DIY plated through holes

[charliex]

yep pretty much

[cpirius]

Alright, I'll try the 3350 laxative powder ;)

Thanks

[bearmos]

good luck, post back with progress/questions/comments :) !

[ric86]

Hi... thanks for sharing, this is very useful.

My question are:
1. How to etch the board?
2. Will thruhole copper be lost during the etching process?
3. How to protect it?

Thanks

[bearmos]

Hi ric86, welcome to DP!

Etching either needs to be done mechanically (by a CNC Mill) or you'll need to apply a reverse mask, tin plate what you want to keep, and use chemical etching.

For the reverse mask, you'll need a negative of the artwork (the opposite of what is used in "normal" toner transfer or UV process). Apply this to the board (via toner transfer or UV etch resist) and then tin plate the PCB. Since the through holes are exposed, they'll also get tin plated.

After tin plating, strip off the resist you applied and chemically etch the board (the tin acts as an etch resist).

There's a comment on the last page (electroplating 4) that addresses this partially.

Think and Tinker has the most comprehensive guide to PCB fabrication I've come across (and is what the Twilight Robotics writeup is largely based on). They explain in-depth what needs to happen (even going into multi-layer PCB's).

Here's the information from Think and Tinker on tin plating - you'll be interested in "pattern plating", as described above.

Let me know if you have any more questions! Also, if you're interested in other prototyping info (soldering SMD's, using solderpaste, good flux recommendations, etc) - be sure to check out the rest of DP - there's a lot of great information here, as well as a lot of great people that are interested in helping others!

[cpirius]

Hey all,

I have my plating tank up and running now and I've used it for a few boards. So far through hole platting is working well, but there is definitely room for improvement. Below are pictures of 41mil header holes that I plated.




The plating is not complete, but it is usually connected, and with plenty of copper to allow solder to wick through the whole and make a solid connection to both sides. I've been reading about making a pulsed constant current load for plating. One that reverses the current for a small percentage of the time at some frequency. This is suppose to disrupt organic buildup and provide a more even platting. Has anyone tried that?

I've also had good success with 26mil via holes, but very poor turnout with 12mil holes. The main problem seems to be activating the holes without them getting clogged with ink. Even when I have an active vacuum going under the board while I add ink, they still clog about 60-70% of the time. Has anyone tried 12mil holes? Any better methods for activating them?

I also put together this video of my overall PCB manufacturing process so far:

https://www.youtube.com/watch?v=yqjKohZX2t8

[bearmos]

Hey there, cpirius, sounds like you managed to source the PEG and you've got something up and running, congrats!

The plating is not complete, but it is usually connected, and with plenty of copper to allow solder to wick through the whole and make a solid connection to both sides. I've been reading about making a pulsed constant current load for plating. One that reverses the current for a small percentage of the time at some frequency. This is suppose to disrupt organic buildup and provide a more even platting. Has anyone tried that?

See the comments section of the constant current source article. Andrew Frazer has shared a schematic that looks pretty promising for reverse plating. Although, others (namely Null Space Labs in this thread) have had fairly good success without reverse plating. They were using a brightener (Janus Green B). Other things that I've heard help are anode bags (a very fine filter material around the anode bank).

I've also had good success with 26mil via holes, but very poor turnout with 12mil holes. The main problem seems to be activating the holes without them getting clogged with ink. Even when I have an active vacuum going under the board while I add ink, they still clog about 60-70% of the time.

What kind of graphite are you using, is it mixed really well, did you follow the ratio outlined in the article? The graphite should be extremely fine. If you have the right kind fo graphite and it is mixed in the proper proportions to the ink, you shouldn't have problems with clogging at all (it sounds like your graphite/ink ratio may be too high).

Has anyone tried 12mil holes?

I've done down to .015" vias without problems using this method (vacuuming out with a shop vac - a really strong vacuum). As a reference, the graphite I'm using is 44 microns (max) diameter.

Any better methods for activating them?

Someone left a comment [url=twilightrobotics.com/prototyping/electroplating1]here[/url] with a link to a purely chemical approach as well as a short explanation. It doesn't sound like they personally tried it and neither have I. Please post back with your experience if you decide to try it.

You also need to be mindful off the current you're plating with (this is a function of board surface area, described here) and plating time.

[cpirius]

Hey there, cpirius, sounds like you managed to source the PEG and you've got something up and running, congrats!

Thanks! I'm using the 3350 laxative powder and it seems to be working, though I don't really have a basis for comparison...

See the comments section of the constant current source article. Andrew Frazer has shared a schematic that looks pretty promising for reverse plating. Although, others (namely Null Space Labs in this thread) have had fairly good success without reverse plating. They were using a brightener (Janus Green B). Other things that I've heard help are anode bags (a very fine filter material around the anode bank).

I have a similar plan for a micro-controller based reverse current setup. I'm going to make a test board with different size holes and run a few of them with different power timings and see what the differences are. At least, I will when I have the time to build the constant current load with current reversing...

What kind of graphite are you using, is it mixed really well, did you follow the ratio outlined in the article? The graphite should be extremely fine. If you have the right kind fo graphite and it is mixed in the proper proportions to the ink, you shouldn't have problems with clogging at all (it sounds like your graphite/ink ratio may be too high).

I'm using 5 micron graphite, mixed in the ratio from the article, or even a little less graphite. I shake it well before using, but it still seems kind of gritty. However, yesterday when I made some more boards I noticed that wiping the ink off the board might be the real problem. After vacuuming I could see light through the holes, but then after wiping it with a paper towel, it pushed material into the whole and some where clogged, and they would no longer vacuum out. Further investigation required.

I've done down to .015" vias without problems using this method (vacuuming out with a shop vac - a really strong vacuum)

I've ordered some more drill bits to try different sizes as right now I have nothing between 12mil and 26mil.

Someone left a comment [url=twilightrobotics.com/prototyping/electroplating1]here[/url] with a link to a purely chemical approach as well as a short explanation. It doesn't sound like they personally tried it and neither have I. Please post back with your experience if you decide to try it.

This sounds awesome! I'm working on sourcing the chemicals needed. Thanks for the info. For some reason the comments section is not showing up for me in chrome, so I missed this. Will let you know how it goes.

You also need to be mindful off the current you're plating with (this is a function of board surface area, described here) and plating time.

I've been plating at 9-10 A per sqft. Haven't tried other currents yet.

[bearmos]

After vacuuming I could see light through the holes, but then after wiping it with a paper towel, it pushed material into the whole and some where clogged, and they would no longer vacuum out. Further investigation required.

I was using a vacuum, oven cure, scrub/wipe. Are you curing the ink before trying to take off the excess. If the conductivity is being broken between the hole walls and the copper layers, this could also be a problem.

Someone left a comment [url=twilightrobotics.com/prototyping/electroplating1]here[/url] with a link to a purely chemical approach as well as a short explanation. It doesn't sound like they personally tried it and neither have I. Please post back with your experience if you decide to try it.

This sounds awesome! I'm working on sourcing the chemicals needed. Thanks for the info. For some reason the comments section is not showing up for me in chrome, so I missed this. Will let you know how it goes.

I'm not a huge fan of additional chemical baths, although if i did this more I would probably consider it - it seems like as long as the chemistry works it would be more reliable (and faster) than going to the trouble with the ink/carbon.

I've been plating at 9-10 A per sqft. Haven't tried other currents yet.

Just making sure:) Sometimes lower current densities can give better results. Anode bags and brightener are probably worth a shot as well if you're planning on sticking with the process.

[bearmos]

It looks like cpirius has made a video of his entire PCB prototyping process:
Drilling holes, electroplating, mechanically etching and cutting out the board.

Results look promising:

routed PCB (with plated through holes)

mechanically etched PCB with plated through holes

Nice work, cpirius!

[zelea2]

Can you please tell me what kind of software are you using for the isolation routing? I also have an A4 CNC machine which is not specialized for PCB making and the problem I have is that the PCB is not planar enough. My machine can also probe height by electrical contact but I don't have any routing software to make use of a height map.

[bearmos]

Can you please tell me what kind of software are you using for the isolation routing? I also have an A4 CNC machine which is not specialized for PCB making and the problem I have is that the PCB is not planar enough. My machine can also probe height by electrical contact but I don't have any routing software to make use of a height map.

PCB-GCODE (linked to in the comments on youtube). Note that this software is supported by the yahoo group (not Dangerous Prototypes). I've done some experimentation with mechanical methods of making sure PCB's were level. You can read about them here if you're curious. The auto-leveling feature of PCB-GCODE appears to be new since the last time I used the software.

[bearmos]

There also a linux based command line program called pcb2gcode which also sounds like it does height mapping:http://sourceforge.net/apps/mediawiki/pcb2gcode/index.php?title=Main_Page. That is hosted on sourceforge. I think ahri has used this before when he was experimenting with milling PCB's.

Can you please tell me what kind of software are you using for the isolation routing?

I personally use EMC2, cpirus is using Mach3. The gcode needs to be generated in such a way as to make use of the height map, it sounds like both pcb-gcode and pcb2gcode do this now.

[bearmos]

Isn’t there any G-code optimizer that would reduce all the unnecessary travels of the routing bit? It seems like 30% of the total time is spent moving the head long distances without actually cutting.

Yes, I used this. It worked well for the etching but not for the drill files (I think). I'm not sure how it would work with the height-mapping that's generated now.

pcb-gcode has a really weird/fragmented community, there's pcbgcode.org which appears to be nearly dead and then there's also a yahoo group, which appears to be semi-active. I've been out of it for a while, so maybe someone with some more recent experience could fill the rest of us in.

[t0mpr1c3]

Isn’t there any G-code optimizer that would reduce all the unnecessary travels of the routing bit? It seems like 30% of the total time is spent moving the head long distances without actually cutting.

That sounds like a travelling salesman problem. Dynamic programming is I think the usual way to solve that. There was a blog feature this week on genetic algorithms which are also applicable.