CV-NUC+ A Miniature CV Motherboard is in the Works

[+mytek]

I think it's time to give this project its own thread.

 

So what is it?

 

I'm trying to create a NEW Colecovision Motherboard that is only 4.5 x 4.5 inches square (11.43 CM).

 

Why you may ask?

 

Because that will match the form factor of a Mini Atari 8-Bit Computer that I previously created which was called the 576NUC+. Simply put, they will look very cute stacked on top of each other   Well it'll also be extremely portable, allowing one to toss it into a small bag with a couple of controllers and perhaps an AtariMax Ultimate SD Cart.

 

It Will Have a Case Very Similar to This (only styled to look a bit more like a CV)

Case design by @Mr Robot

 

 

Here's an initial look at the board layout to be used in this new project (EDIT: The image below was merely a test to see if all the chips would fit, it's been superseded by a new layout as of September 26th 2022).

 

MAIN BOARD (4.5" x 4.5")

 

Super Game Module (Plug-In)

 

Yes it has Super Game Module compatibility, as well as a built-in Pause switch with a flashing LED on the front panel to indicate it's being paused. The SGM aspect in this project can not be plugged into an existing Colecovision or ADAM without completely recreating a NEW board to make that possible. The hardware imitating this feature is only designed to work within the CV-NUC+, and nothing else. Point of fact is the circuit as written will not even work on the standard expansion port -- its been designed for internal usage ONLY.  It doesn't even utilize a programmable logic device, but instead uses discrete 74xx series chips in a unique configuration mostly of my own making (and a portion of it by @chart45), while benefitting from public knowledge presented on these forums and the ADAM Technical Reference Manual. I'd also like to thank @chart45 for sharing his example circuit and his willingness to answer my questions, @opcode for his written description of the theory of operation, as well as @fieroluke for his added discussion of his own implementation.

 

Oh almost forgot... @leaded solder 's blog about his DIY Colecovision was the thing that initially inspired me to pursue this project. And thankfully there are some great schematics by @ChildOfCv for the Colecovision Console which I have been making good use of.

 

I tried as much as possible to stick with THT components, but when it came to the plug-in SGM, there really was only so much room to work with and still have it fit the form factor I'm after. Although it looks like THT is feasible for most everything in the MAIN board, with one exception (FMS6400 Video Driver) .

 

I haven't yet settled on what switches will get used for ON/OFF and Pause.

 

EDIT: When I get the IC Chips all arranged in the best possible way on the MAIN board, then I'll layer in the capacitors and resistors. And just like what I showed on the Super Game Module, there will be a decoupling capacitor assigned to each IC of the MAIN board as well. A problem I still need to solve is to increase the density of the Super Game Module connection -- still need 10 more connections which I think I have a plan to accomodate.

[chart45]

humm i will buy your pcb in a blink of an eye if i can use a f18a on the vdp but i think it wont fit since its too close of the cartridge port. maybe if you flip the vdp it will not sure i have to mesure 

Edited September 26, 2022 by chart45

[+mytek]

11 hours ago, chart45 said:

humm i will buy your pcb in a blink of an eye if i can use a f18a on the vdp but i think it wont fit since its too close of the cartridge port. maybe if you flip the vdp it will not sure i have to mesure 

It's turning out to be every bit as difficult of a layout as the Atari 576NUC+ was, maybe more so because of that large cartridge. I also really want to keep the MAIN board as all THT this go around, since troubleshooting SMD is made so much more difficult because of them being soldered in place, and not having any easy way to grab onto a pin when scoping things out. Anyway to give you a better idea of where things are currently at...

 

MAIN BOARD

 

I think I've got every component accounted for, although I'll know for sure when I start ID'ing stuff and cross checking to my schematic (next step). I did have to move the TI Sound chip to the plug-in daughter board which made things a bit tighter for it, but I think it'll fly. And I added a right angle sip header to the backside to bring in the rest of the signals I needed. The other two headers are dual row 14 pin, and if the PCB is made with a 2mm thickness, the board fits snuggly between the individal rows and allows for them to be side soldered to the board as can be seen below. So this ends up being a tripod-like arrangement which should make the board very stable after being plugged in -- no need for any mounting hardware.

 

SOUND & RAM PLUG-IN DAUGHTER BOARD

 

When those pin headers get plugged into their female counterparts, the overall stacking will allow the proper height for clearance above the MAIN board.

 

When I did my rearrangement, I pushed all the analog video processing away from everything else so as to minimize digital noise pickup and crosstalk. So that area behind the cart is dedicated to video. I can make the lower 40 pin socket be the VDP if that helps with the F18 installation?   EDIT: Although I'd prefer to have the VDP chip just below the cart, with the Z80 below that.

 

Note: For those that may be wondering why is spacing so tight on all of this. It's because I wish to have it be pretty much a drop-in fit for an existing enclosure design, only requiring small changes to accommodate this new system.

[chart45]

maybe i couluse a small pcb to move the f18a like the adam... f18a is a must for me

[digress]

the MK2 f18a would fit anyways. It's much smaller footprint. perhaps put a socket in place where the vdp would go so it's an easy install if people want to the f18a  vdp instead.

 

https://dnotq.io/f18a/intro.html

 

[chart45]

10 minutes ago, digress said:

the MK2 f18a would fit anyways. It's much smaller footprint. perhaps put a socket in place where the vdp would go so it's an easy install if people want to the f18a  vdp instead.

 

https://dnotq.io/f18a/intro.html

 

when it will be aviable... its been a while 

[+mytek]

28 minutes ago, digress said:

the MK2 f18a would fit anyways. It's much smaller footprint. perhaps put a socket in place where the vdp would go so it's an easy install if people want to the f18a  vdp instead.

The PCB is being designed with just enough spacing to allow sockets for everything including the VDP chip

[+mytek]

The MAIN board and the schematic have both been ID'ed and cross referenced, and as I suspected some stuff was missing, and other stuff needed to be moved around. Some things got moved due to interference problems with other components (I print out scale drawings and drop real components on top to check fit).

 

So other than still needing to route traces, I'm feeling pretty good about this layout...

 

 

I took extra pains to make sure that components that got nested inside a chip footprint wouldn't interfere with the plastic cross bars in the socket. Also no stacking of additional sockets for extra height will be required.

 

The only thing that I'm not 100% certain of is inductor L1 nested inside U3. It might need to be soldered to the underside.

 

[+mytek]

MAIN Board has been 100% Routed and Verified Accurate to Crossed Check Netlist 👍

 

TOP

 

BOTTOM

This will be a 4-layer PCB when manufactured, 1.6 mm thick, 114 mm square.

[Captain Cozmos]

Just curious as to why not just set it up to access a full 32k and install an updated rom.
I would buy this just to develop 32k native based games.

Mind you I have not been following this project as well as others and the subject may have already been brought up and answered so please forgive the question.

 

If you have already decided to get rid of the ability to use the roller controller and steering wheel which also does away with the super controller spinners then one would conclude that maintaining the original system is not a priority.

 

[+mytek]

5 hours ago, Captain Cozmos said:

Just curious as to why not just set it up to access a full 32k and install an updated rom.
I would buy this just to develop 32k native based games.

Actually it does have the full 32K ROM and 32 RAM mapping as well as the AY-3-8910 sound chip (and the original TI sound chip) on a separate plug-in board on the right side. Basically full on Super Game Module support built-in. I'm also going to have a built-in 12-key pad in case someone wants to use alternative controllers that don't have an integral keypad.

 

BTW, I'm not doing this for anyone else, nor will I be selling it. However when I'm done testing, everything will get released into the public domain (Schematics, BOM, PCB Gerbers, and case STLs).

 

5 hours ago, Captain Cozmos said:

If you have already decided to get rid of the ability to use the roller controller and steering wheel which also does away with the super controller spinners then one would conclude that maintaining the original system is not a priority.

Yep that stuff had a very low priority, and as you can see there simply wasn't any room for the extra circuitry to support it. Not really that much stuff was done that absolutely needs it anyway. And if a person just has to have it, well they can always use their 'real' CV console for that

 

BTW, the BIOS ROM socket is actually configured for an 8K EEPROM for easy updating (New Stock in current production -- no need to use eBay or AliExpress for this one).

[Falonn]

Wow, that is packed in there really tightly! 😃

 

16 hours ago, mytek said:

New Stock in current production -- no need to use eBay or AliExpress for this one.

Which model EEPROM were you able to find that is 5V tolerant and still being made today?  I've done that same search a couple times and come up empty handed.

 

[+mytek]

4 hours ago, Falonn said:

Wow, that is packed in there really tightly! 😃

Yes it sure is, but still enough clearance for sockets

 

4 hours ago, Falonn said:

Which model EEPROM were you able to find that is 5V tolerant and still being made today?  I've done that same search a couple times and come up empty handed.

Microchip AT28C64B-15PU (DataSheet).

 

Digi-Key P/N: AT28C64B-15PU-ND 1,600+ in stock @ $5.55 each

 

Mouser P/N: 556-AT28C64B15PU  123 in stock @ $5.84 each

 

I used this same chip on another project a while back, XEP80-II and it worked well. In fact I needed two, one for the OS ROM and the other for the Character ROM. Flashed with a TL866II Plus.

[+mytek]

Update: Change of Direction for Super Game Module

 

It was simply too tight for PCB layout using discrete chips, although I had reduced the number of chips down to 5 for the SGM glue logic. But this still wasn't enough, and I really needed to have a reason to tackle the PLD aspect. So out with the 5 chips, and in with a single SPLD. Routing appears complete at this point.

 

CV-SGM FRONT SIDE (PLUG-IN MODULE)

 

CV-SGM BACK SIDE (PLUG-IN MODULE)

 

CV-NUC+ FRONT SIDE (MAIN BOARD)

 

CV-NUC+ BACK SIDE (MAIN BOARD)

 

I have two more boards I want to layout, and then off to JLCPCB to get some samples made

[+LarryL]

Looks beautiful!

can’t wait to build 🙂

 

@mytek can you already share a preliminary BOM?

I know that this is subject to change, but I would like starting to check/source some components - some of them might take some time if only available in China…

 

cheers

Michael

[+mytek]

1 hour ago, LarryL said:

Looks beautiful!

can’t wait to build 🙂

Thank you

 

1 hour ago, LarryL said:

@mytek can you already share a preliminary BOM?

I know that this is subject to change, but I would like starting to check/source some components - some of them might take some time if only available in China…

Not yet, because it only exists in the form of a multitude of orders from independent vendors, some of which contain parts that I no longer need as the project has evolved. Anyway I would prefer to wait and do something a bit more official, and that will be forthcoming in due time. Heck I have yet to make 'real' boards for testing, so putting out a BOM or schematics at this point would not be good in my mind. However thank you for expressing your interest 👍

 

P.S. once I have a fully working prototype, the schematics will get released, with the BOM following close behind.

[Ruggers Customs]

@mytek I've been following your work for a while.  I'm extremely impressed and can't wait to see finished the products.  Excellent work sir!

Edited October 5, 2022 by Ruggers Customs

[+DrVenkman]

Looks great, Michael. As usual.  

[+mytek]

I just got done testing out the circuit and the PIC firmware to allow a separate keypad to be connected to my CV-NUC+ board (there is already a 5 pin header intended for this purpose). The keypad will be mounted in the lid of the case directly in front of the cartridge port. Its purpose will be to allow controllers without integral keypads to be used (e.g., Atari, Sega Genesis). Because of the odd diode matrixed keypad that the CV uses, a translation was required to go from a standard off the shelf telephone style 4x3 matrix keypad over to the CV's way of doing things. Since I'm very familiar with using Microchip PIC MCUs, I figured this was the easiest way to go from point A to point B in this project, and requires minimal hardware to pull it off.

 

Here's the KeyPad I'm planning on using.

SparkFun P/N: COM-14662

Price: $4.95

 

Here's the schematic for the translation circuit.

 

 

And here's the PIC Firmware: CV-KeyPad.hex

 

My bread boarded prototype worked first time out of the box, so now I'm planning on making a PCB based on this design that mounts directly to the back of the keypad via its 9-pin header provision. The ICSP connection is purely optional if the PIC16F630-I/P chip is socketed, in which case it can be flashed out of circuit with a TL866II Plus programmer.

 

Although this is intended for my CV-NUC+ project, there is no reason that it wouldn't also be usable for anything else CV related

 

EDIT: There is a requirement to supply +5V and GND via the ICSP header (J3) on pins 2 and 3. I have added a small two pin header for this purpose on my CV-NUC project, but it is something to keep in mind for applying this to other systems.

[+5-11under]

Very cool! I still have a couple of the keypads with independent switches, but those seem impossible to find these days, at a decent price, at least.

[+mytek]

On 10/6/2022 at 12:59 PM, 5-11under said:

Very cool! I still have a couple of the keypads with independent switches, but those seem impossible to find these days, at a decent price, at least.

Yep when I decided that I would have an integrated keypad in my new console design, my first thought was where can I get a reliably sourced keypad, and then my 2nd thought was I won't find such a thing because of the odd electrical switching requirement. So I started looking at what was available that would match the keys required. Luckily Coleco mimicked the standard telephone style 12-key layout, so that aspect was easy.

 

It turned out that SparkFun offered a few different keypads matching the layout, with the one I picked being at the lowest price point, while still having real individual keys (not a membrane) and easy panel mounting.

 

Having so few components made for an easy PCB layout, and I already have OSH Park making me some as I write this post.

 

1.4 x 1.2 inches --- 3 boards shipped for $8.40

 

Assuming this board works when assembled and tested, I'll put up a share link to purchase the boards or download the gerber files. As I stated earlier this will work stand-alone, and is not specific to my mini CV project (check out the schematic and you'll see where things get wired on the controller ports). Not really much of BOM is required, besides the two chips shown above, it requires one 0.1uF capacitor, four 10K 1/4 watt resistors, and some 0.1" pin headers.

 

EDIT: Mouser Electronics has plenty of both chips in stock: 595-CD4066BE, 579-PIC16F630-I/P

 

EDIT2: I think I'll relabel the header NUC to JOY on the final release.

 

EDIT3: Final Version has been Released. Go HERE to find out more.

[+mytek]

SGM PLD Progress Update

 

Well I just couldn't get everything to fit in the ATF22V10 SPLD, so I had to move up to a CPLD device also from Microchip: ATF1500A-10JU

 

It's still a 5V device, and there is presently a decent stock at Digi-Key going for a very reasonable $3.82 a piece.

 

When I tried to fit my requirements into the former 22V10 device, the two latches would always get in the way, requiring buried nodes to feedback into the equations. If I had a couple more spare I/O bits it would have likely worked. Anyway I believe I've got it solved now with the new CPLD device.

 

So after much searching and head scratching I was able to put together a visual (PLD schematic capture) development platform for free based on the older Protel 99 SE software. It did require also finding a free copy of Windows 2000, and installing everything in VirtualBox, but in the end it really worked out nicely. It also required installing WinCUPL to do the final compilation, since with this new CPLD it would produce an unknown error in Protel when it attempted to do the same. It was 100% successful in WinCUPL. Luckily the error in Protel didn't stop it from producing the PLD source, so it was a simple matter of carrying that over into WinCUPL to let it compile down to the PLA code (kinda different than a JED file). I still need to verify if my TL866II Plus programmer will handle this ok (fingers crossed).

 

Here's a look at what the schematic looks like in Protel using it's specific PLD symbols. This is a very nice way for me to do this kind of development, because quite frankly I just wasn't fully understanding how to use WinCUPL's equation/table approach. I'm so used to working in a schematic format, and this allowed for an easy transition to designing PLDs. However I did need to learn the quirks of Protel's software application, and it sent me down a rabbit hole for a few days until I finally got it. It utilizes a database methodology for storing the design, and also offers a Windows File System approach as well. Unfortunately I found out the hard way to not opt for the later.

Protel CV-NUC+ CPLD schema.pdf

 

I did have to deal with another quirk of using Protel's PLD schematic capture, where the predefined X74_138 decoder symbol didn't like having outputs Y4-Y6 left unattached, so I had to mickey-mouse it into the circuit with some extra unneeded decoding. But hey it seems to work, and kept the compiler happy.

 

Now I have yet to test this in reality, so until I do, there will be no upload of the CPLD code. Unfortunately I will also need to rip-up and do a new PCB layout since this new device is a PLCC-44 instead of the former 24-pin PDIP. However the nice thing about using a CPLD is that it allows for software changes to fine tune the design.

[+Yurkie]

I'd love one of these when available.

[+mytek]

On 10/12/2022 at 9:42 AM, mytek said:

...the nice thing about using a CPLD is that it allows for software changes to fine tune the design.

Quoting myself, I just expanded upon what I was doing with the new CPLD, and have now used up all but 2 of its I/O pins.

So in this iteration, I've moved the 'real' physical 138 address decoder off of my main board and have replaced it with a virtual one (U3) in the CPLD. This buys me back some space on the main board, which may or may not be utilized, but it does reduce the cost none the less. While I was at it, I also decided to route the upper address bits of the 32K SRAM through the CPLD, thus giving it control of the mirroring aspect to better mimic a stock CV's 1K of SRAM. Now it's all just soft bits instead of hardware that can be changed to tweak things into working.

 

16 hours ago, Yurkie said:

I'd love one of these when available.

Well unless someone opts to build and sell these, you'd probably be looking at a DIY build.

 

As I mentioned early in this thread, this project really is about my entertainment and not something I'm trying to make any money from. Although it has already cost me quite a bit. I just picked up a universal burner off of eBay for around $200 the other day, and then there are all those other parts purchases, as well as buying a real CV for reference. By the time I'm done, I'll probably have spent $700-800 for development, after factoring in sample PCBs.

 

For the DIY people, I'm investigating a statement by Digi-Key on their order page for the specific CPLD I'll be using, that says they can program it.

 

 

BTW this chip is not soldered in place, and will be socketed for easy replacement.

 

There are versions that have JTAG capability (e.g., ATF1502), but they are 3.3V and would require down regulation and level shifters. The ATF1500 natively works at 5V, but doesn't have any kind of in circuit programmability, thus the reason why a programmer that will flash it goes for $200 used (although much nicer than it's original $1,000 list price). However despite the lack of JTAG, it's still a great deal, and very much available even in the midst of the chip shortages. But obviously not everyone is going to buy an expensive programmer, so Digi-Key's programming service might be the answer.

[+Yurkie]

On 10/14/2022 at 1:53 PM, mytek said:

 

 

Well unless someone opts to build and sell these, you'd probably be looking at a DIY build.

 

 

If you would be so kind as to PM me, I would love to talk about buying one of your completed test units.