Pi Pico[W] Peripheral Expansion Box Side Port Device

[visrealm]

Thanks for the TI-99 mod trade, Jason. Initial signs look positive

 

 

[JasonACT]

On 8/15/2024 at 7:22 PM, JasonACT said:

It's now running at 210ns (worst case) before A15 changes from high to low (which is when the data is latched in the console).

...

The actual average timing is 246ns at 250MHz...

Worst case is now 230ns, average 266ns - at 250MHz.  I tried a fair few things to make it better than that, but GCC 12.3 wasn't really cooperating and my logic analyser started to show a weird spike in the PSRAM chip select.  I have no idea why.  The Arm GCC 13.3 download didn't work for me either, and from my testing is only a tiny fraction faster than 12.3 anyway.  This might be it, for the bus logic part of my project.

PPEB2.zip

[+chue]

SAMS burn-in ran fine on this latest firmware update.

 

250 mhz, 1 nop

[Artoj]

On 8/16/2024 at 12:41 PM, visrealm said:

This is what an average TI99ers techo desk looks like today, such a powerful arsenal in every instrument and adaptor board, 40 years ago who would have guessed, whereas my desk looked a little like that 35 years ago but with hand made breadboards with socketed 74LS chips and LEDs,  rough soldered wires in a tangled mess, beeping logic probes, multi-meters everywhere, IC books open everywhere and maybe an old chunky transistor/valve Oscilloscope that you shared with a dozen other Electronic enthusiasts. I do miss the fun, but not the seat of the pants hacking, expecting to make magic smoke at any moment, great work Jason, as Aussies you and Troy have revived my hope in the creative aspects of the old time "open-up-and-lets-see-what-we-can-do" approach which served the TIshug members for a decade of building stuff. It is also wonderful to see the cooperation across continents, from the USA, Finland, Canada, UK, France, Italy and Germany, hope is not just a word, regards Arto.       

[visrealm]

Cheap, small batch PCB manufacturing and a wealth of information at our fingertips has certainly made things a lot easier.

[Gary from OPA]

22 minutes ago, visrealm said:

Cheap, small batch PCB manufacturing and a wealth of information at our fingertips has certainly made things a lot easier.

It has.

 

But I miss the original all local touch. I would go to the plotter guy he would take my floppy and print out the layers onto clear film, then I would swing by the little PCB factory, with my roll of films and my drill via tape (yes it was punch tape input) chat with them as they quickly did a sample run look it over give them ok and return in a couple of days to pickup a batch of 50 or 100 boards, head over to assembly shop and watch their row of soldering girls go to work.

 

Just not the same anymore.. and in the end the cost is really still the same as before, cheaper PCB factory, but now we pay for international shipping of boards and parts, so it comes to the same cost as before with using local labour and no shipping as even the parts like the chips, resistors, caps you could all get locally as well.

Edited August 22 by Gary from OPA
grammar

[JasonACT]

8 hours ago, chue said:

SAMS burn-in ran fine on this latest firmware update.

 

250 mhz, 1 nop

That's good news, thanks again for testing  

[JasonACT]

New 2MB SAMS firmware attached: SAMS RAMBO-4000

 

If autoload.cfg has "RAMBO=1" in it, then addresses >4000->5FFF is SAMS RAMBO memory.  With this, you can have 24KB low memory and 24KB high memory (by enabling the MiniMem 8KB RAM 'super cart' feature).  I'm still thinking about real >6000 SAMS RAMBO, but there's ~4MB of space there already which I think you can page by toggling the RAM mode and writing "to switch pages", and SAMS is only 2MB which would limit things more than they are now.  Open to suggestions here.

 

This firmware keeps track of all 16 external peripheral devices, if any DSR ROM is enabled, then SAMS RAMBO-4000 is temporarily disabled (so don't load your DSR access routine in that space!).  If no DSR ROMs are enabled, we're back to SAMS RAM with the usual paging technique (similarly, don't try to page SAMS using code within the >4000->5FFF area).

 

(If anyone ever builds the 8MB SAMS PicoPEB mod, I'll re-start building both firmwares.  Oh, and I've done this because I was thinking about how to get larger CVBasic programs to work on the TI - this adds 16KB more - for 40KB programs with 8KB of variables, with some small mods to the generated asm.)

PPEB2.zip

Edited September 5 by JasonACT

[+chue]

On 9/5/2024 at 3:33 AM, JasonACT said:

New 2MB SAMS firmware attached: SAMS RAMBO-4000

Passes the SAMS burn-in test.

[JasonACT]

Thanks for confirming @chue

[RickyDean]

On 9/5/2024 at 4:33 AM, JasonACT said:

New 2MB SAMS firmware attached: SAMS RAMBO-4000

 

If autoload.cfg has "RAMBO=1" in it, then addresses >4000->5FFF is SAMS RAMBO memory.  With this, you can have 24KB low memory and 24KB high memory (by enabling the MiniMem 8KB RAM 'super cart' feature).  I'm still thinking about real >6000 SAMS RAMBO, but there's ~4MB of space there already which I think you can page by toggling the RAM mode and writing "to switch pages", and SAMS is only 2MB which would limit things more than they are now.  Open to suggestions here.

 

This firmware keeps track of all 16 external peripheral devices, if any DSR ROM is enabled, then SAMS RAMBO-4000 is temporarily disabled (so don't load your DSR access routine in that space!).  If no DSR ROMs are enabled, we're back to SAMS RAM with the usual paging technique (similarly, don't try to page SAMS using code within the >4000->5FFF area).

 

(If anyone ever builds the 8MB SAMS PicoPEB mod, I'll re-start building both firmwares.  Oh, and I've done this because I was thinking about how to get larger CVBasic programs to work on the TI - this adds 16KB more - for 40KB programs with 8KB of variables, with some small mods to the generated asm.)

PPEB2.zip 577.05 kB · 3 downloads

I haven't continued as I've been busy on many other things, I've built one white onean 8mb, but with only 1 psram right now, but it and my last blue 2mb one didn't work. I've messed with them and I believe I got a batch of non working pico's from Amazon, as I can't find any reason the two of them won't work otherwise and the boards are soldered to the pcb, not on risers. 

[JasonACT]

44 minutes ago, RickyDean said:

I haven't continued as I've been busy on many other things, I've built one white onean 8mb, but with only 1 psram right now, but it and my last blue 2mb one didn't work. I've messed with them and I believe I got a batch of non working pico's from Amazon, as I can't find any reason the two of them won't work otherwise and the boards are soldered to the pcb, not on risers. 

It might be worth posting up a picture/photo.

Edited September 9 by JasonACT

[RickyDean]

19 hours ago, JasonACT said:

It might be worth posting up a picture/photo.

Your command is my wish.

[+dhe]

I was trying to go through this thread and make some notes,

   1) Does this project have a github site?

   2) Has anyone tested plugging a pbox into the PicoSidecar?

[Gary from OPA]

3 minutes ago, dhe said:

I was trying to go through this thread and make some notes,

   1) Does this project have a github site?

   2) Has anyone tested plugging a pbox into the PicoSidecar?

I have two of them and PEB works, but sadly since this sidecard basically does everything, there no use of any cards in the PEB except for maybe and IDE or SCSI, the rest is all emulated.

 

I would like to see options to turn certain features on and off but some stuff would require a custom rebuild firmware to disable certain devices.

Edited September 10 by Gary from OPA

[JasonACT]

6 hours ago, RickyDean said:

Your command is my wish.

The blue one looks good, I think I can see the trace cut for the bodge resistor in one of the photos, but it would be worth measuring its resistance now while it's in the circuit..  Just to make sure it isn't shorted via the trace.  The white one is really hard to see, it's all white  and the small u2 and U7 appear soldered, but that's the only thing that looks off.

[JasonACT]

5 hours ago, dhe said:

1) Does this project have a github site?

Nope, but I'll supply you the latest source code for a copy of the ForTI disk software  

[RickyDean]

16 hours ago, JasonACT said:

The blue one looks good, I think I can see the trace cut for the bodge resistor in one of the photos, but it would be worth measuring its resistance now while it's in the circuit..  Just to make sure it isn't shorted via the trace.  The white one is really hard to see, it's all white  and the small u2 and U7 appear soldered, but that's the only thing that looks off.

The blue resistor is reading .470 Kohms, the white one .465 Kohms. U2 and u7 small have been eradicated with a small die grinder, especially u7 so that I can be sure when I put a larger psram there it will not short out. I will clean up the looks later with a spot of white piant or something. Been tearing down a car engine, fixing Ide cards that haven't worked in 5 years, fixed a Horizon 4000B ramdisk and others assorted stuff. But I'll get there, I'm going to buy some pico's from a better source, and I'll have 9 pico 2's by Oct. Have 4 now.

Edited September 10 by RickyDean
spelling

[RickyDean]

The blue one only comes up with a black screen when installed and powered on, on three different TI's. I did make the gap bigger on the trace. and I ensured the resistor was soldered good on the psram and the sdcard holder.

[JasonACT]

5 hours ago, RickyDean said:

The blue one only comes up with a black screen when installed and powered on, on three different TI's. I did make the gap bigger on the trace. and I ensured the resistor was soldered good on the psram and the sdcard holder.

Black screen might mean the TI is being blocked from running any code, U3 is the data bus driver, you could try pulling that chip out (I think that would be safe) and power on the TI to see if one of the 8 data lines is being messed up.  Also check for shorts between socket pins while it's out.  Once that chip is out, it's probably safe to pull U4 and U5 (address bus) chips and see if the TI then boots up.  Same check for shorts on those two sockets.  Then we get to the 3 transistors, they are READY (top one on its own) and LOAD + EXTINT next to each other at the bottom.  Desolder those one by one and try to power up the TI.

 

At that stage, you are pretty disconnected from the TI (except for Pico inputs: CRUCLOCK + WE + MEMEN + A15 + DBIN), so you should know which area the problem is in.

[JasonACT]

While testing the Pico9918 "F18A" enhancements, I had a directory with 22 demo files...  I created another test case for the Plasma demo, which brought the total to 23 (the max number of items on a page) and I got a blank page!  My modulus operator to find the number of items to show reduced 23 to 0  

 

Here's a firmware update to show all 23 items on "the last page" when this happens (this is the only change, there are no timing sensitive changes here for SAMS).

PPEB2.zip

[RickyDean]

@JasonACT my friend, your sharp eye caught that solder off one edge finger touching the other edge finger. I removed that solder and the blue Pico sidecar that has never worked is now operational. Thanks eagle eye Jason.

I had started soldering this board and I usually wet out all the solder points with some flux and used solder wick to make the joining parts take solder better and give a good connection. I started on the wrong set of fingers on this board and went to the other side, forgetting  this one and it bit me.  It is working now, doing memory testing.

[JasonACT]

Another mystery is history!  You're welcome @RickyDean.

[RickyDean]

@chue what seems to be the most useable MHZ and NOPS for you? I am currently running all 4 of my boards at 250MHZ and 1 NOP and they seem to be working okay.

Edited September 14 by RickyDean
spelling

[+chue]

I've been running 1 board at 250 mhz and 1 nop. 

 

My other two boards are in storage, so I have not tested them with recent firmware.  However, they were all behaving identically previously.  That is, if one board worked on a certain setting then the others would as well.