Startup and tuning bug in 0.9 firmware

[IgorGIV13]

Hello!
There are two bugs in the latest (0.9) firmware.

1. After the real processor frequency has been set and saved, if you turn off the computer and turn it on again, it starts at a slower frequency, despite the fact that the real frequency is displayed.
2. Frequency fine-tuning of the preset does not change anything except the numbers on the display (PWM period tuning works). But it saves the value. So it is not possible to adjust the frequency more accurately. If i try to change the second value in Preset.h before building firmware (see eaxmple below - 15000) it ignores it but shows in fine-tuning mode.

"const std::vector ImmutableTable = {
Entry(10000, 15000, DefaultPeriod)};

P.S. I'm using two button mode

[scrapcomputing]

Thank you for the bug report!

I think I fixed both issues. Could you please try the latest firmware rev.0.9b ? https://github.com/scrapcomputing/ThrottleBlaster/releases/tag/rev.0.9b

Thanks!

[IgorGIV13]

Thank you very much for the fix! Now I can fine-tune any preset and it works great, but unfortunately all settings resets after PC was powered off. So I have to fine-tune it again.

UPD. I found a small inaccuracy in source code in the "main.cpp" file in line 234 - you need to replace "DC.setKHz(Presets.getKHz());" to "DC.setKHz(Presets.getActualKHz());"
At least without this correction the ActualKHz parameter in Presets.h is ignored.

But I still can't save fine-tune settings of presets after PC was powered off when I do it via ThrottleBlaster.

P.S. I try to build firmware and can't use rev.0.9b because it has the same bug and another one related to startup - display shows 0 and system doesn't start until any button is pressed.

[scrapcomputing]

Thank you for persisting on this :)

There was some code that would check if values did not change to avoid unnecessarily writing to flash, but this was hard to maintain and buggy so I dropped it. This should fix the issue related to keeping the settings after a power cycle. Please try the latest firmware rev 0.9c, it includes this fix.

Regarding the issue with the display showing 0 (and not letting your PC to start), I can't reproduce this. Could you walk me through how got it to do this? Is this still in two-button mode?

EDIT1: This is probably related to DC.setKHz(Presets.getKHz());. Presets.Idx might be uninitialized, let me look into this.
EDIT2: Preset.Idx gets initialized by both Presets.resetToDefaults() and Presets.readFromFlash(), so I don't think this is the issue.

[IgorGIV13]

1. Thank you very much for your work! Now I can build fully working firmware from the source code with my presets. Everything works great! I can fine-tune presets, save them, then turn off the PC and turn it on again and now all the settings are saved. You are my hero! ;)
2. However I can't use your 0.9c firmware. When I write the firmware to pico, the display immediately shows 0 (although it should usually show 200). When I turn on the PC the display also shows 0 and the computer does not boot until I press any button on the ThrottleBlaster. After pressing the button, the PC start booting and I can configure the ThrottleBlaster as usual. But as soon as I turn the PC off and then turn it on again, it shows 0 again and does not boot until I press any button.

Anyway i'm able to build my own fully working firmware from the source code so I can't even imagine what is the difference between your build and mine.

[scrapcomputing]

But if you are having issues with 0.9c, others will most probably have issues as well, so I it's worth investigating.

1. Are you using a pico 1 or a pico 2 ? I only checked with a pico 1.

2. Did this issue first show up in 0.9b ?

[IgorGIV13]

1. Yes, I use pico1 two-button mode.

2. While I was trying to figure out why the firmware wasn't working, I tried a huge number of combinations of both the existing firmware versions and various changes to the source code to get it working. Therefore, unfortunately, I can't say for sure whether I've encountered this issue with other firmware versions, and if so, which ones was it. And one important thing - before flashing the pico, I always reset the ThrottleBlaster configuration to default settings.

Here I should share a few details that I remember very well.

My first experience was with the firmware 0.8, and everything worked fine with it except for the ability to fine-tune presets. Then I tried 0.8c and 0.8d and with them I had a problem with the initial setup of processor frequency — it didn't save settings and the display did stop lighting.
0.8e solved the initial setup problem, but fine-tuning presets still didn't work.
Starting with 0.9, I first encountered the need to press any button on the ThrottleBlaster, otherwise the computer worked as if at a frequency of 200 MHz - the computer beeped unusually and it took quite a long time to load. And when I pressed any button, the computer started working at its nominal frequency (I have a Pentium 3 750 MHz).
A friend of mine (he is also a retro fan and also has a ThrottleBlaster) connected an oscilloscope to his ThrottleBlaster (with 0.9) and saw a PWM signal after the PC was turned on, which is probably what slows down the computer to 200 MHz.
So it looks like the problem started with 0.9. But in any case, when flashing the pico, the display always showed 200 and only starting with the latest one it began to show 0.

Here are two more important details, in my opinion. It seems there are different versions of pico1. Some had problems with the initial setup at 0.8c and 0.8d, while others did not. Maybe this is the problem? And that's why I have 0 after flash and your pico1 doesn't affected this bug.
But I really don't understand why with the same source code your firmware causes such a problem on my pico, but when I compile the firmware myself everything works fine.
If this helps, I can send you the firmware I compiled from the source code that works for me.

[scrapcomputing]

Thanks for all these details, I will try to investigate further.

Btw, are you using a pico clone or the original ? Not that it should matter for this project, but just to make sure we are using the same platform.

[IgorGIV13]

I don't know if my pico original or a clone (I bought it on AliExpress), so i can only attach the photo of it.

And here is my firmware with presets: 10, 33, 66, 200 and 400MHz. It would be interesting to know if it will work on your pico .

TB_10_33_66_200_400.zip

[scrapcomputing]

This looks like a clone. The original one has the raspberry logo at the bottom, right above the debug pins.

I don't usually test with clones because they are not as well tested as the original ones and you may end up with one that is half working, also they tend to have slower flash chips which may not be compatible with projects that overclock the pico (this shouldn't be case here though).

[IgorGIV13]

I think I should write a complaint to the seller for deceiving customers.
However, this does not explain why the firmware works when I compile it from the same source code.