I mean the whole thread, and yeah this particular test is brutal... I think we're seeing the limits of the design, very nice.
The star of the show is however the programable load (at least for me)
-
Feeling lucky? Check out the AK 24th Anniversary Giveaway thread and join in the fun!
Sansui AU-717: Restoration with new boards
- Thread starter SebastianH
- Start date
-
- Tags
- pcb restoration sansui au 717
The star of the show is however the programable load (at least for me)
SebastianH
Member
Waiting on the arrival of the PCB... I guess I'll have to wait until next week.
In the mean time I'm vibe coding a web-based ui for my measurements (based on my usual measurement code). Wrote exactly one line of code for this to work (didn't bother to ask the AI to fix this issue). And for such a project I don't really care about the code quality too much at all as long as it works for me.
Current state:
The measurement data are stored in numpy files, the rest uses a simple SQLite file.
It's really low effort, so I guess I (or the AI) will add more features
In the mean time I'm vibe coding a web-based ui for my measurements (based on my usual measurement code). Wrote exactly one line of code for this to work (didn't bother to ask the AI to fix this issue). And for such a project I don't really care about the code quality too much at all as long as it works for me.
Current state:
- Support for the QA403 via its REST API
- Support for Siglent SDS2x04X Plus via Ethernet/SCPI (i like to use the scope for example for measuring frequency response because the QA403 is very limited in this regard ~96kHz)
- THD vs. frequency
- THD vs. power
- Noise & Hum
- Create a new Device under test
- Create a new test Session
- Create a new Measurement and select the measurement type.
- Specify the measurement parameters
- Start a Run
- Compare results (here you can see the variation between two measurement runs with the same parameters)
The measurement data are stored in numpy files, the rest uses a simple SQLite file.
It's really low effort, so I guess I (or the AI) will add more features
That's clever, an LXC and voila! What a great idea.
There's a guy in diyaudio doing a very nice software kit, at least it look fantastic to me.
Just in case you haven't seen it
github.com
There's a guy in diyaudio doing a very nice software kit, at least it look fantastic to me.
Just in case you haven't seen it
GitHub - youtube-at-vach/MeasureLab: DIY tools for generating test signals and measuring harmonic distortion. Built together with AI. Feel free to use, modify, and experiment.
DIY tools for generating test signals and measuring harmonic distortion. Built together with AI. Feel free to use, modify, and experiment. - youtube-at-vach/MeasureLab
github.com
SebastianH
Member
Cool stuff, didn't know about it. If I'm not mistaken it's primarily meant for sound cards. As far as I know there is an ASIO driver for the QA as well, but haven't messed with it. Also, it might be somewhat difficult to control the attenuator. I hope QA will improve their software support which is somewhat limited, but other than that it's an absolutely awesome product.
Yes, will deploy it on my proxmox server in an LXC.
It now has a PDF report function too This is too easy.
Yes, will deploy it on my proxmox server in an LXC.
It now has a PDF report function too
This is too easy.
Exactly an avenue to explore.
You probably already know this work, but just in case
QA40xPlot Thread
Hi On plot sw I have this frame On github I see this where there is the Impedance test I haven’t SOmething wrong by me?
forum.quantasylum.com
GitHub - MZachmann/QA40xPlot: Control a QA40x with a standard Windows GUI
Control a QA40x with a standard Windows GUI. Contribute to MZachmann/QA40xPlot development by creating an account on GitHub.
github.com
Yes, will deploy it on my proxmox server in an LXC.
It now has a PDF report function too
SebastianH
Member
Maybe
I think the attenuator thing might be too annoying. Yes, I've seen it before. Unfortunately it's for Windows, so it doesn't really solve my problem. It could be used as an inspiration for using the "bare metal" implementation though (I think)... But I don't really want to invest that much time to get the same results.
For the moment the official software runs on my Windows lab machine (need Windows for a different application as well) with a reverse proxy to expose the API to my LAN. I wish there was an official and supported python library for Linux.
yeah, that would be great.
SebastianH
Member
Just a quick update: left channel driver board rev D arrived & is assembled. Very good clearance now (pre-driver heatsink, capacitor vs. chassis). Did some preliminary testing, seems good so far, pretty much what I expected. Now I have to do some more testing and design the updated right channel driver board.
SebastianH
Member
I did, but deleted those by mistake 
I didn't bother to restore the files from a backup because I have to redo the measurements anyway as soon as the right channel is done.
The boards are similar. I had an LLM write a scipt to mirror the board and then I changed everything I had to (essentially >= 3 pin components and then make everything fit). Unfortunately the multi-turn trim pots are not symmetrical, but have to be accessible so I had to move those to the side a bit. The layout is 99% complete. But I have to thoroughly check everything and place the reference designators.
I might end up tweaking the component placement (especially the trim pots) on the left channel board to match the right channel a bit better, but I certainly won't build this version myself. That would mostly be to please the eyes
Won't have much time to work on the project for one or two weeks, but I expect to at least finalize and order the right channel board. I'll keep you posted.
Right channel board:
I didn't bother to restore the files from a backup because I have to redo the measurements anyway as soon as the right channel is done.The boards are similar. I had an LLM write a scipt to mirror the board and then I changed everything I had to (essentially >= 3 pin components and then make everything fit). Unfortunately the multi-turn trim pots are not symmetrical, but have to be accessible so I had to move those to the side a bit. The layout is 99% complete. But I have to thoroughly check everything and place the reference designators.
I might end up tweaking the component placement (especially the trim pots) on the left channel board to match the right channel a bit better, but I certainly won't build this version myself. That would mostly be to please the eyes
Won't have much time to work on the project for one or two weeks, but I expect to at least finalize and order the right channel board. I'll keep you posted.
Right channel board:
Having done the mirroring manually, I would very much like to look at the script that the LLM created. The mirroring effort was made more difficult by my lack of understanding of the operation of KiCad and digging through the script would be very educational I think.
SebastianH
Member
Yes, KiCAD 10. To make sure: I didn't use the LLM directly to mirror the board, I used an LLM to generate a python script that does this operation deterministically.
Have done it manually before as well. Didn't like it too much... Here you go: Code Browser. 99.9% is AI generated. As is the README and the diagram attached. Image quality isn't really good enough here. If you're interested, I'd recommend to have a look at the diagram in the README.md.
It took a couple of prompts to get something I was happy with. It's neither the most complicated nor the easiest task. The most difficult part is that footprints may have different origins/points of reference. Just don't expect it to produce perfect results. In my instance it was a major help. Probably should have done it as a KiCAD plugin...
Attachments
Last edited:
Thanks, much appreciated! It will be interesting to examine the code.
SebastianH
Member
Nice! But now comes the part where you have to check and (most of the time) re-route all >=3 pin components. 
yupe, but it saves a good chunk of time if someone needs to mirror a board.Nice! But now comes the part where you have to check and (most of the time) re-route all >=3 pin components.