New 576NUC+ 4-in-1 Expansion Module - The NUCplus4

[+mytek]

@electrotrains inspired me with his recent A8Pico Cart project, and I decided to create a new daughter board for the 576NUC+ that would incorporate that as one of four aspects.

 

I introduce to you the NUCplus4...

 

 

Here are the four main aspects...

1. Additional memory expansion, nearly doubling the original 576K to a grand total of 1088K (512K doubled to 1024K+64K base RAM). Thanks @HiassofT for the original 512K extended RAM circuit, and @tf_hh for further expanding this concept up to 1024K.
2. It adds an ICD compatible R-Time 8 battery backed RTC (based upon the original M3002-16P1 chip which has gotten far easier to reliably source).
3. It adds an SDrive-ng which includes the swap disk feature (rotates-right disk slot 1-4), and allows for hot swapping SD Cards on the fly (thanks @kbr).
4. It adds a 16MB USB flashable high speed Multi-Cart similar to a UNO Cart without the need for an SD Card (thanks @electrotrains).

 

I've tried to acknowledge the main creators of the original hardware designs being integrated into this project, but for the full listing of all those involved in the evolution of these individual projects check out the board's silk screen. And as part of the integration into the NUCplus4, there were several circuit modifications that got made above and beyond the original designs to insure that these four aspects work well in concert, are trouble-free, and hopefully idiot-proof 🤪

 

And for a future FujiNet add-on, there is an additional pass-thru header provided that breaks out all the SIO related signals, as well as the PS/2 keyboard signals. The idea will be to have this future FN add-on board only do the Wi-Fi related FN activities, sans SD Card socket, and be fully controlled via PS/2 keyboard shortcuts instead of by physical switches.

 

Disclaimer: This board will not support a 'real' cart, nor be usable with the cart version of the 576NUC+, or any other previously offered daughter boards.

 

Today I will be placing a sample PCB order to be sure all is good with the design, and then it will be released on my website thereafter for anyone to use or produce. I will not be making these for sale myself, so please don't think this is a pre-order thread or PM me about getting one

[Noelio]

Noice!!!!!!!!!!!!!
Things are getting very good in the A8 world!

[+Stephen]

4 hours ago, Noelio said:

Noice!!!!!!!!!!!!!
Things are getting very good in the A8 world!

Sadly I think there's so much new stuff coming out continually, I can no longer have it all

[+mytek]

Here's an initial look at the board in action...

 

 

This demo is with the first prototype which did require some fixes after the fact, a few tacked on components and wires, and was missing the SIO pass-thru header. The next version (V1.4) should be production ready and have all this extra stuff integrated.

 

EDIT: I didn't show much of the SDrive in action since that's already been covered fairly well in other videos. The SDrive-ng which this is based upon pretty much does most of what the SDrive-Max is capable of, minus a few aspects such as working with CAS and ATX files. Swapping up to 4 disks is possible via a momentary push button, and the SDrive can be assigned as D1-D4 via a dip switch. Currently all HSIO divisors from 2 on up are possible, and I am hopeful that 0 and 1 will be possible in a future update.

[+mytek]

4 hours ago, Stephen said:

Sadly I think there's so much new stuff coming out continually, I can no longer have it all

Stephen for consideration, this board will cost around $75 if you DIY.

[+Stephen]

14 hours ago, mytek said:

Stephen for consideration, this board will cost around $75 if you DIY.

Yeah but I'd need a second 576NUC because my current one has cart port + FujiNet and I can't give that up.  But, they are small, and I do have a pair of matching Sony PVM 8040s and I did want a black case.  Hmm - I see where this is going.

[+mytek]

Here's a better look at the prototype board used in the demo video (Code Name: RPSX). This version got it 85% right, and served as a good platform to run experiments on, that would eventually get it another 15% towards the finish line.

 

 

The Antic and the GTIA have been heat-sinked on the 576NUC+ board underneath, which is good practice for this system because of the chip density compromising the two graphics related chips ability to shed heat. The black jumper wire is providing the /CCTL signal from a header on the 576NUC+, and is the only 'extra' connection required, with everything else coming thru the piggybacked chips and the internal SIO header making this a solder-less installation.

 

In the new upcoming version (V1.4) The battery size has been reduced, the SD Card socket has been changed to one that is in current production, and the bodge wires and extra components have been incorporated. Also the SIO signals will be passed thru to a top mounted female header for future expansion. The R-Time 8 only uses the battery when the unit is unplugged from the wall, otherwise it gets a constant power connection even when the 576NUC+ is in an OFF state, thus it doesn't need to be very big battery for back-up.

 

SW1 is the reset for the A8Pico Cart, SW2 is for the SDrive swap disk feature, and SW3 allows selecting the SDrive's physical drive assignment.

 

 

EDIT: BETA, V1.0, V1.1, V1.2 never materialized as real hardware, but were instead conceptional revisions that occurred only on paper instead. Some got close to reality, and even had PCB layouts, but simply didn't make the cut.

 

EDIT2: Also wanted to mention that there are some blog posts that I made earlier about this project on my website.

[Mr Robot]

While you are playing with the A8PicoCart design, why not make a couple of low profile ones that would fit into the XEL or XLD?

[+orpheuswaking]

Stop already 😝 I'm with Stephen, I want this but can't justify the cost of another board. 

[kenames99]

On 9/23/2023 at 7:23 PM, mytek said:

Here's an initial look at the board in action...

I watched you video when it was out for 33 minutes. thanks for all your work mytek!

 

Ken

 

[+mytek]

20 hours ago, Mr Robot said:

While you are playing with the A8PicoCart design, why not make a couple of low profile ones that would fit into the XEL or XLD?

Not a bad idea, and maybe I'll look into doing that when I get a chance

[+mytek]

Project Update

 

Well allowing for the addition of FujiNet became a top priority last month. So in order to make that happen a plug-in module was devised, which also required two more small headers on the main NUCplus4 daughter board. So besides the new FujiNet board, there will also be a new main board, bringing it up to revision level V1.5.

 

Here's a mock-up of the new Fujinet Plug-In Module.

 

 

The actual board footprint is just under 2" square, with the FTDI board's USB-C jack overhanging to match up with the depth of the USB-C jack on the A8Pico Cart Module (they'll be side-by-side with access from the rear panel). And of course the ESP32 WROVER WiFi Module overhangs the front side where its antenna is located.

 

This a WiFi only version, with the SD Drive being handled by the NUCplus4 daughter board.

 

The PIC chip interfaces with the 576NUC+ PS/2 keyboard (luckily I added those signals to the 576NUC+ internal SIO header as part of the original design), and via the function keys on the right side of the keyboard, substitute for the original FujiNet Button A, Button B, and Soft Reset switches. There will also be a keyboard equivalent to controlling power to the board in order to provide enable and disable of the FujiNet aspect (with the last state stored in non-volatile EEPROM within the PIC chip).

 

PCB layout is complete, and there will be a sample board run pretty soon. Stay tuned for the results of the tests on an assembled version.

[+mytek]

Well scratch that last board proposal.

 

After running extensive tests, I wasn't seeing good results when trying to share the SDrive with FujiNet, so I've changed direction on this. I've decided to let FujiNet do its own thing when it comes to the SD Card aspect.

 

So how this will work is: when there is no FujiNet Plug-In Module in place which is after all optional, then the NUCplus4 Main Board will revert to a standard SDrive-ng peripheral. However when the FujiNet Module is present, then FN can take over the SD Card, and kick the SDrive-ng to the side. This will be under PS/2 keyboard control, but essentially only one or the other has full use of the SD Card - not both at the same time.

 

Here's the NUCplus4 FujiNet Plug-In Module schematic.

And here is the SDrive section of the NUCplus4 Main Board schematic

The 7-pin header (P2/J2) is the SD Card bridge connection between the two boards. All SIO related signals still come from the 576NUC+.

 

Both boards also have PS/2-AUX PIC microcontrollers for communication and decoding of control signals related to their operation emanating from the 576NUC+ keyboard connection.

 

So instead of shutting down power on either the ATMEGA or the ESP32, each processor can be taken out of action by holding its respective RESET line in a low state, resulting in all I/O on that processor reverting to inputs. Taking this approach stripped out a lot of power control components and circuitry. Also by piggybacking the 3.3V power output of the Pi Pico Module that provides a A8Pico Cart functionality (not shown), no independent 5V to 3.3V regulator is required either. And with some 2N7000 MosFet logic, the need for power steering diodes was also eliminated. The diodes wouldn't have worked anyway for sharing power from the USB +3.3V or the Pi Pico +3.3V due to too much voltage drop incurred thru the diodes. So at least for flashing of the ESP32, the 576NUC+ will need to be powered ON since it will provide the sole +3.3V power source via the NUCplus4's Pi Pico module.

 

Note: The +3.3V coming from the FT232RL module is strictly used as a logic output to insure that the ESP32 is placed in a state for USB programming.

 

In place of dual 74LS07 buffer chips normally used for level shifting and isolation, a single SN74LS641 Octal Transceiver Chip is employed instead. Just like the 74LS07 its outputs are Open Drain, and because it has an enable pin, all connections can be tri-stated, effectively removing all ESP32 connections to the SIO bus when desired (all except audio which is tied in via a resistor for SAM). And unlike the 74LS07 buffers, the SN74LS641 also provides schmitt trigger inputs which will help eliminate any noise on the SIO bus from getting to the ESP32 (couldn't hurt).

 

So a single SD Card serves both the SDrive-ng and the FujiNet when present, without any compromise to either.

 

Thus far I've tested most of the new logic and things seem to be working properly, but for a complete test I will need new PCBs. So that will be my next job to do.

[AtariGeezer]

7 hours ago, mytek said:

Well scratch that last board proposal.

 

After running extensive tests, I wasn't seeing good results when trying to share the SDrive with FujiNet, so I've changed direction on this. I've decided to let FujiNet do its own thing when it comes to the SD Card aspect.

 

So how this will work is: when there is no FujiNet Plug-In Module in place which is after all optional, then the NUCplus4 Main Board will revert to a standard SDrive-ng peripheral. However when the FujiNet Module is present, then FN can take over the SD Card, and kick the SDrive-ng to the side. This will be under PS/2 keyboard control, but essentially only one or the other has full use of the SD Card - not both at the same time.

 

 

Yep, Fujinet uses SIO devices from 0x70 to 0x7F. which clobber SDrive and SIO2SD device types. My Atari WiFi Modem code uses SIO2SD routines as an alternate to using the Screen Configuration for SD Cards...

[AtariGeezer]

10 hours ago, AtariGeezer said:

Yep, Fujinet uses SIO devices from 0x70 to 0x7F. which clobber SDrive and SIO2SD device types. My Atari WiFi Modem code uses SIO2SD routines as an alternate to using the Screen Configuration for SD Cards...

Make that 0x70 to 0x78...

[+mytek]

Still making progress...

 

Look Ma, No Switches!!! Everything is now controlled by the same PS/2 keyboard that also controls the 576NUC+ down below. And we are still looking at a completely solder-less upgrade.

This is the 3rd physical prototype of the NUCplus4 daughter board including the first plug-in FujiNet module to make it from the design table to an actual PCB. I decided to go with an RGB LED for a status indicator on the FujiNet module so that only a single hole would be needed in the lid. A single 74LS641 bus transceiver chip provides the buffering and voltage level translation between the ESP32 and the SIO bus, and a small 8-pin IC (PIC12F629) is the PS/2 sub-processor. There is also a similar one on the main board as well. These translate the PS/2 keyboard signals into switch closures that operate everything.

 

This diagram gives a fairly good overview of the complete system as it stands now, and how it gets controlled via the common PS/2 keyboard.

 

 

 

 

When in SDrive mode, the SD card is assigned to that usage, and when FujiNet is selected it takes over that SD Card for itself. Bottom line neither side is aware of the other, and they are only meant to be used in a singular fashion. This avoids any possible conflict, while also giving FujiNet full control when it's active. To make this possible, each side is held in constant reset when not desired thus putting all of the I/O for either the ATMEGA328 (SDrive) or the ESP32 (FujiNet) in an input state. This also applies to the SPI bus that communicates with the SD Card. So whichever side is not being held in reset, can interact with the SD Card as well as the SIO bus.   EDIT: I go into a bit more detail about this HERE.

 

The '4' in NUCplus4 represents 4 active devices that are accessible by the 576NUC+ when this daughter board is plugged in.

1. A8Pico Cart
2. SDrive or optional FujiNet
3. R-Time 8 (8-bit RTC)
4. 1088K Total System RAM

At the moment I'm working on the firmware for the PS/2 aspect via the 3 microcontrollers (Main TK-II and two PS/2 sub-processors). Getting pretty close to done

 

 

EDIT: All other PS/2 control aspects are handled in the same fashion as the original 576NUC+ system.

 

[+Stephen]

I was pretty sure I was done building new Atari hardware, but this little guy in a half height case (compared to my 576 NUC+Cart) would be the absolute ultimate in portable Atari goodness.  I'm already starting to count my spare LSI chips.  I'd love to print my own case using my resin printer, but I think it's just barely too small to manage.

[+mytek]

6 hours ago, Stephen said:

I was pretty sure I was done building new Atari hardware, but this little guy in a half height case (compared to my 576 NUC+Cart) would be the absolute ultimate in portable Atari goodness.  I'm already starting to count my spare LSI chips.  I'd love to print my own case using my resin printer, but I think it's just barely too small to manage.

Speaking of which, I will be working on case modifications next to accommodate the changes. It looks like it'll be 12-13 mm taller than the non-cart case, and be missing a few holes on the front panel, and have two USB-C sized holes in the rear.

 

I was trying to print some test stuff the other day, but ran into troubles which required a few new parts - those arrived today from Amazon. Resin printing would be sweet, but I still don't have one of those.

[Panther]

I thought you were going to really impress us all and fit this into the same space as the basic unit.  😁

 

Looking forward to the final version!

Guess that means I'd better dig my workbench out from under this pile of stuff...

 

[+mytek]

7 minutes ago, Panther said:

I thought you were going to really impress us all and fit this into the same space as the basic unit.  😁

Well I almost could have if I had not provided headers and such for things like the A8Pico module and just soldered everything in permanently. But I didn't want to do that and cut-off easy troubleshooting and possible upgrades.

 

However I did make a mistake in my calculations and it's looking like the overall height might only need an increase of around 7-8 mm over the basic unit. Still pretty tiny

 

EDIT: I'm starting a 3D print of the 1st prototype NUCplus4 base right now - fingers crossed.

[Panther]

I've most of the main components on-hand, so assuming I locate my workbench under this mess I'll be able to build one fairly quickly once you've finalized it.

 

[+mytek]

BTW, I wasn't the first to come up with the idea of having both the SDrive and FujiNet coexisting on the same upgrade for the 576NUC+. @Dropcheck has the claim on that, as evidenced from this PM post she made in the BETA Tester group over 3 years ago.

 

 

Me being a narrow minded idiot and just trying to get the 576NUC+ released with a simple FujiNet only upgrade, expressed my disinterest in going down Lenore's alternate path for the initial release. Of course I didn't use all my words to properly say what I needed to say, and I also didn't put the brakes on this far earlier, so we lost a BETA Tester that day, which really wasn't what I wanted to see happen and that's entirely on me. Personally I feel she had some great ideas in her proposed implementation.

 

Anyway more technology came along, such as the A8Pico Cart and a method to boost the memory from 576K up to 1088K, so these things got incorporated as well. Keep in mind that at the beginning of the 576NUC+ project all I really wanted to do was keep it simple but have it be a strong well thought out base configuration -  and most of all get it released early. Then the idea was to observe what people did with it over the next couple of years, and eventually do an All-in-One upgrade such as the NUCplus4. Also by not being rushed and under a deadline, better ideas came along, thus allowing for full PS/2 keyboard control of everything.

 

EDIT: I also gear my projects as much as possible towards the average DIY solder jockey and avoid SMD in my designs if at all possible. However, sometimes there just isn't any THT alternative for what I need, so in that case I go for the biggest SMD I can find that will still do the job. If you take a close look at Lenore's boards you will see extensive use of SMD, which suggests these weren't intended for DIY and would have likely only been for sale as fully assembled devices. I would have not been able to release the gerbers for her boards as I normally do with mine. Albeit in the time period we're looking at, and due to the SMD on the back side of the 576NUC+ board, the original thought was that this wasn't going to be a DIY project. That changed.

[Panther]

Just thinking out loud how since I'm about to place an order for some PCBs, it sure would be nice to be able to add some NUCplus4 boards to my order...

[+mytek]

37 minutes ago, Panther said:

Just thinking out loud how since I'm about to place an order for some PCBs, it sure would be nice to be able to add some NUCplus4 boards to my order...

Sorry not ready to do that yet. Still chasing the tail on one last bug that I need to iron out and flatten. Until that happens, which also requires time I don't presently have, it stays in limbo.

[Panther]

Time...just start moving very fast.