• Please review our updated Terms and Rules here

Cromemco dazzler replica project

As @daver2 says, there are IC locations in places in the Dazzler, especially in the sync generator circuits that must be the original 74 family to work. You will notice that in places they delayed the rise & fall times of some gate outputs by applying capacitors to them and the amount of delay there is significantly different if an LS part is substituted in for plain 74.

When I built the first Dazzler replica I was very fastidious about selecting all vintage TTL chips with early date codes, to be sure they were 100% genuine 74 TTL. The reason was that I had already experienced similar issues in TTL based video games where a pulse arriving 20nS early or late could cause trouble or not, so I had been down this road already and was familiar with this issue, but a number of people who built the Dazzler since didn't know about this issue and found LS parts easier to get and used those instead. Unfortunately over the years newer parts have been relabeled and sometimes they are functional equivalents but other logic families. So for this project I completely avoided any new parts from the major suppliers. Or ones that didn't look like new old stock, laser engraved markings for example.

You can detect a genuine 74 part because of the input pin current it sources, which is higher for 74 parts than LS. I also used the vintage type of crystal originally used which were popular in the color decoders that American TV's used at the time, the type Cromemco would have used. AS a result I had no trouble with my Dazzler and it worked first pop, but I have only tried a limited amount of software because the only S-100 computer I have is the SOL-20 which runs the 8080 and its code. So it is still technically possible that some program could upset my Dazzler too and that I simply don't know that yet.

One thing I would wonder about is if there is a defect in the gating that is allowing picture content to interfere with either video blanking or video sync time, and the problem is just happening to show up on the one program and not the others. Probably you have inspected the vertical interval on the scope in detail using a scope's delayed time-base function and you can check each field using that too for any signals that might be corrupting the vertical block. Of if there is some defect in the sync generator it may enable you to diagnose what is happening to it.

As for the color, the color encoder is super primitive and is not a suppressed carrier system (as it should be), but as far as simplicity goes it would have to win a prize, but it won't produce perfect colors, or white because there is always a truckload of residual carrier. Still when it is adjusted as good as it can be, the only color that looks significantly off is blue and it tends to look a tad purple as I noted in the article I wrote about Dazzler.
 
Hey all.
It turns out most of the vertical jumping/jittering was related to U72 (74LS04) on all three boards. Apparently the Dazzler doesn't like the 74LS04 ICs that I used, even though they work on other parts of the board and test fine with a Retro Chip Tester. They are labeled:
24ARSS8E
SN74LS04N
There is a stylized V or something like that to the left of the 24...

I still have about a one-line vertical jump about once a second, which I'm ignoring for the moment. Otherwise, all three board sets are working fine.

My vectorscope helped immeasurably when fine-tuning the colors.
Thanks to all!
Terry
 
I am on the motorway on the way home at the moment - so no access to my documentation. I have just visited Gary!

Was U72 originally specified by Cromemco as a 7404 rather than a 74LS04?

If so, it would be what we said - that some 'critical' devices HAVE to be none LS parts. Either that or sine LS parts will work (the faster ones) and some not (the slower ones)...

The chip tester only tests the function of the device, and does not even remotely attempt to test the device at speed.

Likewise, on other location on the board, the devices will be less susceptible to the speed (or slew) parameters.

They could also be substandard devices or 'fakes' of course, especially with the marking 'ARSSE' 😉!

I will check tomorrow myself.

Dave
 
Hey all.
It turns out most of the vertical jumping/jittering was related to U72 (74LS04) on all three boards. Apparently the Dazzler doesn't like the 74LS04 ICs that I used, even though they work on other parts of the board and test fine with a Retro Chip Tester. They are labeled:
24ARSS8E
SN74LS04N
There is a stylized V or something like that to the left of the 24...
These LSTTL parts with what looks like a "mermaid" logo have been the topic of several discussions in the S100Computers Google Groups forum. They look like this (photo courtesy of a poster to the S100Computers Google Group):

Counterfeit chips edited.jpeg


No one seems to know who produces these parts, but they are distributed by many eBay vendors as well several established secondary market parts distributors like Jameco and UTSource. I suspect they are factory rejects that are being picked up and rebadged by whatever company is behind this. At best they "sometimes work" and sometimes but not always will get a "pass" from basic TTL chip testers like the "Retro Chip Tester". My advice to everyone who finds an IC with this distinctive logo - even if they do "work" - is to throw them away. In the long run that save you a lot of troubleshooting grief.
 

Attachments

  • Counterfeit chips edited.jpeg
    Counterfeit chips edited.jpeg
    1.1 MB · Views: 9
Last edited:
I mentioned to avoid modern looking chips that have Laser markings for their identity, because they are re-labelled fakes most likely made post the year 2000 and may well be another logic family. A chip tester is useless to detect that. Most 74xxx and 74LSxxx chips from the late 1960's to late 1980's with those date codes, and an appropriate markings that are not fresh are genuine. After that lots of the original 74 TTL's were not re-manufactured because they had been replaced by initially by 74LS and then quickly by other cmos logic which was less power hungry. And now there is huge chip refurbishing business going on with transistors and IC's and somebody else decides what might suit your application, but the parts, though "chip tester equivalents" don't suit all applications. So even though you selected the correct part number for U72 you got caught out by the part not being genuine. I have never seen a 74 or 74LS part with Laser markings that are genuine.

For example if you sell any of the boards you have made, you may well get them returned with Egg on your face. Obviously there would be places/locations where a clone part of another family with Laser markings could be a good substitute, but a thorough study has not been done to confirm exactly where these locations might be yet and gates in a single IC are shared in the circuit sub-sections. So it is not worth the risk of using other than the exact vintage genuine IC's. If you have any other chips that don't look like genuine old stock, it may well be better to replace them now.




.
 
Yes. Dave asked what the spec was.

I wonder what they really are. HCT versions relabled ?
You could work it out with some tests that an IC tester does not do, but I would not waste my time. Because they are pretending to be something they are not and have been the subject of the refurbishing industry, and even if you could prove what logic family they belonged to, they are totally untrustworthy and belong in the bin because their unknown history likely includes the chance of a lot of electrostatic damage. You would be 100% better off to buy a vintage part in a Radio Shack package from 1985. If you put your mind to it, you can spot fake non vintage IC's from a mile away. There are many features that give the game away.
 
tfox, If your 74LS04's turn out to be mislabeled beasts and if the source of your jitter, I have some '70's to '80's LS04's. Truthfully, I also have period 7404's if push comes to shove too. If you are in the USA and will pay the postage and I can send 3 of which type you want off to you.
 
DeltaDon. thank you for the offer. I have some 74LS04s from TI that are working fine (except for a one-line vertical jump about once a second).

These three Dazzler boardsets are for myself and the person that I built them for. I don't plan on selling them to anyone. It wouldn't be right to sell them as original Dazzlers anyway, as the boards themselves are obvious reproductions.
Thank you all for your help!
Terry
 
So after VCFMW last year I told myself that I wanted to do a Cyclops exhibit next (this) year - but alas, I'm no closer now than I was then to acquiring a Cyclops or building a reproduction. So I figured in the meantime I could make a simulated version. Here's my first attempt to get something working. It's using a modern computer with a webcam. I wrote a C# program which is converting each frame to a 64x64 image within the Dazzler color palette, and sending the data encoded over serial to a Z80 ASM program I have running which basically just feeds the data into the display. I think it's working well enough that I will continue down this road so I at least have something new/fun for visitors to check out - at least I think they'll have fun waving at themselves. I also have tried the 128x128 mono mode with a few different dithering patterns. Each have tradeoffs but its fun to play around. At 9600 baud over serial I'm getting like maybe 1 frame every 4 seconds (0.25 FPS?). I am going to be switching over to a parallel connection and using the TU-ART, I think I should be able to get up to a full 30FPS then but we'll see. If this all works well, then I'll switch to some kind of full hardware version maybe with an ESP chip.

PXL_20250619_023627842~2.jpg
 
Last edited:
For those wo like to play around with the cyclops, here is a way for those who own neither Dazzler nor Cyclops. It is based on (my favorite) Z80/8080 emulation framework, the z80pack by Udo Monk.

It runs natively on Linux. All you need is a working z80pack emulation, and here are the steps:

1. Install the packages required for z80pack

For Fedora:

sudo dnf group install development-tools
sudo dnf install git SDL2 SDL2-devel SDL2_mixer SDL2_mixer-devel SDL2_image SDL2_image-devel


For Debian/Ubuntu:

sudo apt install build-essential libsdl2-dev libsdl2-image-dev libsdl2-mixer-dev

2. Get the latest z80pack sources

git clone https://github.com/udo-munk/z80pack.git

3. Apply two changes on the source code

Enable Cyclops by removing the comments in line 58 in file z80pack/imsaisim/srcsim/sim.h to

#define HAS_CYCLOPS /* has simulated I/O for Cromemeco 88 CCC/ACC Cyclops Camera */

Correct a typo in line 102 in file z80pack/iodevices/cromemco-88ccc.c (variable is t instead of t1) to

tleft = j * 1000 - (long) (get_clock_us() - t);

4. Enter the z80pack directory & build z80pack

make

This is the standard build option for X Windows, z80pack also offers a SDL2 based build option, which however currently won't work with the VIO emulation

If all packages in step #1 above have previously been installed, everything now should compile without errors.

Now you are ready to start the emulation. For your convenience, I have created a short recording:

The are already existing videos available on YouTube for the same demo, but there was no step-by-step tutorial yet.

One note for Windows users: z80pack now recommends utilizing the Windows Subsystem for Linux (WSL). If someone needs help on that, please give me a note.

Have fun

-Ansgar
 
So after VCFMW last year I told myself that I wanted to do a Cyclops exhibit next (this) year - but alas, I'm no closer now than I was then to acquiring a Cyclops or building a reproduction. So I figured in the meantime I could make a simulated version. Here's my first attempt to get something working. It's using a modern computer with a webcam. I wrote a C# program which is converting each frame to a 64x64 image within the Dazzler color palette, and sending the data encoded over serial to a Z80 ASM program I have running which basically just feeds the data into the display. I think it's working well enough that I will continue down this road so I at least have something new/fun for visitors to check out - at least I think they'll have fun waving at themselves. I also have tried the 128x128 mono mode with a few different dithering patterns. Each have tradeoffs but its fun to play around. At 9600 baud over serial I'm getting like maybe 1 frame every 4 seconds (0.25 FPS?). I am going to be switching over to a parallel connection and using the TU-ART, I think I should be able to get up to a full 30FPS then but we'll see. If this all works well, then I'll switch to some kind of full hardware version maybe with an ESP chip.

View attachment 1302982

Hi,

It would be neat to have an S-100 board that had a "Type A" USB connector on it to input the webcam video. Because of the scarcity of chips and especially the camera module, it does make more sense to have an ESP32 on the board doing the work ... and maybe doing the parallel data out to the S-100 buss over to the DAZZLER.

That kind of setup would also make it affordable for more people to build than trying to build a reproduction version.


.
 
The are already existing videos available on YouTube for the same demo, but there was no step-by-step tutorial yet.

Thank you for this! I've seen other YouTube videos but had no idea how to get it all set up like this myself, definitely going to give it a try!
 
Back
Top