Cromemco Cyclops 88 CCC board set reproduction

[hmb]

But, have you actually tried them yet in the application? They may well be just fine.

Unfortunately, I have had some interesting experiences with sellers that do not post a photo of the supposed old stock parts they are selling, because they don't actually have them, they acquire parts after an order by a customer is placed, and they have a thing called "the production department" where various devices that some engineers think are compatible, get put in costumes and makeup. So as soon as you have tested these parts in the application, can you report how they work ?

I've dealt with many fake and/or rebadged parts, but have little doubt as to the authenticity of the AMI S4008-9 sold by LittleDiode. My point in mentioning the lack of a photo on the vendor's web site was merely to make potential purchasers aware, and not to imply these parts were possibly fake or rebadged. It will be weeks before I have my Cyclops reproduction built so that I can test them myself. It is more likely someone else will beat me to it.

As for the Pomona 3301 enclosure - it is still in Pomona's current catalog. Mouser does not currently have any in stock, but they are accepting back orders. As mentioned above, DigiKey does have 2 in stock right now, and when they are gone I expect you will be able to place a back order from them as well. There are also other distributors of Pomona's products other than Mouser and DigiKey.

 

[Hugo Holden]

A question on the D mount lens mounting, ( I am familiar with C mount not D mount) was a thread cut into the case, or did it use a threaded mounting ring attached with screws, perhaps taken from a 8mm movie camera ?

And the other thing I was wondering, is about De-capping the IC, the better metthod to do it, presumably they just un-soldered the top ? and then also the issue of the die being exposed to humidity, and if it should have a glass window put over it ?

 

[hmb]

A question on the D mount lens mounting, ( I am familiar with C mount not D mount) was a thread cut into the case, or did it use a threaded mounting ring attached with screws, perhaps taken from a 8mm movie camera ?

And the other thing I was wondering, is about De-capping the IC, the better metthod to do it, presumably they just un-soldered the top ? and then also the issue of the die being exposed to humidity, and if it should have a glass window put over it ?

Based on all the photos I've seen of the original Cromemco Cyclops, the D-mount thread was cut into the case.

With regard to de-lidding the IC... there are several described methods. I did some digging into methods for de-lidding, and the safest description I found is to use a sharp knife to create a slit underneath the lid at one of its corners, just enough to be able to lift it. Then, to use pliers to roll/peel back the lid starting at that corner. Sort of like peeling the cover off a sardine can. I suppose de-soldering it with a hot air tool might work, but I have not seen it described for this purpose. I have also seen a description of just "knocking it off" with a chisel, but that idea worries me a bit.

Cromemco supplied the de-lidded chips with a glass cover glued on to replace the metal cover. Descriptions of the reproductions suggest the chip just be left uncovered, which should be okay since it is sealed inside the enclosure. Nevertheless, I have considering looking for an inexpensive source of small-sized microscope slide cover glass to glue on as a cover to keep out dust and humidity.

 

[Hugo Holden]

Based on all the photos I've seen of the original Cromemco Cyclops, the D-mount thread was cut into the case.

With regard to de-lidding the IC... there are several described methods. I did some digging into methods for de-lidding, and the safest description I found is to use a sharp knife to create a slit underneath the lid at one of its corners, just enough to be able to lift it. Then, to use pliers to roll/peel back the lid starting at that corner. Sort of like peeling the cover off a sardine can. I suppose de-soldering it with a hot air tool might work, but I have not seen it described for this purpose. I have also seen a description of just "knocking it off" with a chisel, but that idea worries me a bit.

Cromemco supplied the de-lidded chips with a glass cover glued on to replace the metal cover. Descriptions of the reproductions suggest the chip just be left uncovered, which should be okay since it is sealed inside the enclosure. Nevertheless, I have considering looking for an inexpensive source of small-sized microscope slide cover glass to glue on as a cover to keep out dust and humidity.

Thanks.

I was thinking that a good trick to avoid having to cut the thread would be to use a standard D to C mount adapter and machine off the threads for the C and turn it into a mounting ring . Then it would only require a clean hole in the case and some screws through the ring's perimeter.

https://www.ebay.com/itm/167915939030?

Also it looks like a D to Pentax Q adapter could be machined down for the task:

https://www.ebay.com/itm/194384788663?

 

[Hugo Holden]

I have been working on the software in the 88-ccc document, to try to get it into a form with hex values so I can assemble it.

I'll probably have a few questiions since this is always the most difficult part for me. The first thing I noticed on page 6, is about the control word for Output Port B: In the manual it says that bit 0 to bit 3 specifies the number of fields to be stored in memory. And that bit 4 and bit 5 (LSB & MSB respectively) specify the number of 2 mSec increments separating each field.

Going to the example program CCC-3 (to run the Cyclops with the Dazzler) on page 12, they load a value of 013 (Octal) = 0Bh to port B and say in the comments that this specifies 12 fields at 2mS spacing, but I think that would store 11 fields at 0mS spacing. I would have thought from the description of the port B and the bits, the value required (in Hex) to be sent out to port B would be 1Ch (or 034 octal) to get the 12 fields with the single 2mS spacing.

So I must be doing something wrong, or could there be a mistake in the document ? or is there something about the octal to hex conversions that I'm getting wrong ?

 

[daver2]

@Hugo Holden, do you want to post a link to the manual you are using for the software and I will take a look?

EDIT: This one: https://www.hartetechnologies.com/manuals/Cromemco/Cromemco 88 CCC Manual.pdf?

Dave

 

[daver2]

There is certainly an inconsistency within the documentation. Now, where it is...

I suspect that a value of 00 in bits 4 and 5 may signify 2 ms between frames.

I also suspect (like you) that the frame counter value in test program CCC-3 may be out by one.

The octal value should be 014 = 00001100 = XX_00_1100. 00 = 2 ms and 1100 = 12 frames.

However, it is also possible that the frame counter (like the interframe delay) is one frame if the value is 0000?

The only real way of finding out (without analysing the logic on board 1 in fine detail) is to try it, see what you get, and update the documentation!

D4 and D5 (the interframe gap) are stored in U33 (C and D latches). These feed the B and A inputs of a 74151 8 input mux (U12). With CBA = 000 at the input of the U12 8:1 mux, the output (at pin 6 /W) will be a '1' (because the input at pin 4 is '0').

Likewise, for D0 to D3 (the number of frames) are stored in U34. This value is latched into U24 when demanded.

Now we start to follow the snake...

I have some work to do, but I may come back to this to see if I can logically work out what is going on.

Dave

 

[Hugo Holden]

I have some work to do, but I may come back to this to see if I can logically work out what is going on.

Dave

Thanks Dave.

I will soon post the original publication of the "Octal" coded program and my interpretation of it, in the usual hex coded .ASM program that I hope to assemble.

I would be extemely grateful if you could check it and see if I have made any monumental foul ups (which you know with software I am prone to do) so that it is basically good, before I attempt assembly.

 

[daver2]

You could try and write it in assembler of course - old dog new tricks !

Dave

 

[nullvalue]

I knew Bill must have had something already compiled, so I asked him and and he sent me back this:

Here's something more useful. This is my assembler to read the ccc and put it on a dazzler... with bells and whistles:

http://wsudbrink.dyndns.org:8080/cyclops/ccc3fsr.asm

Note that "fsr" stands for "faster".

 

[Hugo Holden]

In the other example program that they give to store a single frame of 15 field per frame (on page 8) they send octal 017 = 0Fh to port B so at least that one makes sense. But they also say its 2mS, so maybe it is that the two bits, bits 4 & 5 for port B do code for one 2mS time, but it would be an odd way to do it.

 

[Hugo Holden]

I knew Bill must have had something already compiled, so I asked him and and he sent me back this:

Wow, that looks incredibly involved. For now at least I'll stick to the original Cromemco program and try to get that working first.

 

[Hugo Holden]

I have attached a first attempt. I put address references to include numbers that matched the octal memory locations in the original Cromemco listing so to reduce the chance of fouling up the jumps, but it appears that one address is decfective, the JNC at octal memory location 202 in the original program, it jumps to octal memory location 000,211, but there appears to be immediate data at this location 211, not an opcode. To fix that of figure out where it was supposed to jump to, I would have to understand how the repacking program works, I don't yet, that part of the program is hardly commented. Unless I have somehow messed up. Thanks for checking it.

 

[daver2]

I have printed out the program. I am at work at the moment and have to go out this evening, so I will take a more in-depth look tomorrow.

One thing I did notice is that the instruction at address 0C240 "LXI H" should be "LXI H,0". The operand is missing.

I do like the mixed use of Hex and Octal in the address!

You are correct regarding the "JNC 0C211". There is something strange there. Either it is a fault in the original CCC-3 listing or (as sometimes does happen) people combine instructions and jump into the middle of an instruction. In this case, the jump would be to opcode 10h (which is a DJNZ on the Z80 CPU and a 'do not use' on the 8080) - so this is not intended behaviour.

As you say, we have to understand what the intent of the program is before we can work out where the instruction should jump to in reality.

Back to my program at work...

Dave

 

[daver2]

I have converted the mnemonics from Intel to Zilog mnemonics (so I can understand the assembler source better) whilst retaining 8080 object code compatibility.

I can start to see patterns emerging in how the camera image bits may be being translated into Dazzler display bits.

But time will tell!

Dave

 

[Hugo Holden]

I received the board set, Thanks Aron.

I started to assemble the boards and on fitting the regulator, I noticed that on board 2 there was no foil on the solder side under the regulator, but looking further I think that there is more ground foil missing too, that should have extended around the board perimeter shown in Red. Board 1 looks ok in this respect. I'm just working on a way to try to remedy that.

I would guess that unlike the Dazzler, Cromemco did not publish the original foil patterns for these boards ?

 

[nullvalue]

Hey Hugo yeah, unfortunately I've not been able to find any foils. I've seen some photo evidence that they were distributed, but a complete set has never shown up online.

I started placing sockets in my boards yesterday and also noticed the same thing. Really annoying. Not sure how that happened since Board 2 was a copy of 1, but I'm guessing when I copied & pasted I must have missed the bottom layer an never noticed.

The good news is that both the top & bottom layers for that section are GND. I think the trace on the top layer is plenty thick as-is so I don't think it'll affect anything.

 

[Hugo Holden]

Hey Hugo yeah, unfortunately I've not been able to find any foils. I've seen some photo evidence that they were distributed, but a complete set has never shown up online.

I started placing sockets in my boards yesterday and also noticed the same thing. Really annoying. Not sure how that happened since Board 2 was a copy of 1, but I'm guessing when I copied & pasted I must have missed the bottom layer an never noticed.

The good news is that both the top & bottom layers for that section are GND. I think the trace on the top layer is plenty thick as-is so I don't think it'll affect anything.

Do you think all the tracks are accurate ?
I saw some odd things like vias placed between the fingers on the board. I have not seen that before on any S-100 board. I hope thay don't interact with the socket connections. I wonder how they got there.

 

[Hugo Holden]

Problems with pcb's seem to really bug me.

So I fixed the board 2 problem by adding some 0.4mm thick Brass plate screwed down with some 1.4mm Hex head miniature Brass screws threaded into the pcb. I had to make sure it avoided the slides on the side of the board.

The thing with S-100 boards, many had earth/common connections that were not so wonderful and any help they can get is good, so while it might work, I think its better to have the trackwork.

The thing that worries me somewhat; is if this big track area got missed, how sure are you that the interconnects between the IC's are all correct. Have they been manually checked one by one to be 100% sure, or did the schematic capture thing do it and then auto-route the boards ?

When I did the Dazzler boards I burnt the midnight oil to make sure it was all good. Even then, after many hours of study, I missed one pullup, but the boards still worked, because the pin assumed a logic high on its own.

 

[nullvalue]

Yes the vias between the pads should be ok - but I should have moved them - I moved them on Board 1 after the autorouter finished just because I didn't like the look of it but must have missed that on board 2.

Nice job on the ground foil fix.

Yes everything was verified against the schematic in Kicad using the DRC. I am confident it's electrically correct, everything is hooked up as it should be (according to the schematic - there's still a chance I did something wrong there!). But everything should be within generous tolerances and the Kicad DRC has never failed me yet.

The reason that big blob got missed is because Kicad doesn't consider it a trace - it's just a polygon on an electrical layer. So I copied it from Board 1 to Board 2, but missed one of the layers (there are 4 polygons all together and I must have just grabbed 3). The DRC didn't catch it because that regulator is already connected to GND on the front layer.