apache-labs.com

ct1iqi wrote:@laurence:
tnx for background info!
Using the Armbian environment made an image on basis of arm64 architecture and kernel 6.5, but with just enough in the OS to let pihpsdr show its GUI. So basically just cli plus X11 and Openbox Have patched 'your' xdma driver into it and for now called it xdma2 as the xdma name is already used in that kernel. Had to add some small patches to make it compile as some functions, deprecated in kernel 5, were removed or changed in 6. I add the patches, just remove the .txt as .patch does not upload here.
The good news: the audio chopping is gone, using embedded pihpsdr
The bad news: 384 k rate with two receivers is the max. That brings CPU load close to 100%. The massive calculus that wdsp has to do is just to much at higher rates.
p2app works fine and even copes with 1.5 M rate. At 2 x 768k rate it gets the CPUs in the 30-40 % load range. So xdma as such works just fine.
When using embedded pihpsdr there is some effect noticeable in the audio from events in the GUI. When by mouse movement in a menu over a clickable box that box changes color, some effect in the audio is audible. So there are some issues with interrupt priorities it seems.
For some reason with embedded pihpsdr the microphone input is not working while using p2app and an external pihpsdr it works. wdsp, pihpsdr, and p2app are all 100% the current github ones, no changes made.

ct1iqi wrote:@truckercm4_saturn.jpg

laurencebarker wrote:

ct1iqi wrote:@laurence:
tnx for background info!
Using the Armbian environment made an image on basis of arm64 architecture and kernel 6.5, but with just enough in the OS to let pihpsdr show its GUI. So basically just cli plus X11 and Openbox Have patched 'your' xdma driver into it and for now called it xdma2 as the xdma name is already used in that kernel. Had to add some small patches to make it compile as some functions, deprecated in kernel 5, were removed or changed in 6. I add the patches, just remove the .txt as .patch does not upload here.
The good news: the audio chopping is gone, using embedded pihpsdr
The bad news: 384 k rate with two receivers is the max. That brings CPU load close to 100%. The massive calculus that wdsp has to do is just to much at higher rates.
p2app works fine and even copes with 1.5 M rate. At 2 x 768k rate it gets the CPUs in the 30-40 % load range. So xdma as such works just fine.
When using embedded pihpsdr there is some effect noticeable in the audio from events in the GUI. When by mouse movement in a menu over a clickable box that box changes color, some effect in the audio is audible. So there are some issues with interrupt priorities it seems.
For some reason with embedded pihpsdr the microphone input is not working while using p2app and an external pihpsdr it works. wdsp, pihpsdr, and p2app are all 100% the current github ones, no changes made.

Rick is looking at the microphone issues with piHPSDR, but can't reproduce a problem.

We do know that the tip and ring of the 3.5mm jack are reversed between front and rear: the Thetis or piHPSDR settings should be correct for the rear connector.

Can you tell us exactly what happens please, with what settings?

w-u-2-o wrote:It would be good to get familiar with how to use Linux, if you are not already. See viewtopic.php?f=48&t=4724

laurencebarker wrote:somehow the primary image has been corrupted.

What FPGA image filename did you try to program using flashwriter? it looks like it was saturnprimary.bin which I think won't work. Try saturnprimary2023V11.bin unless you are using the newest Thetis dev release in which case use V12.

for a while now I've added new images with their version number in the filename.

The issue is: when we got fallback config properly working, we had to change to x1 configuration speed rather than x4 because that's all that xilinx supports with fallback. I suspect saturnprimary.bin uses x4

Torrey wrote:

laurencebarker wrote:somehow the primary image has been corrupted.

What FPGA image filename did you try to program using flashwriter? it looks like it was saturnprimary.bin which I think won't work. Try saturnprimary2023V11.bin unless you are using the newest Thetis dev release in which case use V12.

for a while now I've added new images with their version number in the filename.

The issue is: when we got fallback config properly working, we had to change to x1 configuration speed rather than x4 because that's all that xilinx supports with fallback. I suspect saturnprimary.bin uses x4

How do I navigate to saturnprimary2023V12 in Flashwriter? I am using the latest Thetis dev-2 version. The only available file via Flashwriter is saturnprimary.bin, saturnprimary(1).bin, saturnprimar(2)y.bin (multiple identical downloads).

Tnx and 73,
Torrey N9PY

KC2QMA wrote:Although I don't have the skills there must to be a way to create a small updater application to automate Firmware/Software updates on the G2 radios.

I agree its good to learn Linux but I believe there would be less chance of people possibly bricking their radios and also cut down on people asking the same questions over and over on How to update the G2.

The idea would be to have a Linux "G2 Firmware Updater" application installed on the G2 with a shortcut on the desktop.

When the user double clicks on the Updater icon on the desktop the application will automatically check GitHub to see if newer Firmware/software is available.

If no new Firmware/Software is available a message box would pop up and say “Your Firmware/Software is up to date”

If new Firmware/Software is available a message box would pop up saying "New Firmware/Software is available would you like to update now" (Click Yes Or NO).

If you click Yes and the application goes to GitHub downloads the necessary file(s) and shows a progress bar.

Once update is complete a message box pops up and says “Update Complete Please Restart Anan G2”

And that’s it, update complete!
Anyway just an idea to make using and updating the G2 easy and safe!

laurencebarker wrote:

Torrey wrote:

laurencebarker wrote:somehow the primary image has been corrupted.

What FPGA image filename did you try to program using flashwriter? it looks like it was saturnprimary.bin which I think won't work. Try saturnprimary2023V11.bin unless you are using the newest Thetis dev release in which case use V12.

for a while now I've added new images with their version number in the filename.

The issue is: when we got fallback config properly working, we had to change to x1 configuration speed rather than x4 because that's all that xilinx supports with fallback. I suspect saturnprimary.bin uses x4

How do I navigate to saturnprimary2023V12 in Flashwriter? I am using the latest Thetis dev-2 version. The only available file via Flashwriter is saturnprimary.bin, saturnprimary(1).bin, saturnprimar(2)y.bin (multiple identical downloads).

Tnx and 73,
Torrey N9PY

click "open file"

then navigate to /home/pi/github/Saturn/FPGA

you should find it there.

laurencebarker wrote:We are in the process of changing the firmware, p2app and piHPSDR to work with enlarged FIFO buffers in the FPGA. Debug code has identified that we were stalling the software waiting for a speaker FIFO to empty. Hopefully a combination of enlarged buffers and a different way for the software to transfer data will fix that.

Yes, I noticed the GUI activity can cause lost data problems with Thetis too. As at today it is very stable - no lost data packets - in normal operation at FS=1536KHz with one receiver; if RX2 is enabled a tiny amount of data may be lost.

2 receivers at 384KHz is at the upper end of what I'd expect from piHPSDR, I consider that to be success. Hopefully in a few months there will be news of a pi5 compute module.