Hugo Holden
Veteran Member
Thanks!
- VCF South West - June 14 - 16, Davidson-Gundy Alumni Center at University of Texas at Dallas
- VCF West - Aug 2 - 3, Computer History Museum, Mountain View, CA
- VCF Midwest - Sept 7 - 8 2024, Schaumburg, IL
- VCF SoCal - Mid February 2025, Location TBD, Southern CA
- VCF East - April 2025, Infoage Museum, Wall NJ
-
Please review our updated Terms and Rules here
Cromemco dazzler replica project
- Thread starter Hugo Holden
- Start date
Gary C
Veteran Member
Hugo, do you have an image of the silkscreen for the component ID's
I have done as many as I can from an image of a board but some are obscured by components.
I have done as many as I can from an image of a board but some are obscured by components.
Thank you for that recommendation Hugo. Miniature PC mount trim capacitors are also becoming obsolete and new parts distributors stock only a very limited selection.Don't forget, apart from the rare IC, you will need a pair of these connectors for the ribbon interconnect cable, there are very few sellers with these now :
* NEW * 16 Pin DIP 0.3" Plug Male Header IDC Connector Flat Ribbon Cable 4 pcs | eBay
Find many great new & used options and get the best deals for * NEW * 16 Pin DIP 0.3" Plug Male Header IDC Connector Flat Ribbon Cable 4 pcs at the best online prices at eBay! Free shipping for many products!www.ebay.com
It pays to make a tool out of a pair of plates at least 5 or 6mm thick, or longer than the pins, with holes drilled, 2.5 to 3mm diameter or so where the pins are in one of the plates. The 16 holes can be marked with a prototype pcb board as a template. Then squeeze the connector assembly & ribbon cable together between these two plates in a small vise. This way it avoids any damage to the relatively thin pins on this connector.
Also for the IC socket on the Dazzler board in this location, it is probably better with a dual wipe than a machine pin socket to mate with this connector for multiple insertions. For mine I was lucky enough to have a pair of blue 16 pin dual wipe sockets with gold plated contacts, but these are pretty rare now and I have still not figured out who made them.
This is the variable capacitor:
NEW 2PCS CV31B250 IC CAPACITOR Trimmer Ceramic 350V VARIABLE , 7 TO 25PF 2-PIN | eBay
Find many great new & used options and get the best deals for NEW 2PCS CV31B250 IC CAPACITOR Trimmer Ceramic 350V VARIABLE , 7 TO 25PF 2-PIN at the best online prices at eBay! Free shipping for many products!www.ebay.com
I'll try to find where I got the Xtal.
My set of boards is on its way from LD Electronics. I am awaiting delivery to measure the lead spacing for the capacitors and trim pots - it's hard to do that accurately from Cromemco's original foil patterns. Would you happen to know the part number you used for the two trim potentiometers?
Hugo Holden
Veteran Member
The potentiometers are standard larger sized Bourns types that are vertically mounted. I did this so that the adjustment slots faces backwards to the board rear. Cromemco's original ones faced upwards, which is awkward when the boards are plugged in beside other boards.
It was these from Digikey, part 3386C-1-501LF
These are the correct version, the pins are in a linear array with 2.54mm spaced pins:
Also I think I used inductors from Jaycar, ones from Mouser, Digikey etc would be fine too.
22uH RF Choke | Jaycar Electronics
An RF choke is an inductor that blocks all High Frequency RF Signals and only allows DC Signals to pass through it.
www.jaycar.com.au
( It is easy to work out the spacing of any of the holes on the pcb, because the IC pins are exactly 2.54mm apart and it makes for an accurate scale).
When you get your blank boards from LD, can you take a photo of them and post here, it might be of help to GaryC' efforts.
Last edited:
Hugo Holden
Veteran Member
GARYC: The only diagram I have that includes the ink markings is the one I posted for one board on post #114, I will look for the other one for the other board.
Also, it is not important with the vias, but make sure where the component leads go that the board coating is also removed from the pads on both sides of the pcb.
Last edited:
Hugo Holden
Veteran Member
This is the exact crystal I used from this seller, so it is known to work:
I think this other HC-6U one here would also probably be ok, said to be for parallel resonance. The Dazzler circuit relies on series resonant mode, but I think it would still work in series mode & come into adjustment with the trimmer cap :
3-Pce lot of 3.579545 MHz Crystal in HC-6U case with wire leads + ground wire | eBay
Find many great new & used options and get the best deals for 3-Pce lot of 3.579545 MHz Crystal in HC-6U case with wire leads + ground wire at the best online prices at eBay! Free shipping for many products!
www.ebay.com
I think this other HC-6U one here would also probably be ok, said to be for parallel resonance. The Dazzler circuit relies on series resonant mode, but I think it would still work in series mode & come into adjustment with the trimmer cap :
(2) NTE358B Quartz Crystal 3.579545 MHZ - Lot of 2 | eBay
Item is in new old stock condition. See photos for detail and condition. Listing is for 2 parts.
www.ebay.com
Last edited:
Gary C
Veteran Member
Ah, found your images, that helps.
Good point. I need to expand the area that the coating is removed from.
Might need to upgrade to Kicad 6 as doing images in 5 is painful and the logo across the top looks a mess as just text.
But basically, the boards are all laid out now and its prettifying and labelling left to do.
daver2
10k Member
I have DAZCLS (Dazzler Clear Screen) working and I am starting to debug DAZPLOT (Dazzler Plot a point).
I don't have a working Dazzler card yet (I still have to build my cards) so I am running my code in a Cromemco Emulator and looking at the memory I have allocated to the Dazzler screen buffer.
I know it sounds daft, but to see the HEX values for the colour I have cleared the screen to and an individual RED, GREEN, YELLOW and BLUE pixel at the point I have actually specified is impressive...
Question: Does anyone have a working Dazzler with a Z80 CPU card and would be willing to perform some tests for me?
I will do a little more debugging first though...
Dave
I don't have a working Dazzler card yet (I still have to build my cards) so I am running my code in a Cromemco Emulator and looking at the memory I have allocated to the Dazzler screen buffer.
I know it sounds daft, but to see the HEX values for the colour I have cleared the screen to and an individual RED, GREEN, YELLOW and BLUE pixel at the point I have actually specified is impressive...
Question: Does anyone have a working Dazzler with a Z80 CPU card and would be willing to perform some tests for me?
I will do a little more debugging first though...
Dave
Hugo Holden
Veteran Member
Gary, I think you are doing amazingly well. Of course once you have figured this out, the first time around, the next time will be much quicker to complete the task.
Maybe we could form a team for vintage computer related PCB replicas. I can draw up the tidy replica pcb artwork required and you could convert it to Gerbers via Kicad. Then we make the Gerbers available for everyone on the forum to use a pcb house of their choice, to acquire the pcb replica.
Gary C
Veteran Member
Right, V1 of the Gerbers are now completed as far as I want to go with them at the moment. I have untented the vias so that they can be used for component mounting where necessary.
Free for anyone to use but be aware they are not tested.
If anyone wants to have a look and see if there are any obvious errors
Free for anyone to use but be aware they are not tested.
If anyone wants to have a look and see if there are any obvious errors
Gary C
Veteran Member
Gratuitous picture shot
silk
But I noted the TM silk on the top was misplaced in the zip files posted earlier
so adjusted for this image and new zip file included. The Kicad 3d viewed doesnt take notice of the instruction not to tent the via's so they look as if they are covered, but in the Gerbers, they aren't.
silkBut I noted the TM silk on the top was misplaced in the zip files posted earlier
so adjusted for this image and new zip file included. The Kicad 3d viewed doesnt take notice of the instruction not to tent the via's so they look as if they are covered, but in the Gerbers, they aren't.
Gary C
Veteran Member
Just a last comment, any updates I make will be to https://github.com/Gary-Clark/Cromemco-Dazzler
I have ordered a cheap as chips set from JCL to see if they are correct, but still don't know if the dazzler works with a Northstar horizon as that's the only S100 machine I have.
Anyone tried it ?
Except to add - Error in board 2 detected by JCL, two disconnected pads found. Errors corrected and gerbers uploaded to Github. WIll post when tested
I have ordered a cheap as chips set from JCL to see if they are correct, but still don't know if the dazzler works with a Northstar horizon as that's the only S100 machine I have.
Anyone tried it ?
Except to add - Error in board 2 detected by JCL, two disconnected pads found. Errors corrected and gerbers uploaded to Github. WIll post when tested
Last edited:
My understanding is that all Northstar Horizon configurations used dynamic RAM, and if so it is almost certain a Dazzler will not work in it.
Gary C
Veteran Member
Cheers.
Hugo Holden
Veteran Member
Are these a fairly standard S-100 bus though, could one not plug in a static ram card like I did in my SOL ?
like this or one of the CompuPro boards:
Artek Electronics 32K Static RAM S-100 Board for Altair, IMSAI, SOL-20, and More | eBay
Find many great new & used options and get the best deals for Artek Electronics 32K Static RAM S-100 Board for Altair, IMSAI, SOL-20, and More at the best online prices at eBay! Free shipping for many products!
www.ebay.com
Gary C
Veteran Member
I was thinking that.
Found a couple of unconnected tracks on the layout, can an admin delete my attachments ?
Found a couple of unconnected tracks on the layout, can an admin delete my attachments ?
Yes - if the dynamic RAM boards in the Northstar Horizon are replaced with static RAM boards then the Dazzler *might* work.Are these a fairly standard S-100 bus though, could one not plug in a static ram card like I did in my SOL ?
like this or one of the CompuPro boards:
Artek Electronics 32K Static RAM S-100 Board for Altair, IMSAI, SOL-20, and More | eBay
Find many great new & used options and get the best deals for Artek Electronics 32K Static RAM S-100 Board for Altair, IMSAI, SOL-20, and More at the best online prices at eBay! Free shipping for many products!www.ebay.com
As John Monahan discussed in this writeup:
there can still be other "issues". I suspect the Dazzler board does not have a wide tolerance for bus timing beyond what was originally specified by the design of the Altair and IMSAI 8080 CPU boards, hence the problems that article refers to with use of some Z80-based systems. Also, many "newer" static RAM boards produced by CompuPro don't work without modification - the CompuPro RAM17 board is apparently the "best" choice among CompuPro's boards but it is also one of the hardest to find. A good choice for a static RAM board are any of the older boards that were designed prior to the development of the IEEE-696 standard, with none of the enhancements that support more than 16 bits of addressing or 8/16 bit data transfer - for example, the Artek board that Hugo has provided an eBay link for.
Gary C
Veteran Member
So the Dazzler accesses 2K of memory on the bus via DMA and makes the CPU wait using Hold, but the RAM must be fast enough for the Dazzler ?
As the NS RAM cards do their own refresh (I think) then we need a 2K static RAM card ?, sounds like a new project
But, it really sounds as if I need to find a known compatible machine.
As the NS RAM cards do their own refresh (I think) then we need a 2K static RAM card ?, sounds like a new project
But, it really sounds as if I need to find a known compatible machine.
Last edited:
Hugo Holden
Veteran Member
I tried the Dazzler with three types of static ram boards in the SOL all work, Two boards were the Seattle Computer 16k and their 64k board (set for 48k) and the Compupro RAM 17.
No mods were required to make it work.
The Dazzler timing looks after itself internally, but obviously the section of RAM it is using must be fast enough as they specified in the Dazzler information and the computer must not attempt to access that RAM during the proceedings.
(I couldn't find anything that caused a conflict for the Dazzler in the SOL, but this was not the case with the Matrox video boards, where the SOL used a line on the bus in a way that upset them with a conflict where the output of two gates fought each other, I would have to look that up).
In any case, I think the reason the Dazzler boards sometimes didn't work (I've mentioned this before somewhere) is that there fundamentally were two versions. The late version of the board (the one I made and now GaryC) had the problem fixed and removed some troublesome bus connections of the early version. I would have to look back at the schematic. I think I concluded at one point that the early version would have given trouble in the SOL (and probably other computers). Though I think their schematic in the manual is the late version, but I found out about the other version/s from photos of original dazzler boards on the net and examining the tracks and I found the altered bus connections.
If you want a static RAM card and you don't want a large amount of extra memory, go for this 16k one, they have very elegant address decoders and work very well. They also have an interesting power rail distribution where there are insulated conductors that run perpendicular to the upper board surface. This can make the IC's more difficult to remove from the sockets, due to reduced access getting under them, unless you make a small tool:
Though I might have mentioned, on every vintage computer card I own I have replaced the pale blue CTS DIP switches. They are not sealed to the ingress of fumes or moisture. They get significant internal corrosion. And contact cleaner doesn't help. If the contacts have been open for some decades and then are selected closed, they make an unreliable connection and will give you a fault finding headache. I replace them with machine pin IC sockets and to those I fit new Omron DIP switches that have the form factor of an IC. The overall height ends up the same as the original blue CTS switch. I just did this on the Taylor-Wilson printer interface. Of course, there is nothing wrong with the CTS switches, when they are new, but >40 years later it is another story. One could argue its a good term of service, so I won't criticize them, it is an age thing. If these switches are used in modern equipment with an expected life of less than a decade, they would be just fine.
No mods were required to make it work.
The Dazzler timing looks after itself internally, but obviously the section of RAM it is using must be fast enough as they specified in the Dazzler information and the computer must not attempt to access that RAM during the proceedings.
(I couldn't find anything that caused a conflict for the Dazzler in the SOL, but this was not the case with the Matrox video boards, where the SOL used a line on the bus in a way that upset them with a conflict where the output of two gates fought each other, I would have to look that up).
In any case, I think the reason the Dazzler boards sometimes didn't work (I've mentioned this before somewhere) is that there fundamentally were two versions. The late version of the board (the one I made and now GaryC) had the problem fixed and removed some troublesome bus connections of the early version. I would have to look back at the schematic. I think I concluded at one point that the early version would have given trouble in the SOL (and probably other computers). Though I think their schematic in the manual is the late version, but I found out about the other version/s from photos of original dazzler boards on the net and examining the tracks and I found the altered bus connections.
If you want a static RAM card and you don't want a large amount of extra memory, go for this 16k one, they have very elegant address decoders and work very well. They also have an interesting power rail distribution where there are insulated conductors that run perpendicular to the upper board surface. This can make the IC's more difficult to remove from the sockets, due to reduced access getting under them, unless you make a small tool:
Seattle Computer Products 16 K Std S-100 Board -( ALTAIR,IMSAI,CROMEMCO) | eBay
Find many great new & used options and get the best deals for Seattle Computer Products 16 K Std S-100 Board -( ALTAIR,IMSAI,CROMEMCO) at the best online prices at eBay! Free shipping for many products!
www.ebay.com
Though I might have mentioned, on every vintage computer card I own I have replaced the pale blue CTS DIP switches. They are not sealed to the ingress of fumes or moisture. They get significant internal corrosion. And contact cleaner doesn't help. If the contacts have been open for some decades and then are selected closed, they make an unreliable connection and will give you a fault finding headache. I replace them with machine pin IC sockets and to those I fit new Omron DIP switches that have the form factor of an IC. The overall height ends up the same as the original blue CTS switch. I just did this on the Taylor-Wilson printer interface. Of course, there is nothing wrong with the CTS switches, when they are new, but >40 years later it is another story. One could argue its a good term of service, so I won't criticize them, it is an age thing. If these switches are used in modern equipment with an expected life of less than a decade, they would be just fine.
Last edited:
Additional to what Hugo said...So the Dazzler accesses 2K of memory on the bus via DMA and makes the CPU wait using Hold, but the RAM must be fast enough for the Dazzler ?
As the NS RAM cards do their own refresh (I think) then we need a 2K static RAM card ?, sounds like a new project
But, it really sounds as if I need to find a known compatible machine.
A basic no-frills static RAM board (which includes most of the early static RAM boards including the ones specifically mentioned by Hugo) will work in most cases. The board does not need to be exceptionally fast - if it is able to run with no wait states in an Altair 8080, it will be fast enough.
The problem with using dynamic RAM and a board that uses DMA like the Dazzler is:
- if a dynamic RAM board is busy doing a refresh cycle, the requesting bus master will need to wait before it can complete a memory read or write. The Dazzler board has no ability to wait for a RAM board that is not immediately ready for a data read cycle.
- many dynamic RAM boards are dependent on the ability to "steal" machine states when the bus is not being accessed - these occur frequently when a CPU board is the only device accessing RAM. The Dazzler board is accessing RAM on a constant regular basis, which may not allow the RAM board to "steal" a sufficient number of machine states to allow the RAM board time to perform its required frequency of refresh cycles.