Pi Pico[W] Peripheral Expansion Box Side Port Device

[JasonACT]

I've done some more testing.  A single file ("device") can have 8 partitions of max 0x7C000 sectors (260MB each).  Add 1 sector for the partition-table, makes a file of ~2GB, being the largest that can be supported by the IDE DSR.  There can be no slave device, as the device IDs (IDE1-8) are all now taken/used.

 

Attached is an archive of this file, it is an empty set of formatted partitions, with zeros in unused sectors - so it compresses nicely down to 2MB.

 

I was able to copy this to a freshly formatted exFAT thumb-drive (micro SD card in a USB converter), formatted with 128KB clusters, and save-to/load-from IDE8 (the end partition) in the TI without any issues.  I think this is as large as it gets.

 

7 hours ago, RickyDean said:

changed to your new firmware, still no joy on the images being seen. Then I tried a usb to sd card adapter, And a sdcard to micro sd and got joy.

This could be 'that' issue with the FAT library I'm using, you may need to run "diskpart" "list disk" "select disk x" "clean" (where x is the thumb-drive number you see listed as plugged in devices).  Then create a partition and format as exFAT (Windows: Computer Management - Storage - Disk Management).

test.zip

[JasonACT]

On 10/30/2024 at 3:28 AM, RickyDean said:

Can the usb drive be used concurrently with a usb keyboard, mouse, and joystick, with a usb hub?

I am seeing issues when multiple devices are plugged into different USB ports on the hub, it looks like a 4 port hub has ports 1, 2, 3 & 4 and the USB stick needs to be in a lower port (1) if another device (gamepad) is plugged in (2 or higher).

 

Also, the SD->USB converter device matters when other devices are plugged in, my cheap one doesn't work (the issue you were having maybe) with my gamepad plugged in, but my more expensive converter does work.

 

EDIT: And, I've just tried another 4 port hub, it has the same priority issue, but it does also allow one of the cheap "all-in-one" converters to work - where that one didn't work in the other hub.  Wow.

Edited November 2 by JasonACT

[JasonACT]

Trying to wrap this testing up...

 

My final all-in-one device needed to be in a higher USB port, the complete reverse to all the other devices, this one has a CF port and that worked too.

I found one more 4 Port USB hub (I know I've got another somewhere, but I can't find it) and it shows the same issue with priority, but only has a problem with my cheapest SD to USB converter (like the other good hub, so only one hub has issues with 2 devices) 3 other converters worked OK with it.

I've got a SanDisk Cruzer Switch 8GB (really old) USB stick which works in both RAW and exFAT.  Had to clean it, only got the FAT32 option when adding a partition, but Windows Explorer allowed it to be re-formatted to exFAT afterwards.

I've got a SanDisk Cruzer Blade 4GB (oh, so old) USB stick which works too in both modes, only got the FAT option after cleaning, but could re-format like before in Windows Explorer.

 

So..  I have 1 micro SD to USB cheap converter that can't be made to work "somehow" - when another USB device is plugged into the hub.

[RickyDean]

Folks, I believe that we now have Ide functionality on the Ppeb. There may need to be a bit of testing done, but with this work, you should be able to pull in a drive image from the SD card using TIIM and fill it with files and use it on the TI99.

 

@JasonACT can we still use 2 drive images if we choose, just partitioning each at 4 partitions apiece? Thanks for your contributions and hard work.

[JasonACT]

5 hours ago, RickyDean said:

Folks, I believe that we now have Ide functionality on the Ppeb. There may need to be a bit of testing done, but with this work, you should be able to pull in a drive image from the SD card using TIIM and fill it with files and use it on the TI99.

 

@JasonACT can we still use 2 drive images if we choose, just partitioning each at 4 partitions apiece? Thanks for your contributions and hard work.

I don't see why not @RickyDean but I haven't tested it yet, I was more concerned with "the largest possible file" passing tests.  I'll test that today.

[JasonACT]

6 hours ago, RickyDean said:

@JasonACT can we still use 2 drive images if we choose, just partitioning each at 4 partitions apiece? Thanks for your contributions and hard work.

I'm not seeing any issues copying from my internal SD card to the end partition of the slave image file, and no issues copying from there to the end partition of the master image file.  So I think it's good.

 

I'll take a look at the IDE DSR source code today and see if I can find the timeout value, then try and work out what sort of timing these files on exFAT are using to seek to sector positions in the file.  If they're not close, then that'll also be good.

[JasonACT]

52 minutes ago, JasonACT said:

I'll take a look at the IDE DSR source code today and see if I can find the timeout value, then try and work out what sort of timing these files on exFAT are using to seek to sector positions in the file.  If they're not close, then that'll also be good.

Looks like the command timeout is about 458ms, and with "exFAT/128KB" all reads are within 12ms.  Plenty of headroom, I'd say.

[JasonACT]

Just to close this out, so to speak, reading about exFAT today at lunchtime...  There's a flag on each file that says if it's fragmented, where if it isn't (and exFAT goes out of its way to not fragment files) then it bypasses the FAT chain processing and uses a fast position calculation based on the in-use-bitmap.  That's why even small cluster sizes worked with exFAT.  Should only become an issue with heavily updated partitions, which we can easily avoid.

[RickyDean]

2 hours ago, JasonACT said:

Just to close this out, so to speak, reading about exFAT today at lunchtime...  There's a flag on each file that says if it's fragmented, where if it isn't (and exFAT goes out of its way to not fragment files) then it bypasses the FAT chain processing and uses a fast position calculation based on the in-use-bitmap.  That's why even small cluster sizes worked with exFAT.  Should only become an issue with heavily updated partitions, which we can easily avoid.

Thanks again. I took your test.cf7 file and opened it in TIIM3, not the latest though, and loaded about 6 of the partitions with TI goodies. I then saved the file back to the SD card and can see the files in all the partitions that I just loaded, outstanding!

I will take and load files on the two smaller images and see if I get the same results. I expect all to work well.

[+dhe]

First PI Pico Expansion Box case spotted in North America.

 

 

[JasonACT]

@dhe has asked me "can you do TI's debugger card?"

 

OK, you can't single step in the last 256 bytes of high-memory, but I don't care.  And yeah, it's in the RS232 Card's CRU space, after all its mapped devices - so only gets enabled with my Digi-Port device implementation, again I don't care.  I kind of do care about the extra run-time being incurred, so I'll think about this until Friday and I may have an updated firmware.

 

 

[JasonACT]

Does anyone know the maximum memory accesses a single instruction can possibly do?

[+mizapf]

Probably a BLWP or XOP instruction with indexed addressing, as a first guess.

[visrealm]

16 hours ago, JasonACT said:

Does anyone know the maximum memory accesses a single instruction can possibly do?

According to this: tms9900.htm

 

XOP: 8

[JasonACT]

7 hours ago, visrealm said:

According to this: tms9900.htm

 

XOP: 8

Yep, thanks, that's it, along with indirect auto increment adding 2.  So 10 max.  I'll keep mine at 16 to be safe.

[JasonACT]

PicoPEB New Features: Console Tester + TI Debugger Card:

 

The config file should have TSTR=1 added, and ROM file "constest.rom" copied to the SD card root folder.  With this set, a LOAD interrupt on initial power-up will fire which runs my version of the "TI-99 Console Tester".  Single LED flashes are it running, when it's done, 2 flashes are a VRAM error, 3 flashes are an SRAM error, 4 flashes are both in error.

 

This version will "talk you through" any errors, literally..  In case the VDP isn't showing anything on the screen (in which case @visrealm's  Pico9918 would probably help, given it may be the VRAM or a bad power choke making that part of the console fail).

 

I've pulled out my GROMs to test this, to make sure you can hear what the issue is from the speech synthesizer, but if it's a READY line being held low (bad GROM or sound chip) I've made the LED on the Pico stay lit constantly when tester mode starts.  Pull all 4 socketed chips, re-test and re-insert them one by one to see which one has failed.

 

For the TI Debugger Card feature, PSNDO=1 should be added to the config file (enable Digi-Port) since its CRU-bit is shard by that device.

 

PPEB2_TSTR.zip

Edited November 15 by JasonACT

[RickyDean]

3 hours ago, JasonACT said:

PicoPEB New Features: Console Tester + TI Debugger Card:

 

The config file should have TSTR=1 added, and ROM file "constest.rom" copied to the SD card root folder.  With this set, a LOAD interrupt on initial power-up will fire which runs my version of the "TI-99 Console Tester".  Single LED flashes are it running, when it's done, 2 flashes are a VRAM error, 3 flashes are an SRAM error, 4 flashes are both in error.

 

This version will "talk you through" any errors, literally..  In case the VDP isn't showing anything on the screen (in which case @visrealm's  Pico9918 would probably help, given it may be the VRAM or a bad power choke making that part of the console fail).

 

I've pulled out my GROMs to test this, to make sure you can hear what the issue is from the speech synthesizer, but if it's a READY line being held low (bad GROM or sound chip) I've made the LED on the Pico stay lit constantly when tester mode starts.  Pull all 4 socketed chips, re-test and re-insert them one by one to see which one has failed.

 

For the TI Debugger Card feature, PSNDO=1 should be added to the config file (enable Digi-Port) since its CRU-bit is shard by that device.

 

PPEB2_TSTR.zip 1.09 MB · 0 downloads

I was testing this some last night. Had 4 old consoles that work, but wanted to try it out. All but 1 showed as good, the one would not boot and the Pico flashed one time when I hit the switch then stayed off. I guess it's something other than sram or vdp. I'll try this new file and see if it does better. I'll try with some dead boards too, as I have about 30- 50 boards in storage, to include a 99/4 board, and a couple OI ones.

[JasonACT]

3 hours ago, RickyDean said:

to include a 99/4 board

I didn't think to include the good-checksum values for a 99/4, so it will say all the ROMs have a checksum error on that machine, if it gets that far.

[JasonACT]

Now includes all 3 console ROM good-checksum values.

constest.zip

[JasonACT]

19 hours ago, RickyDean said:

All but 1 showed as good, the one would not boot and the Pico flashed one time when I hit the switch then stayed off. I guess it's something other than sram or vdp. I'll try this new file and see if it does better.

Maybe one of the line drivers (2 x 74LS367 and a 74LS244 - U503, U509 & U510) aren't carrying out an address line to the side port reliably.  I would use my digital multimeter in logic mode to see if one of the address lines is stuck in a high or low state on the external port.

[+chue]

Finally got around to trying the console tester functionality on one of my non-booting TIs.

 

Here's a video of it in action... When the console is turned on

- the screen turns green, and it counts down from 21 to 1

- then it does some testing on the SRAM and VRAM.

- then it does some checksumming on the ROMs and GROMs

 

 

 

 

According the tester, my non-booting TI has nothing wrong.  Wait, what? Well it is actually correct!

 

So why does my non-booting TI boot?  Well when I first got it, it did not boot.  So I pulled out the power supply to use in another TI.  During that process I realized the -5V regulator failed.  I fixed the power supply but never put it back into this TI until now.  Guess it works!

[JasonACT]

7 hours ago, chue said:

Finally got around to trying the console tester functionality on one of my non-booting TIs.

 

Here's a video of it in action... When the console is turned on

- the screen turns green, and it counts down from 21 to 1

- then it does some testing on the SRAM and VRAM.

- then it does some checksumming on the ROMs and GROMs

 

...

 

According the tester, my non-booting TI has nothing wrong.  Wait, what? Well it is actually correct!

 

So why does my non-booting TI boot?  Well when I first got it, it did not boot.  So I pulled out the power supply to use in another TI.  During that process I realized the -5V regulator failed.  I fixed the power supply but never put it back into this TI until now.  Guess it works!

Another happy customer!  

[JasonACT]

So, TI's Debugger Card...  The only software that seems to use it is the TIBUG Debugger (which comes with the Editor/Assembler package - and has a bug in the provided source code [a double period, in the command table]).  Little does anyone know, but I spent a lot of time using this program in my mid teens, and made many enhancements to it which are now lost to time.  But I've now brought one enhancement back, and turned the program into a cartridge/Supercart - so on my device you need to configure MiniMem mode as 8KB RAM in the config and "CALL CARTMMO" to turn cart RAM mode on.

 

Here's a screen shot of the upgrade running:

Debugger.zip

[visrealm]

On 3/23/2024 at 9:26 AM, JasonACT said:

They call this beige?

 

OpenSCAD 3D model attached, bottom part "no support needed", top part needs support.

 

Zfinal.zip 1.32 kB · 32 downloads

 

How about a "clay" one? 😁

 

[JasonACT]

58 minutes ago, visrealm said:

How about a "clay" one? 😁

That's turned out very nicely, I'm really not a fan of the beige filament I have.

 

I'm still holding out for a 2 tone case, I live in hope

 

To be honest, my first TI was beige, which the power supply went pop within 12 months..  And TI replaced it with a black/silver model, which I used for years and years.  But all my units are now beige, and I'm not unhappy with that.  I don't feel I ever got to really know the beige model.