Mike Bell
rebelmike.bsky.social
Mike Bell
@rebelmike.bsky.social
Electronics tinkerer, narrowboater, puzzle hunter, board gamer and occasional geocacher.

Also on fedi @mike@rebel-lion.uk
There’s my DV HSTX library - though I thought Adafruit were already using it on the fruit jam.

github.com/MichaelBell/...
GitHub - MichaelBell/dvhstx: DVI for HSTX
DVI for HSTX. Contribute to MichaelBell/dvhstx development by creating an account on GitHub.
github.com
November 10, 2025 at 12:46 PM
I did ponder using edge connectors

But decided to stick close to RC2014 so went for standard pin headers.
October 28, 2025 at 8:49 AM
Shape wise, I had been thinking more rounded - this is the current state of the CPU module (most things not even placed yet):
October 26, 2025 at 9:49 AM
Then you have the possibility of compatibility with dual voltage RC2014 modules, but at least you shouldn’t have a high current 5v connected if you do put an RV4028 card in an RC2014.

Think the RV4028 modules should also have some warning text on them, and be a different shape.
October 26, 2025 at 9:40 AM
I think the best solution might be: provide 3v3 on the equivalent of the RC2014 5v pins, but don’t use it on RV4028 modules.

Instead use a 3v3 provided on some of the spare pins, which are unused by most RC2014 modules.
October 26, 2025 at 9:32 AM
If this is self assembly it could be made a recommendation but I don’t know how many people would bother.

One possibility would be to offset say the last 2 pins by 50mil, but that would be a little tedious as you then have to break your 40 pin dual row headers to solder them in.
October 26, 2025 at 9:28 AM
The only reason not to do this would be to allow using RC2014 modules in an RV4028 if you built them with 3v3 compatible components. E.g. use a 74hc instead of 74hct and the clock module would be compatible.

But that’s probably possible with sufficiently few boards that losing that is no big deal.
October 26, 2025 at 9:05 AM
Just moving/disabling the supply voltage pin is likely not enough as there will be 5v on the other pins if you were to plug a module into an RC2014.

Although you might get away with that, at least briefly, so this is probably sensible.
October 26, 2025 at 9:02 AM
It's likely that self assembly would be less possible - at least for regular hobbyists - as most of the chips would be surface mount not through hole.

But there would still be plenty of soldering to do for the backplane connections!
October 25, 2025 at 5:06 PM
The basic theme is to double everything on the RC2014 - double the address bits, data bits and clock frequency. Hence the name RV4028.

One thing that isn't doubled is the voltage - at 3.3V it should be easier to interface with more modern hardware.
October 25, 2025 at 5:06 PM
It reminds me of this video of simulations of some #TinyTapeout VGA designs
www.youtube.com/watch?v=J81W...
The Digital Heartbeats (preview)
YouTube video by Programmable Artificial Life
www.youtube.com
October 24, 2025 at 9:38 PM
This idea became lodged in my brain so I started putting some thoughts (and proof of concept gateware) together.

Not sure how much further I might take it - but let me know what you think: github.com/MichaelBell/...
GitHub - MichaelBell/RV4028: RV4028 - A hackable Risc-V computer
RV4028 - A hackable Risc-V computer. Contribute to MichaelBell/RV4028 development by creating an account on GitHub.
github.com
October 24, 2025 at 8:52 PM
You could then have some level of compatibility with RC2014 modules by using ~same pinout on one of the rows - maybe with the top few bits of the address switched to high bits of the 32-bit address so address based module selection still worked.
October 21, 2025 at 12:06 PM
You only need to cut the trace if you want to plug the QSPI Pmod in directly but then use uio7 from somewhere else (e.g. the RP2).
October 21, 2025 at 11:43 AM
No, the audio Pmod itself will pull up uio7 on the pass through socket, so if you have the QSPI Pmod plugged into the audio Pmod you don't need to cut the trace.
October 21, 2025 at 11:43 AM