Playing with this one: github.com/jrowny/p4kvm work well. There is an nice patch proposed: github.com/jrowny/p4kvm.... This will be a good use for my stack of esp32-p4 boards! The esp32p4 seems to be very suited for these types of tasks. Also made my immich frame version 3.0 compatible.

Anybody knows what this is? I just soldered 4 pcbs from same batch. Out of the 4; 1 has these weird grey/silver on the pads.. All soldered on the hotplate within a few minutes from eachother, same temperature and time. I had this before with aisler PCBs, curious what it is? works fine, just ugly

Today saw a website that sells a fake casio FX-82MS with build in bluetooth for p2p communication, googling and notes. It was expli. sold for cheating on exams. These are impossible to spot, it looked identical to a €5 euro fake one I bought a while back. Consequence, no more calculators on exams.

Fun bug, via mqtt I could control all the boards from 00 up to 40, but not board 08 and 09. Locally via buttons it worked fine. The number parsing uses strtoull. I am happy it took me only a few mins to realize my mistake, as I would not have been surprised if it would have cost me hours to find.

3 more done between meetings.. 8 left.. Luckily they work as expected and work significantly better then the old setup.

I got, slightly unreasonably, annoyed when asking how a feature worked in an application and I had to go to the source code to figure it out, the answer was. > Try to use AI for this? it has an MCP integration. Maybe I don't want to hook up my app to an AI to use it feature to draw a graph.

Making my own life hard, by having a nicely packed pcb. Fun by hand! So far it seems to work, now to add the remaining channels. Weirdly enough, the channel (that use the copy layout in kicad) had some production error, 7/90 (18pcbs) failed. It passed DRC and all. Should be fixable with a bodge.

Doing stepper driving on an ESP32. Never used the mcpwm blocks before, mostly used fpga's for this. I think I got it working. At the moment I can do a full speed homing cycle and control it with 0.02mm steps.. Hopefully it works with all the UI stuff added and multiple axis, if not ULP.

I just noticed, that the memory in my 1 year old pc now costs €100+ more then I spend on the whole PC. (CPU, mobo, psu, expensive cpu heatsink, case and memory) Crazy.

Thanks to AMD's direction with xilinx, working on alternatives. @raspberrypi.com trying to plan for the next 10 years (~700 boards a year), are there follow up plans for the pi zero 2 W? I see its production up to 2030. Will a follow up be same form-factor/pinout?

Finally got the 25MHz crystal in, I should have gotten a bigger one. But it sits, and it works.

@timberborn.bsky.social There is an award for 250 beavers, 500 beavers, but no award for 1000? The beavers do play hell on FPS, even though both CPU and GPU have low-ish utilization.

Love you can now buy a esp32 display module, create a PCB for it. Then with 0 effort, download a 3d model (where you can enable/disable parts) and generate a same model for your PCB. Match everything up, and see if you made any errors in all the sizes, alignment, etc. It makes things so easy!

Watched this one this morning, www.imdb.com/title/tt0327.... Enjoyed it, wonder how I missed it up to now.

Played a bit with wifi settings and tweaking the queue'ing and got it to play without deadline miss. I wonder if I can push it to RGB666 instead of RGB565 (uses 3 bytes per pixel, so a big increase in space needed).

I hate shipping companies.. I was home, did not ring the doorbell, it was small, so it would fit in mailbox. No, just drop it in front of apartment complex.. Luckely I noticed it in time. and this is how it looked..

2nd version soldered.. Shacking hands is getting a bigger issue, needed two hands to place components, one to hold the tweezers, one to stabilize first hand. The hotplate way luckely helps straightening out the components. It all worked out, but I wonder at what point I cannot do this anymore :-/

Got my own si5351c board in from @aisler.net. The solder job is very bad (shaky hands today), but so far it is findable over i2c. And yes, I forgot to order the 25MHz crystal, but luckely I ordered this one because it can take in an external clock (all the hobby boards I found have the si5351a).

Another ali haul. Got a round 1.54" screen. I've been using 1.5" screens for years now, they are cheap ( ~3€), high res (240x240) and good viewing angles. This one is *slightly* larger, higher resolution 360x360 pixels, cheap (3.90 €) and is qspi. Hopefully this one is the new go to screen.

Made a small case for the Icepi Zero. Link here: makerworld.com/en/models/26...

The opensource tools for the zeropi/lattice are great. Got a platform with our riscv32i cpu up and running with little effort. A synthesis takes 10~14 seconds, even for a small design that is pretty amazing. Got PLL working, Tri-state input/outputs and the uart that is hooked up to the usb.

Got a icepi zero to play with. First step is to see if I can get my RiscV running on it and the sdram to work. Hopefully all with opensource tools!

*sigh* where ever you go, spam/e-begging will follow you around..

Now replaced the npn transistor in my flux 8120 by one a bit more appropriate. It still works, did a quick calibration check and it seems to be still pretty good, but not into spec. Still better then several cheap dmms I had lying around.

ESP32-P4 PPA weirdness. I wanted to make images fit the frame a bit better by cutting up a small slice at the top/bottom and scaling it a bit less to fit. However when I add that to the ppa_srm_oper_config_t, I get these weird little block pattern. As far as I can see, I do nothing wrong.

My ugliest fix ever.. and regrettable not without damage. (to desolder transistors, 2 pads (that did not connect) where lost, I managed to save a third. But fluke nixie dmm works again. The NPN died, PNP measured fine out of circuit. After replacing the NPN, the PNP failed. 8's and 9's all around!

Dove into my broken Fluke 8120A multimeter. A nice one with nixies. Found why it won't display 8/9 well. Most likely one (or two) transistors in the ADC 4th bit feedback line are broken. Now to figure out if i can still get these, or to find a suitable alternative.

The Reflective LCD waveshare board I've been using has some odd quirks: The RTC (with bat) seems to 'loose' state flashing the ESP, or pulling the main battery. Kinda defeats the purpose of the main RTC. I did manage to get the idle usage down (no deep sleep) to 2mA with some spikes on redraws.

A while back I got quiton jc-esp32p4-m3-dev and made an immich photo frame out of it. I saw a really cheap small esp32p4 board from dfrobot, the firebeetle 2 esp32-p4. At a local store ~15 euro, but has same spec (16mb flash, 32mb psram) and a DSI/CSI port, no eth. Drop in replacement for the frame

PCB came in today to duplicate my 10MHz signal. Now I have 6 50Ohm output on my reference clock. (forgot to take picture of soldered board before installing) So far it works fine. My Siglent AWG now uses this as reference clock (and according to my scope/freq counter) it increased it accuracy sign.

My desktop CPU is weird.. It has a large cpu cooler. Under full load the core sits at 78C. All cores have a 5.1GHz boost. Even then the cpu cooler is ~ 30C.. With heatcam it all looks cool, even at the cpu. I think to get cooler core temp I need to delid the cpu. Not sure I want to take the risk.

Finally got the 0.5x adapter for my microscope in. The different in image quality is night and day. (I still wonder if the 0.35 I had was broken.. ) No more purple glow around all bright stuff (I couldn't get rid off) and very limited focused area. Hopefully I can make better images of my work now.

Weird vintage find at work.. A BNC to banana plug without ground. Also allows a wire to be screwed in. For my next 200MHz measurement 😝

Had a nice example of how DDR memory keeps state for some time after powering off. Working on an Zynq setup, where FPGA writes to a FB in the shared DDR. I triggered an overcurrent causing it to power off. After remotely resetting the efuse, starting it up the test image was there, with some noise.

Wayland rendering is "fun".. You only know after the first 'frame' what exactly info is of the target monitor. This includes DPI and scaling factor.. This makes showing an UI at a certain physical size hard. Fractional scaling adds more fun, the application only gets an integer factor.

Working on running 80 fpga zynq boards over nfs, I wonder if I should keep using ubuntu.. 24.04 has many things broken (working in 18), only in the arm version. Got it working in the end, but required several 'firstboot' fix scripts to get everything going correctly. Maybe time to switch to debian.

@aisler.net oops it's broken.

Almost done, waiting for PCB to arrive from @aisler.net to add the finishing touch. (actually wire up the 6 10MHz outputs)

Took a picture of frontplate I was working on.. it got a whole black 'spot' after processing (it looked fine, then after a sec it showed this). Took the same picture in 'pro' mode and its fine.. Wonder WTF google camera app is doing..

Making a PCB to 'duplicate' my 10MHz reference signal. A quick test on perf board to test this out. Green below is the output. Not perfect, but good enough.

Ordered a 0x5 c-mount adapter on aliexpress. It listed 'DHL, Post-NL,..' as shipping. They used PDN. They made 1 attempt in the middle of a working day. Claimed they 'called' me. No more attempts, no way to contact them. The one number (mobile) I dug up, gives full voicemail box. The Worst...

Not wanting to drill into the case, I've 3d printed some caps for the holes that normally has the carrying handle. One has a nice 3d printed 3/8" thread that nicely takes my bnc connector.

My whole 'stable' clock started after rescuing a HP5315A. I wanted to calibrate it. Now it spiraled to give it an upgrade, a OCXO or external input. I saw people hacking it in using the OCXO option, but this was expensive (100 Euro). Going through the PCB I saw the extr. input header was there.

This is control that does 'rapid' control first if the offset is > 0ppm. Then, after 1 hour, it goes to slower control with a 1 hour window and updates every 20minutes. It looks like I need to switch to a longer window/interval for the control later. Currently on chip I keep around 2 hours of data.

Learned the hard way: If you want va_arg with 64bit arguments on 32bit hw you need to make sure the top of the stack is 8byte aligned. GCC generates an alignment ptr = ((ptr+7)&-8); and then uses that to find the 2 32bit values on the stack. It will put 64bit arg. on A2&A3, A4&A5 slot in stack.

On my OCXO I keep track of offsets in 30 seconds interval (PIO counts for 30 seconds). From this I can see that during an hour I have 12 blocks that have an 'offset' (6 +1, and 6 -1) . Not sure if this is instability in my measuring or the OCXO. Still very good and better then I have use for.

In my reference clock project we now can: * accurate count the 10MHz clock * Count against GPS PPS to get offset over longer period. Next step is to 'tune' the OCXO based on the calculated offset. The board uses a 204k pot-meter, allowing you to tune OCXO Vref between 1.96-2.32V.

Fun with SMD markings.. A divider, with 363 resistors, and 688 or is it 68B? 688 makes no sense.. but the last 8 looks more 8 then B, yet it was. Took me an (embarrassing) while that some where in 3-eia others in eia-96..

In latest ALI haul, got a few SI5351 clock generators in. Not for myself, but to help somebody out (results will be published). They are easy to use, not accurate, but good enough. Another thing in this haul was a bnc to banana plug cable, I bought one before locally, this was <1/3 the price.

hmm 'bug' in software, my opnsense router eaten through a large amount of data on the nvme... Spare: 100% Spare Th: 10% % Used: 26% Read: 140,971,432 [72.1 TB] Written: 56,692,629 [29.0 TB] In 23530 hours..