend up to a couple hundred dollars. Here are some 1101 with a 1981 date code, but they have plenty in stock. Price seems to be slowly ciimbing on these, as I think I paid less than $3.00 each a few years ago.
http://store.americanmicrosemiconductor.com/mm1101a-a1n.html

8008-1 is faster, the -1 part can run to 1Mhz

The auction says that particular 8008 is a clone, hard to say if they were generally available. Price seems wicked high to me, I don't really know why 8008s command such a premium price on ebay, as they were made into the 80's.

regards,
Mike Willegal

Yeah there are some 1977 vintage 1101s on ebay but asking price is $17.00 per. Ugh. $4-5 sounds better.. albeit for 1981. I'll keep looking.. I don't anticipate doing much with this unit so 1k probably is enough to demonstrate capabilities. It seems like a number of the surviving Mark-8s I've seen stuck to 1k or people built higher capacity boards.

Tor
October 13th, 2016, 03:35 PM
AFAIK the 8008 was 0.5MHz and the 8008-1 was 0.8MHz.

falter
October 15th, 2016, 07:17 AM
Could a Mark-8 run with an 8008-1?

Another question.. were date codes stamped on the bottom of 8008s a later on thing? I see many old looking 8008s that have sold for significant dollars but no date code. I'm wondering how you identify that on chips where they don't have the bottom marking.

falter
October 17th, 2016, 11:44 AM
I found some date code correct RAM.. C1101a back to 1968.. 32 of them. Expensive though.. $14 each! What do you guys think? Too early? All the Mark8s Ive seen this far where the mem board was shown used p1101 plastic chips.

Still no luck finding earlier N8263s. But I'll keep at it.

falter
October 17th, 2016, 05:02 PM
Woohoo! Found 5 Signetics N8263s, 1973 date codes. On a roll today!

glitch
October 18th, 2016, 04:38 AM
Yes, you can use an 8008-1 at 8008 clock speeds. You also don't need the skewed clock that Intel talks about in the datasheet, you just end up with slightly non-optimal timing and therefore slower execution speed (not sure how much, for 99% of things it probably doesn't matter).

Watch out for fake 8008s, there's enough of a market for them that supposedly relabeled ones have appeared. I purchased one from China years ago, when a Chinese seller had them for $20 shipped, and it seems to be both genuine and working.

From what I know MF8008 and U808 chips are fully 8008 compatible, but not Intel-produced silicon. I don't know if they're licensed second-source production or if they're pirate copies. I want to say the U808 was East German production, so I wouldn't be surprised at all if it was a pirate copy. The electrons probably don't know, though :D

mwillegal
October 18th, 2016, 01:07 PM
I have found that with the SCELBI, that the clocks (there are two), must approximate the data sheet specs or the system hangs. There is quite a bit of latitude, but the relationship between the two clocks should be considered or the processor will probably not run properly.

regards,
Mike Willegal

glitch
October 18th, 2016, 05:35 PM
I have found that with the SCELBI, that the clocks (there are two), must approximate the data sheet specs or the system hangs. There is quite a bit of latitude, but the relationship between the two clocks should be considered or the processor will probably not run properly.

regards,
Mike Willegal

It's been my experience that they do need to be strictly non-overlapping, but I've run non-skewed phases with the few 8008s I've got and have been able to run small diagnostic routines. Haven't gotten enough system built up to attempt to load BASIC or even a ROM monitor. My little project design is based off of schematics/reference designs from Microcomputer Design by Martin Research.

I'm not familiar with the Mark-8 or SCELBI schematics -- how difficult would it be to construct a RAM board using a modern(ish) IC like the 6116 or 6264, and avoid the 1101 cost issue?

snuci
October 18th, 2016, 05:55 PM
I'm not familiar with the Mark-8 or SCELBI schematics -- how difficult would it be to construct a RAM board using a modern(ish) IC like the 6116 or 6264, and avoid the 1101 cost issue?

You might want to take a look at the 8008 based computer I was able to acquire. I've scanned and published the documentation that came with it. It uses 2102's which are not too expensive and can be easily found. The schematics for the RAM boards are there. http://vintagecomputer.ca/bill-1-8008-computer-documentation/

Problem is, this computer uses a Intel C8201 clock generator and driver that is impossible to find. The Intel 1975 Data Catalog on Bitsavers (and only that year) lists it as:

8201 CLOCK GENERATOR AND DRIVER FOR 8008 CPU
The 8201 generates the two phase clock signals used by the 8008 CPU. Both TTL and MOS level signals are available. Only an external crystal is required for the 8201. A reset signal generator is also provided for power on or external reset requirements. The internal divider is selectable with the MODE line.

I'm not sure why that chip seemingly disappeared. Sounds like it might have provided solid clock signals for the 8008 but it was gone from the 1976 Data Catalogs and beyond.

Hope it's helpful.

Corey986
October 18th, 2016, 11:27 PM
The Scelbi 8B used 2102. I think it really has to do with when the machine was designed. 1973-74ish and it would be 1101, 1975-76 would be 2102. The 1101, is a very power hungry chip and with a full 4k, could heat a room. Trust me, you can see the heat coming off a fully loaded Scelbi 8h.

I spent a lot of time playing around with the clocks on the 8h with my logic analyzer seeing tolerances as a troubleshooted a bad 8008 that I finally gave up on and replaced with a spare. The 8008 does some weird things when the clocks aren't right, but they can be a little sloppy, enough that the Scelbi used a simple clock circuit with one shots and not a crystal. The Scelbi used milspec caps so that the clock wouldn't drift too much as the machine heated up.

Cheers,
Corey

glitch
October 19th, 2016, 06:38 AM
You might want to take a look at the 8008 based computer I was able to acquire. I've scanned and published the documentation that came with it. It uses 2102's which are not too expensive and can be easily found. The schematics for the RAM boards are there. http://vintagecomputer.ca/bill-1-8008-computer-documentation/

Very cool! I look forward to reading through that!

Clock generation shouldn't be too much of a problem with modern components. I'm currently using a RC-timed clock made up of some inverters and NAND gates, but a proper skewed clock shouldn't be too hard either -- I just haven't gotten past a protoboard, and optimizing the clock isn't a high priority for my little experiment.

I'm using a 6264 8K x 8 SRAM for memory, seems to work fine. I don't know if the SCELBI/Mark-8 bus is tied to some peculiarity of the 1101, but I'd imagine even if it is, a modernish SRAM could be adapted, for those who don't care about having something built 100% the old way. I'd even push for something newer than the 2102, since they're getting less common and, at least with the bucket of 2102s I've got, I'm starting to see higher failure rates in NOS devices. I've got enough projects on my plate that I'm not wanting to jump into yet another memory board for yet another system :)

In the end, it's just parallel address + data, and state signal timings. No magic, should be able to interface anything with acceptable timing parameters!

falter
October 19th, 2016, 07:38 AM
The Bill-1 looks really neat. Have you attempted to fire it up yet?

I've just ordered the C1101s and those N8263s. That way I have them *if* I go to build my original board set. Which is still very unlikely. I'd really like to find some plastic P1101as though to experiment with my clone. I have N8263s from 1977 also for that purpose, so I don't cry if there's smoke during the smoke test.

falter
October 19th, 2016, 08:25 AM
My Mark-8 board set has 4 1k boards. I wonder how heavy duty you would have needed to go on PSU to make that all work. I've looked at a lot of pictures of Mark-8s and seldom see any with more than 1K, I guess for that reason?


The Scelbi 8B used 2102. I think it really has to do with when the machine was designed. 1973-74ish and it would be 1101, 1975-76 would be 2102. The 1101, is a very power hungry chip and with a full 4k, could heat a room. Trust me, you can see the heat coming off a fully loaded Scelbi 8h.

I spent a lot of time playing around with the clocks on the 8h with my logic analyzer seeing tolerances as a troubleshooted a bad 8008 that I finally gave up on and replaced with a spare. The 8008 does some weird things when the clocks aren't right, but they can be a little sloppy, enough that the Scelbi used a simple clock circuit with one shots and not a crystal. The Scelbi used milspec caps so that the clock wouldn't drift too much as the machine heated up.

Cheers,
Corey

glitch
October 19th, 2016, 09:14 AM
My Mark-8 board set has 4 1k boards. I wonder how heavy duty you would have needed to go on PSU to make that all work. I've looked at a lot of pictures of Mark-8s and seldom see any with more than 1K, I guess for that reason?

Price was probably the main motivator, you can always split the power rails and use multiple supplies.

Marty
October 19th, 2016, 09:25 AM
Hi All;

When I built my Wire-Wrapped Mark 8, on some 44 pin Vector Boards, and I used the clock 7400 circuit and I made the 8263's out of 7400 type of Ic's and I used 6116 Ram memories, I had NO problem with it running and working..

THANK YOU Marty

Chuck(G)
October 19th, 2016, 10:03 AM
Marty, that's a good point about the genesis of the Mark-8. Back when, the object was getting hold of a reasonably-priced 8008; I suspect that numerous variations were constructed by hobbyists, wire-wrap included. So a set of PCBs is perhaps not as significant as some think. While I was still looking for an affordable 8008, the 8080 came out, so my own project was aborted--but I did have colleagues who did go ahead and "roll their own". I still have my faded copy of the 8008 publicity squib.

mwillegal
October 19th, 2016, 10:59 AM
The SCELBI 8H preceeded the Mark 8 by a few months. LIke the Mark-8, it supported 4K of SRAM using 1101 spread across 4 boards. It was followed a year later by the 8B which supported 16K of SRAM (2102) and/or EPROM (1702) also spead across 4 boards. The EPROM board could be ordered with MEA programmed into it. MEA was a complete integrated development environment and was used to develop SCELBAL BASIC for the 8008.

Actual power on the 8H is about .6 AMPs on the -9 volt rail, per 1K SRAM board. A 4K system would require about a 3 AMP -9 volt supply. Here is the text from the SCELBI HW manual.

THE BASIC SCELBI CARD SET CONSISTING OF ONE EACH OF: SCELBI #1100 CPU CARD, SCELBI #1101 DBB & OUTPUT CARD, SCELBI #1102 INPUT CARD, AND SCELBI #1104 FRONT PANEL CARD; REQUIRES A MAXIMUM OF 1.5 AMPS AT +5 VOLTS AND 100 MILLIAMPERES OF -9 VOLTS.

EACH “PAGE” OF MEMORY IN THE SYSTEM (256 WORDS) REQUIRES 200 MILLIAMPERES OF +5 VOLTS AND -9 VOLTS.

I use 5 or 6 AMP 5 volt supply for both SCELBI 8H and 8B.

regards,
Mike Willegal

snuci
October 19th, 2016, 02:52 PM
The Bill-1 looks really neat. Have you attempted to fire it up yet?

No, not yet. I want to save what is on the EPROMs first and have an ME1702A programmer that I am building. I need to get back to that as it's been on the back burner for a bit.

falter
November 3rd, 2016, 05:13 PM
Got some vintage ICs today:

34160

They appear to be legit. I'm stunned I found 8263s of the correct vintage. The RAM I'm not sure -- there's no obvious date codes, which if I read correctly means they are pre-73 (I read Intel started doing date codes that year). I'm not sure if a Mark-8 builder in the day would have used RAM that dated 4 or 5 years prior... but I guess it's better than being into the late 70s-early 80s.

I also got a bunch more late 8263s. So I now have 11 of them. :) I thought that chip was really going to be a bear to locate.