Making PCBs at home, or yet another guide to toner transfer

[andri]

I just finished creating my first PCB using the toner transfer method. There are a zillion of web pages on the Internet on how to do that and at first I found the whole process a bit daunting: what if I don't heat the paper enough, what if I leave the PCB in the etchant for too long and so forth.

Turns out that actually it's darn easy to make your own PCBs. If you're a beginner / amateur like me, I highly recommend trying it out. Obviously I wouldn't use this method for creating hundreds of PCBs, but for the one-off it's great.

1. I created my board layout in Eagle:


I forced the pads to be 20mil in size to make soldering easier and used wide 50mil traces (mostly because I didn't know how well narrow traces would transfer to the copper and what happens in the etchant).

2. From the nearest supermarket, I got the cheapest inkjet photo paper (1.80 € for 20xA6 sheets, 170 g/m²). I hid every layer in Eagle except bottom, pads and dimension and printed it on the glossy side of the paper. I also cut out a small piece of PCB that was slightly larger than my design; again, I didn't want to go too small as I was afraid if I cut it out too small it would get stuck in the laminator.

3. I bought the cheapest laminator I could find from an office supply store, cut out the printout, placed it in top of the PCB and ran it through the laminator for about 10 times or so, varying the angles a little bit.

I didn't really know how many times I should keep doing that; in the end, I used my gut feeling when the copper was so hot that you couldn't really hold the piece with bare hands... and ran it through the laminator twice or so more for good measure.

4. Although people recommend using hot water for removing the paper, I just filled a small tub with lukewarm water from the tap and dropped the board in. After 10ish minutes I tried to pick it a little bit with my finger; if it doesn't come off easily, let it soak for some more. After the paper had softened enough, it was able to remove most of it and the results were looking rather promising:



I was in no hurry and let it soak some more until I was able to get most of the paper off with just rubbing the board with my thumb. I didn't want to push too hard or use a brush because I was afraid I would damage the toner.

It was looking good in the water, so I dried the board and saw that there was a fair share of paper left on the board (it was just transparent when it was wet):



I tried using a sharp scalpel to remove some of the paper; while I was able to break the larger parts I wasn't really able to remove any and I think it's not worth doing that.

At this point I dropped the board back in the water and decided to take the risk of scrubbing the board with a rough cloth. It worked pefectly and my fears were unbased as I didn't damage the toner on the board at all; in fact, I think I could be a lot more aggressive with cleaning the board without damaging the traces.



There were some minor faults where the toner hadn't transferred well (or I had scrubbed it off -- who knows), so I touched those places up with a fine-point waterproof marker that I usually use for writing on CDs.

5. For etching I bought some sodium hyposulphite (I'm not really sure about the English name, the formula should be Na₂S₂O₂) from the local electronics store. The guide said to mix it 1:5 with ~50C water and the board should be etched in about five minutes.

I think I messed up the ratio and as I didn't have a heater it took me a while to figure out how to keep the solution heated (place the plastic cup into a larger cup of hot water), so it took me about 40 minutes to etch the board. At some point after 10 or so minutes I was wondering if it was working at all as I could see no change on the board; soon enough the solution started to turn blueish and after 20 or so minutes the board was done:



After removing the toner with some acetone:



I think the board turned out pretty darn good. It's interesting to note that the small border line that was transferred somewhat OK was almost completely gone; I guess that is below the lower bound of what's achievable with this kind of home etching.


Next up, finding a drill, transferring a "silkscreen" on the top and soldering the board.

[Kiriakos-GR]

I had do lots of PCB's just with a marker and acid . 

More complex designs and multiple thin lines , needs an more high tech process.

[andri]

Given how well the "V1.0" came out, I'm fairly confident you can go down to at least 12 mil traces; probably even narrower. But indeed, this is something that is useful for making small one-off boards in your home, it's not an industrial process

[Zero999]

Nice work!

I use clay coated paper from old magazines for toner transfer, it's the best thing I've come across and it's free.

Do you know of a quick way to set force the hole size for all of the components in Eagle? I like to make the holes on the mask tiny (0.1mm) so I don't need to use a centre punch before drilling and it would be helpful if Eagle could do that, otherwise I can see myself doing it in MS Paint .

[TheDirty]

You can go down to 8mil or even 6mil with the same process you are using with practice.  You should drop a ground polygon fill in order to reduce the amount of etching needed.

There's a ULP called drill-aid you can run to decrease the size of all the holes.  It uses a separate layer to overlay the pads to make them smaller.  You can run it using file->run or simple type 'run drill-aid 0.7' in the command line.  Change 0.7 to whatever works for you.

EDIT: Leave the paper on the traces as long as it does not go outside the toner area.  The paper helps block the etchant and reduces rash.

[DavidDLC]

Another option instead of silkscreen is to use something like Liquid Thin from MG chemicals to do a tinplate

I have used it and I'm very happy with the results.

[shodan]

I used a solution of muriatic acid (hydrochloric acid, but they call it that at the hardware store)  3$ for a liter of 30%, and some hydrogen peroxide 30% at the drug store (I think it's to bleach your hair ?) 50%/50% in a plastic tray, etch fast, and it's transparent so you can see what is happenning and how far the etching is done

I used a clothes iron to heat the board it worked great on the 1st try

you can use http://cpc.farnell.com/1/1/19102-tin-plating-powder-600-021-mega.html to add a tin plating (there's probably a cheaper way to do this)

next up is directly printing to the board with a modified inkjet printer
or machining the copper right off the board with a diy bench-top mill (good for production) and you can do that reasonnably cheaply

it's probably possible to do multilayer & vias on the cheap too

[Zero999]

I've done a silk screen before. I just did toner transfer on the other side of the board and covered it with a conformal coating to protect it from being scratched off. It worked quite well and looked quite professional, despite being black. I think using a silk screen is a good idea for more complicated boards, as it helps the assembly, debugging and repairing and it doesn't cost much and is easy to do.

[DJPhil]

I used a solution of muriatic acid (hydrochloric acid, but they call it that at the hardware store)  3$ for a liter of 30%, and some hydrogen peroxide 30% at the drug store (I think it's to bleach your hair ?) 50%/50% in a plastic tray, etch fast, and it's transparent so you can see what is happenning and how far the etching is done

I've been working with this too. I've just finished aerating two small jars worth of solution that needed recharge. When I first started I used the raw mix as described above, but it didn't stay active long and tempted me to discard it. I found that I hadn't prepared the mix completely.

Done properly it's indefinitely rechargeable by bubbling air through and adding acid occasionally. You'll have a steadily growing supply, so you might eventually need to give some away or dispose of some.

Here's the link I used to get started. If anyone's struggling with this method I'd be happy to offer advice.

Hope that helps.

[Zero999]

You can go down to 8mil or even 6mil with the same process you are using with practice.  You should drop a ground polygon fill in order to reduce the amount of etching needed.

There's a ULP called drill-aid you can run to decrease the size of all the holes.  It uses a separate layer to overlay the pads to make them smaller.  You can run it using file->run or simple type 'run drill-aid 0.7' in the command line.  Change 0.7 to whatever works for you.

EDIT: Leave the paper on the traces as long as it does not go outside the toner area.  The paper helps block the etchant and reduces rash.

Thanks for that, I'll give it a go, be sure to get more questions.

[Simon]

or machining the copper right off the board with a diy bench-top mill (good for production) and you can do that reasonnably cheaply

it's probably possible to do multilayer & vias on the cheap too

for a decent setup your looking at £800+ and yes you would need something that decent for production, Although I'd love tobe able to do it for prototypes and a few one offs

[shodan]

you can spend a lot less or a lot more than 800 gbp on a pcb milling machine

this is what I had in mind



although I would use something better than drawer slides !


"I told him what I was going to use it for and ... he didn't believe me"

[Simon]

not a bad idea, I suppose my worry is getting the electronics to work with the mechanics and the software, The idea is good but I'd heavily question the accuracy of that milling table, just one screw holding a bearing in place that stops the table going haywire ? yea right, typical yanks ! but the principle is good, I might even have a go one day

[shodan]

you can make this several orders of magnitudes more reliable by using slightly better components
I think the guy wanted how simple it could get and probably set himself as a rule to use only "hardware store technology"

then you start improving and tinkering at the worst points

there's tons of software to control the electronics, a very popular one is Mach3, there are many OSS project for this too
basically they are G code interpreter with an interface to let you know what is going on and to tweak settings on the fly

software side isn't much more complicated than hardware side

if you want to buy a ready made product with all the accuracy constraint resolved you can get a 500$ micRo
it's here (but for the price I would want more Z travel so I'm not getting one !)

http://lumenlab.com/d/index.php

personnaly my approach will be to start with a press drill and just make a stiff 2 axis stage

[Zero999]

I always cringe with I see instrucables, most are crap but this is one of the better ones. I wish they would not allow articles which haven't been subject to peer review and would get rid of the compulsory sign up to view large pictures.

Simon,
You're right, it's not good enough for PCBs.
http://www.instructables.com/id/Easy-to-Build-Desk-Top-3-Axis-CNC-Milling-Machine/step9/Engraving/

Now that I got the machine back together after making the instructable and I did some engraving and made a PC board.
Cutting plastic is no problem but doing the PC board the bit went a little too deep on the left side of the board and took out all the finer traces. This is when you start tweeking on it. Just take some aluminum foil and put it under the rail of the Y axis. So as the stage travels left to right the height of the bit should stay the same.
Notice I'm just holding the material down with masking tape. What I like about this thing is, it's easy to fix these kinds of problems because it's all made from simple elements.

Software could probably used to help calibrate the Z axis If you know the thickness of the copper on the PCB, you should be able to build a map of where it's cutting too deep/shallow and subtract it from the Z dimensions.

It'll probably never be good enough to make complex PCBs, but it's probably more than accurate enough for drilling holes for components which would save a lot of time. Another option is to use it to make the thicker traces and etch the rest to save PCB etchant but that would waste time and it wouldn't be as accurate.

[Simon]

bottom line: you need a damn accurate CNC mill for PCB's thats why you best just getting the sable or being very clever

[scrat]

I worked sometimes with a professional cnc machine for PCBs (it was from LPKF). Milling head is suspended on an air cushion to stay precisely on the board, while you can adjust trace width by rotating a screw knob (since the mill is conical). High end ones have a camera, which can be used to calibrate and read fiducials (for top-bottom alignment). On our machine, in spite of specifications, precision in positioning all over the board was around 2/100 or worse, and drills were always out of center.
I wonder how accurate homemade CNCs can be...

[charliex]

my cnc does drill off center, its going off for a new floating head design thats supposed to fix it, but its only a mil or two. so i basically create the PCB in away it won't have problems with a slightly off centre drill,s o i make the pads slightly larger. I can cut traces of 6mil all day long. the camera fiducial is a doddle to use.

basically just work around the limitations.

[scrat]

my cnc does drill off center, its going off for a new floating head design thats supposed to fix it, but its only a mil or two. so i basically create the PCB in away it won't have problems with a slightly off centre drill,s o i make the pads slightly larger. I can cut traces of 6mil all day long. the camera fiducial is a doddle to use.

basically just work around the limitations.

But it is an industry-made one, isn't it? If it was a homemade machine, well, my compliments to the maker!
About our CNC, when we complained for that off centered drills (which were not regularly offset, they were quite random) they tried to deny the evidence, and at last admitted that the previous versions of the machine (even the low end) did not have that problem (as I also verified on my own). Since ours was a high end one, it seems really absurd to me. They justified by saying that the head was not well designed for the camera weight, but I think that in their recent models they are saving costs for mechanical parts, thus robustness is going down...

[charliex]

yeah mines a 'pro' machine like the lpkf. i did have some of that same thing going on, but they offered me the upgraded to the new model for free so i can't complain.

it is a bit annoying the drill is off slightly, and i can't really see the reason for it. but it is small enough that its easy to work around, the new head is supposed to make it go away altogether.

[Zero999]

There's a ULP called drill-aid you can run to decrease the size of all the holes.  It uses a separate layer to overlay the pads to make them smaller.  You can run it using file->run or simple type 'run drill-aid 0.7' in the command line.  Change 0.7 to whatever works for you.

I've just tried it and it works brilliantly - thanks.