If you are interested what feature request issues exist here is a link with the correct filter. If you are a KiCad user you might want to skim through that. There is a lot 712 issues at the time of writing… https://gitlab.com/kicad/code/kicad/-/issues?sort=popularity&state=opened&label_name%5B%5D=feature-request&first_page_size=20 #kicad #pcb #electronics #opensource

If it is not clear, if you think resizable logos/graphics in KiCad PCBs would be a useful feature consider upvoting the linked issue! Thanks~~~ ;) #kicad #pcb #electronics #opensource

Something kind of like that. Header pins sticking down on the bottom that will plug into the ROM socket. Then on the opposite end a ROM socket on top, and a set of header pins on each side of the ROM socket sticking up. Wires running over the top for one row, underneath for the other row.

The header pins next to the ROM socket would be useful for jumpering to adapt 32 to 28, and also to hook a logic analyzer onto.

Maybe I could make a "socket riser" out of proto PCB that plugs into the 28 pin ROM socket on the bottom, has a bunch of wires running up to a 32 pin socket on the top, and whatever jumpering necessary to go from 28 to 32. That's a lot of little wires to solder. And of course the MSX ROM is oriented one way and the CoCo is the other, so if I want to do both I have to make two adapters.

I was thinking I could maybe use an IDC DIP 28 pin connector to a ribbon cable to feed off the board. These things open up and you insert a ribbon cable without stripping the ends. Then you smash it back together and the [I]nsulation [D]isplacement pierces the insulation and connects to the wire inside.

But don't have a 28 pin. Don't think cutting down a 40 pin will work, and I don't see a way to add 4 pins to the 24 pin one.

Turned out to be a simple typo in scikit-rf. ​​ #electronics

Somehow my tests are still failing after the fix. Then I caught a yet another bug in scikit-rf while I'm trying to fix the original bug. ​​ Looks like a textbook shallow copy bug. #electronics

Found a fix to the UnknownThru sign ambiguity bug in scikit-rf. Instead of estimating the sign of the square root in comparison to an ideal Thru calibration (which is unreliable if the Thru is not so ideal), we'll just run the calibration twice with both signs, and pick whatever sign at each data point that produces a calibration that is closest to the user supplied approximate definition. Brute-force, but it makes no assumption of the imperfect Thru such as having a linear phase, so it should always work (if not, the user input is wrong, the algorithm is not wrong). #electronics