Quote  

Not giving up yet, I need to know....

I know that feeling ;-)
Have you tried the firmware with no options set apart from the I2C, see if that will run, if not I think it is hardware, somewhere, :-( if it does, then add in the options one at a time, until hopefully it fails.
Regards Kevin

sdcard emulator for 90€ maybe of interest as I never read about it @backshed

is it really  unthinkable?
Edited 2025-05-07 17:04 by dddns

  Quote  

Just checked, enabling PSRAM is the problem. If disabled it works, enabled as discribed

Sounds like you are dropping characters. You need to set a line delay in whatever is sending. In teraterm set the serial port delay per line and/or the copy/paste delay per line to 10mSec

Try to only execute a autosave and press F1 with PS enabled and it happens without even attempting to transfer data


if disabled it says "saved 5 bytes"
Edited 2025-05-07 17:31 by dddns

@Kevin, Peter,

No Aardvark, but I found this... and it actually works.

I2C sniffer on pico

Only disadvantage is that this is volatile. You need to load the uf2, and then it works. At a re-power you have to re-load the uf2. SDA=GP0, SCL=GP3. Maybe they load the uf2 directly into RAM, and run from RAM.

Will diagnose further when at home....

Volhout
Edited 2025-05-07 19:14 by Volhout

  Quote  

Just checked, enabling PSRAM is the problem. If disabled it works, enabled as discribed

If reporting an issue, please indicate which board you are using, the options you have set and exactly which firmware is running otherwise I have no chance to replicate.

I don't personally care too much for bucks, just wanted to be helpful. I'm posting this now I think for the third time

  dddns said  

> option list
PicoMiteVGA MMBasic RP2350B Edition V6.00.02RC22
OPTION FLASH SIZE 16777216
OPTION COLOURCODE ON
OPTION KEYBOARD US
OPTION PICO OFF
OPTION RESOLUTION 800x600 @ 360000KHz
OPTION DISPLAY 50, 100
OPTION SDCARD GP41, GP40, GP39, GP42
OPTION PSRAM PIN GP47

Autosave problem I just double checked,

The code flows by and is transferred to the last byte. When I reconnect pico is alive and autosave is still listening. When hitting F1 or F2 pico crashes with "invalid adress - resetting"

The board type might be important. We have seen operational differences because of that before.

simply look away and type shorter lines :) But its a PGA2350

and runs since 500 hours without problems..

I think you got one of those..can you confirm the bug?
Edited 2025-05-07 21:24 by dddns

EDIT: delete post
Edited 2025-05-07 21:44 by Volhout

Page 1 of this thread?

sorry for my niggling @Mixtel
Edited 2025-05-07 21:45 by dddns

800x600 works fine on the RP2350 VGA as does 848x480
The PGA2350 is extremely marginal running overclocked particularly when PSRAM is enabled. I spent many many hours playing with the various QMI timings to get it to work overclocked particularly related to PSRAM timings. My all-in-one designs are much more robust for overclocking as I use a linear regulator for DVDD, have better routing and use a oscillator rather than a crystal.
When you press F1 or F2 you are writing to flash on a board where the timings are already marginal so I suspect you are seeing bus issues that I can't do anything more about.
Please note, it is important for everyone to list things like options in each post as I can't be expected to look back through a thread to find a historical post from the same person to investigate each issue raised.

  matherp said  

800x600 works fine on the RP2350 VGA as does 848x480
The PGA2350 is extremely marginal running overclocked particularly when PSRAM is enabled. I spent many many hours playing with the various QMI timings to get it to work overclocked particularly related to PSRAM timings. My all-in-one designs are much more robust for overclocking as I use a linear regulator for DVDD, have better routing and use a oscillator rather than a crystal.
When you press F1 or F2 you are writing to flash on a board where the timings are already marginal so I suspect you are seeing bus issues that I can't do anything more about.
Please note, it is important for everyone to list things like options in each post as I can't be expected to look back through a thread to find a historical post from the same person to investigate each issue raised.

Excuse me please, didn't want to offend. I follow your development now since 5.08 doing almost every update to watch the development. Three weeks ago we had that issue with the editor which was as well related to PSRAM. Now you really worked out all issues there were and I definitely thought that this is one you easily figure out because its seems to be so obvious a bug.

But I can't judge as I have no insight (and being not able to) and I'm accepting this answer completely!


Would you please do me and favor a try this on one of your reference boards

My PGA runs stable at 378Mhz and LCD
Edited 2025-05-07 22:11 by dddns

  Quote  

Would you please do me and favor a try this on one of your reference boards

I have - it works. LCD is much simpler, only one core running and no continuous DMA bus transfers happening
Edited 2025-05-07 22:33 by matherp

Thank you!

If there is somebody in EU who is willing to sell one of the 2350B reference designs, please drop me a PM. I will paypal first, then you can send..
Edited 2025-05-07 22:54 by dddns

dddns

Try this - I've just done some fairly arbitrary recoding - it may help?????

PicoMiteRP2350VGAV6.00.02RC23.zip
Edited 2025-05-08 01:38 by matherp

Yes it does! Just executing autosave and terminating it works. Receiving data and then F1 or F2 works as it gets saved and executed. Many thanks for investigating!

The problem that I now have is the 10ms delay and how to do that with cat or something. If I find a solution then it will be possible to integrate in geany.
Perhaps you have a very brief hint? :)

  Quote  

The problem that I now have is the 10ms delay and how to do that with cat or something.

Don't understand

only 4.3 of 50k are received. I thoght it would need a delay between each line sending

The code looks completely scrambled

But wait, I think it was already ok with disabled PSRAM. I will check in a second

no, its scrambled too.
Edited 2025-05-08 02:08 by dddns

  Quote  

only 4.3 of 50k are received. I thoght it would need a delay between each line sending

The code looks completely scrambled

You need a delay between lines - I use 25mSec in Teraterm but this is probably overly conservative.
You also need a delay before you start sending - the change I made was to clear the flash first before reading the stream into the buffer and this takes time.
The best way to do this is to send a line and wait for the echo. When you have received the echo it is safe to send the next line.
Edited 2025-05-08 02:12 by matherp