My design would have a pivoting arm holding the PCB in the fluid and bump or rotate the arm with a cam. Why put the big container full of caustic liquid in motion when you can just move the PCB?

I've been nerd sniped thinking about this. Latest thinking: simple sliding bar clamp to hold two edges of the PCB. The PCB hangs down into the solvent tank. The bar on the clamp drops into vees at both ends. One end of the bar is bumped with a motorized cam, the other has a spring. The clamp+PCB oscillates back and forth and the arms agitate the fluid. When the cycle is complete you just lift the clamp+PCB out of the tank. The tank doesn't move. Tank shape can be optimized to minimize working fluid and agitation.

The YouTube video acknowledges the over engineering in the title.

No, I came to the comments after having this exact reaction.

> Or just scrub the board with a bit of ferric chloride and a sponge.

If you use ink pen and transfer pads like I did decades ago, the sponge would almost certainly wipe the pads off. The best way that I'm aware of to ensure the copper is being consistently covered by the solution is to use an air pump to produce bubbles under it. Those used in aquariums are quite cheap and durable.

Yes! The sponge technique takes a little practice to get consistent results, but it works well (at least for small boards), and lets you use microscopic amounts of etchant, which is good for cleanup. I find it’s especially good when trying to do relatively small traces, because you get such fine control over the process. (I have to say though, nowadays I’d usually rather wait six days for JLCPCB to do it for me.)

Hah, that's a great idea!

Can you recommend a Website/service?

>Why would you want to make them yourself?

Welp, not everyone lives in the US or Europe. Living in South America my only options are either spend ~$100 for a two weeks _fast delivery_ or have it for free ~2 months later. Not many options for the hobbist maker over here. There are local services but they are more expensive (and slower if that's even possible) or just middleman for PCBWay or JLCPCB.

There are reasons.

Mine: phased Vivaldi antenna arrays. Iteration is painful if you have shipping in your run debug loop.

My thoughts exactly. It's fun as a hobby I guess, but sitting down and soldering a non-trivial board sounds like a PITA vs paying $50 for a respin of 5x boards with full pick-and-place assembly, delivered in <5 days.

€40 to receive it in 3 days or €0 to have it in 3 hours.

This is not a cleaning process (in the sense of "removing dirt"). One way to create PCBs is to use a board with a copper surface, put a protective mask on the locations that should have traces or pads and then etch away copper from everwhere (else). In the end, only copper under the protective mask will remain. Etching is faster when the solution is moved around during the process. This machine does exactly that.

It would make a poor machine for marinating food. Most of the foods you'd marinate should be refrigerated during the process. A much better way to marinate is throw everything into a ziploc with as much air removed as you can to maximize contact. Flop it around every now and then if you like.