When are jumper wires on a breadboard entirely optional? When it’s the latest version of [Kevin Santo Cappuccio]’s Jumperless, which uses a bunch of analog crosspoint switches (typically used for handling things like video signals) to create connections instead of physical wires. There’s even an RGB LED under each hole capable of real-time visualization of signals between components.
If this looks a bit familiar, that’s because an earlier version took second place in the 2023 Hackaday Prize. But things have evolved considerably since then. There are multiple programmable power rails, ADC channels, a rotary encoder, and much more. The PCB design itself is fantastic, including the probe which acts like a multi-function tool for interacting with the whole thing. The newest version will make its debut on Crowd Supply in just a few days.
It’s open source and made to be hackable, so give the GitHub repository a look if you want a closer peek. You can watch it in action in a brief video posted to social media, embedded below.
We (and I cannot stress this enough) are so back.
Jumperless V5 lets you prototype like a nerdy wizard that can see electricity and conjure jumpers wherever you want. And the wait is nearly over for this particular superpower, it launches September 23rd on @crowd_supply. pic.twitter.com/VRr5H9sJD5
— Kevin Santo Cappuccio (@arabidsquid) September 20, 2024
Okay, very clever. what happens when you have to connect a bunch of signals between the top and bottom sides? Seems like it would run out of LEDs to show the connections pretty quickly, making this useful for only the simplest circuits.
After a certain number of “wires” where it would become too dense, it switches over to “lines” mode, where it just lights up the whole row and doesn’t draw anything between them.
Each net gets its own unique color, and if you’re unsure, you just tap the row and it will highlight everything that’s connected to it.
The (new?) tap-to-wire functionality shown in the video looks very intuitive. I can’t wait to test one out at Supercon!
So the probe on the OG Jumperless was a pure hack, like I thought of it after the hardware was made and luckily had 2 extra GPIO pads in the corner. So it needs to disconnect everything and then scan through every row looking for a PWM signal from the probe. That also means if you have parts on the board that connect 2 rows, it has no way to tell which side you’re touching and you need to click through the options to pick one.
The new probe is also a hack, but at least it part of the design from the beginning. It uses 92 resistors different valued and an ADC to know what you’re touching (honestly I couldn’t believe that actually worked). So you can do it with the connections in place, which is way more useful.
But yeah, I’ll definitely have some floating around at Supercon for everyone to play with.
Weird. Not having looked into it, I was under the impression that it would use a blunt canule with a polymer optical fiber in it, terminating into a photodiode that would pick up the phase or pulse width encoding the correspnding LED coordinates.
Apparently not the case. Could make the pointer wireless though.
I wonder what switches are used inside a “electronic patch panel”, the ones without any cables hanging out the front.
This: https://www.valiantcom.com/patch_panels/images/vcl-xc-64-e1-f.png
vs this: electronic patch panel: https://www.reddit.com/media?url=https%3A%2F%2Fi.redd.it%2F09x8g20r9vcy.jpg
Script kiddies for HW?
Could just slap a cpld or fpga and just vhdl or verilog wires to pins
Digital breadboard
Just flash the connection and attach to chip
Or just flash the entire circuit to the chip
I had a look at the KiCad project, and the GND plane is absolutively atrocious. It’s a 4 layer PCB, quite filled with “stuff” and tracks, and although there is overall some adequate fill with GND zone sections, the interconnections between all these is absolutely minimal. This PCB needs at least some 200 extra stitching via’s to connect the parts in the GND plane.
And also, in the schematic, he puts comments like:
Decoupling Caps (the number of caps on the minimal RP2040 reference design was total overkill)
And he’s writing that just above the only 5 decoupling capacitors for the whole PCB.
The PCB apparently still “works”, but signal integrity and EMC will be horrible, with all the switching for the WS2812 LED’s going on, combined with analog opamps.
While staring at the A0 schematic on a single sheet, I’m also having some trouble finding the “programmable power rails” For a project like this it would be really nice to have some power supplies that can do +/- 12C @100mA to feed some opamp or other analog circuits. There is a switched capacitor charge pump that generates + and – 9V. Verifying connectivity in such schematics is quite horrible experience and I did not fully analyze it. But in the end I did accidentally find the L272D used in the circuit. That’s a power opamp that can deliver up to 1A at it’s output. That is a nice find and usage in this circuit. Unfortunately it takes this power from the switched capacitor’s output (+/- 9V) and that has a sag of 5V with a 100mA load. That is a design choice I do not understand.
I do wish this guy lots of success with his (now going pretty commercial) project / product, but for a commercial project this is very simply unacceptable. My advise is to at least consult someone with electronics / PCB design experience, and also listen and implement to the advise you get from such a person. If you don’t understand the advice, then educate yourself until you do.