@Luke: In these two years I have not done much improvement to the process, I only just used it as-is to make some more PCBs, but I was keeping track of the various solutions to update the guide with information about the long-term storage of the solutions.

What may be interesting to you, is that after about two years of storage in a well capped container, my palladium activator solution turned from dark brown to pale yellow, and when it did, it stopped working. The pale yellow solution leaves a brown coating on the PCB copper that is not catalytic towards the electroless copper solution, and the same brown coating cannot be easily removed, making it necessary to restart from scratch with a new PCB.

If I find some time I will explore more the chemistry of the palladium solution, but the TL;DR is that if your palladium solution is pale yellow it does not work, while if it is dark brown it works. This can also be used as a clear visual indicator that the solution has gone bad.

I updated the Process 01 document with this information.

So, I tried some electroless plating using a modified version of this excellent documented process from Hydrogen Time.

 It didn't actually work in my case. I would like some peer-review or help to check this process.

I don't care about actually making a PCB, so the other steps such as PCB drilling, photolithography and electroplating thicker Cu aren't relevant. I'm only interested in the steps of electroless Cu deposition onto a nonconductive dielectric substrate such as glass.

I decided to modify this process using a Na2PdCl4 stock solution. This isn't part of the original process, but I decided to add this step myself and I don't think it should make any difference. We're mixing PdCl2 and NaCl in the Pd activator solution anyway, so it's exactly the same as making a Na2PdCl4 stock solution from PdCl2 and NaCl. PdCl2 itself does not dissolve in water, so this is the way to get it into a soluble solution.

This is easier to measure and dispense in small portions, easier than measuring out 20mg of PdCl2. 1.0 g of PdCl2 is added to 50ml H2O and 0.659g NaCl, with thorough stirring. H2O is added to make up to a total volume of 100ml, stirred more, the solution is filtered to remove any insoluble residue (there was essentially none). This Na2PdCl4 solution provides a water-soluble alternative to PdCl2, which is itself not soluble. The solution is deep brown but clear, like cola or dark tea in appearance. 10ml of the Na2PdCl4 stock solution is equivalent to 0.1g PdCl2 combined with 0.0659g NaCl in aqueous solution.

Then I made the Pd activator solution, using the above Pd stock solution, SnCl2 and NaCl.
200 ml H2O
10ml Na2PdCl4 stock solution
11.25g SnCl2 dihydrate
38.69g NaCl
11.8 ml 36% HCl
Stirred thoroughly, and H2O added to make up 250ml final volume.

I don't think any of these changes are significant - it should be pretty much equivalent? There is so much NaCl in the formula that the Pd would form the [PdCl4]2- complex when the NaCl is added anyway. I scaled-up the volume by 5x, I used the Na2PdCl4 stock solution, slightly reduced the HCl volume because my concentrated HCl is 36% not 34%, and I reduced the NaCl by a tiny amount due to the equivalent NaCl in the Na2PdCl4 solution. The final reagent has a pale yellow colour. However, the videos show a dark brown, almost black liquid. But it should have everything in exactly the same concentration.

Then I made the SnCl2 post-activator reagent.

375 ml H2O
42.5 ml 36% HCl
10g SnCl2 dihydrate
Thoroughly stirred and made up to a total final volume of 500ml with H2O.

(Scaled up to a volume of 500ml.)

Volume of HCl reduced slightly as I used 36% HCl, not 34%. This dissolves completely to give a colourless solution, unlike the cloudy solution shown in the original video.

Then I made the Cu plating reagent.

I didn't have any disodium EDTA salt, only acid EDTA so I used the equivalent amount of EDTA combined with two equivalents of extra NaOH. I think Na2EDTA will be OK to substitute with EDTA combined with an appropriate amount of extra NaOH. (2g * 2.5) + (0.69g * 2.5 * 2) = 8.45g NaOH. My formaldehyde is a bit old and had some precipitate of white, insoluble paraformaldehyde in it, so I filtered it before use. This is commercial aqueous formalin which contains some stabilisers such as methanol.

375 ml H2O
6.3 g CuSO4 pentahydrate
12.6g EDTA
8.45g NaOH
10 drops Triton X-100
4 ml 38% Formaldehyde
Stirred thoroughly, topped up H2O to 500ml final volume.

Volume scaled up to 500ml. Forms a clear blue solution with no undissolved material.

A glass slide is cleaned with ethanol and dried, then immersed in the Pd reagent with gentle agitation for 5 minutes, then removed with no rinse and put in the Sn reagent for 5 min with gentle agitation. Then the glass substrate is gently washed with H2O and put into the Cu reagent, for 40 minutes. There are no gas bubbles and no visible reaction, no copper deposited. DI H2O is used for all steps where H2O is needed, not tap water.

After spending so much time in this process, my plans are currently to exploit it "as is" for a couple of hardware projects. I'll be pushing it towards smaller traces and via sizes though, as I'll be making SMD boards.

As for reducing the number of steps, the big improvement would be an electroless-only process, skipping all the electrolytic plating steps.Of course, a quick and sure way to shave off steps is to tin plate the board (or just leave it bare copper) instead of going for a solder mask.

Many of the projects I make are one off, the prototype is also the final unit, so I do care about making a "good looking" PCB that will last and not oxidize. I may consider adding steps and go for silkscreen, maybe. On the opposite end, I guess some may be interested in the process to make a rapid prototype to test an idea quick, and then go for a professionally made PCB after they're sure it works. For them, an electroless-only process that stops at the bare copper would be the quickest and most effective.

Hey, I really enjoyed watching you work through this process. Really looking forward to improvements -- how many steps do you think you can shave off? Great looking PCBs too -- very professional! Maybe you should try silkscreening them too...