Home made SMD PCBs

[RCMR]

A lot of my prototyping and small-run jobs use mostly through-hole components and I frequently make my own PCBs using the toner-transfer method.

For runs where you only want 2-3 units and turn-around times are important, this turns out to be the most cost/time-effective way (for me) to do it.

However, I have a couple of projects that will mandate the use of all-SMD components (size and other factors dictate this).

The thing I'm rapidly discovering is that when using very small SMD components in a dense layout, 2-layers is really not enough.

Does anyone have any tips for making life easier?

[sacherjj]

I used DorkBots PC batch orders.  I've only done 2 layer boards, but they also do 4 layer.  For any serious SMD, you can't really so it at home unless you have some way of doing a solder mask.  Almost impossible to do 0.5mm chips without masks.

[Psi]

You can add unnecessary 0R resistors (in a package large enough to allow a track underneath) to make the layout a little easier.

But when dealing with really small SMD (below 0805) i found the toner transfer method starts to fail. Pads would end up smaller than they should be due to side etching. And overall track thickness was a bit too small for the toner transfer method to produce good results.  However, if you've found a way to make it works then great

I agree with the above post though, sending away to get boards done really does make a huge difference. Not just because they look awesome but they are so much easier to solder and work with compared to home made boards.

[RCMR]

I used DorkBots PC batch orders.  I've only done 2 layer boards, but they also do 4 layer.  For any serious SMD, you can't really so it at home unless you have some way of doing a solder mask.  Almost impossible to do 0.5mm chips without masks.

I regularly use some SMD devices (TQFP with 0.8mm pitch) without a solder mask and it's no problem to solder these.  Likewise with SOIC packages.

[sacherjj]

I used DorkBots PC batch orders.  I've only done 2 layer boards, but they also do 4 layer.  For any serious SMD, you can't really so it at home unless you have some way of doing a solder mask.  Almost impossible to do 0.5mm chips without masks.

I regularly use some SMD devices (TQFP with 0.8mm pitch) without a solder mask and it's no problem to solder these.  Likewise with SOIC packages.

Your Toner Transfer boards must come out better than mine.  One thing is for sure, doing SMD eliminates all the drilling.  I think drilling is the most annoying thing of doing my own board.

If two sides are not enough, you might be able to do 2 sided plus a single layer for a 3 layer setup from home.  Although you would only want the bottom layer to have traces and would need to solder the vias first, before you put the single layer over it with the copper down.

Theoretically, you could continue to add single layer boards for at home multilayer.  However, it seems like it would get harder and harder to keep registration and if you missed a via solder before you added the next layer, you would be hosed. 

[ruku]

Just did this not an hour ago with the toner transfer method. It was my third attempt, but this is only my first time doing toner transfer! Dual layer board, smallest traces are 12 mil for the MSOP packages. There's a TQFP-44 (16 mils) and several SOPs (20 mils). 20 mils pullback on the copper fill.



The 16 mil traces on the TQFP were a little sketchy, and the 12 mil traces on the MSOPs got through on a wing and a prayer. God bless SO packs and their fat 20 mil wide pads! I had to doctor the board just a little with some masking tape in a few places, but I have continuity everywhere I've checked so far. The big question is getting all the "through holes" soldered and aligned! Thank goodness I have access to a CNC mill and some resharpened carbide bits for this one!

A few things I do to make life easier... make your via's wide, and make the drill hole about 10 mils wide. This gives you a big fat copper pad to solder to, which comes in handy when your layers turn out misaligned. I'm also using the top layer as a power plane. Perhaps inadvisable, but it was nearly impossible to route the board otherwise. (There's actually two power planes in this pic... 5V all around, and +3V3 under the main IC.) Everything goes to ground through a via... There's about 97 of them. Bleh. Maybe I should have made the bottom layer my power plane.

Was using the backing for a sheet of envelope labels as my transfer paper. The waxy surface worked like a charm! I didn't even need to wet it down to remove it, just carefully peeled it off after ironing it for a while. If I saw any breaks, I'd just fold it back down and iron a little more. Usually worked! I originally was pre-heating my boards, but that didn't seem to positively affect the process. On a whim, I tried using an overhead transparency. I started to get a little transfer, but ended up melting the sheet all over the bottom of my mom's iron. Whoop!

Having done toner transfer, the UV sensitized board method, and having a PCB shipped, this is definitely the least reliable for what I'm doing... but it was the cheapest! It was also really nice to be able to scratch off toner with a pick where I didn't want it. If I had access to UV materials though, I'd definitely do that for this board. This was a stretch, and added a few too many gray hairs!

PSL mentioned 0R resistors... Maybe I'd try that with 1206's. 0805 seems too small for that... but it seems feasible. What are you all using for your transfer paper? Everyone seems to be using photo paper, but I can't fathom it! Seems far too thick! Then again, the backing sheet *was* a little flimsy...

(Edited a bajillion times for content and formatting.)

[RCMR]

I was using inkjet photo paper but it often leaves bits of paper between traces that even the most enthusiastic soaking and scrubbing with a toothbrush won't shift so now I find a nice glossy "House and Garden" or fashion type magazine and use pages from that which have very little color ink on them.

Once you find the right magazine, this paper is not only very cheap but far better than the glossy inkjet paper.  After transfer, the paper comes away with a little soaking in cold water and there's no bridging of even the finest gaps.

One way to ensure good line-up is to place diagonally opposing registration points on the two layers.  Once you've transferred the first layer, drill the holes (about 0.8mm) and then use pins through the same holes second printed piece of paper and through the registration holes.

Doing this, I seldom end up with more than 0.1mm of offset between layers.

The biggest problem these days is finding a source of echant though.

[Psi]

I use press-n-peel blue paper for pcbs, it works really well but there is an art to doing it right and even when it works perfectly there is a limit before toner tracks are just too thin and fall off before etching is finished. Especially when using ammonium persulfate as it only works when the water is HOT and the heat tends to melt/soften the toner.

Here is a 100tqfp i did with press-n-peel blue. This pushed the method as far as i found was possible

[nukie]

Psi- That's a good work. I haven't try those press-n-peel before, after seeing your finished board I think I might try it.

[joelby]

Have you tried ferric chloride? The Ultrakeet site compared ferric chloride with ammonium persulphate using PulsarProFX (a variant of press and peel) and found the former to be superior for a number of reasons. I've lost all enthusiasm for etching boards at home but will try ferric chloride next time.

[ruku]

I was using inkjet photo paper but it often leaves bits of paper between traces that even the most enthusiastic soaking and scrubbing with a toothbrush won't shift so now I find a nice glossy "House and Garden" or fashion type magazine and use pages from that which have very little color ink on them

...

Do you need to find a completely blank page, or does it not matter what's on the page? Have you ever wound up with text or pictures from the page on the PCB after ironing?

[baljemmett]

Have you tried ferric chloride? The Ultrakeet site compared ferric chloride with ammonium persulphate using PulsarProFX (a variant of press and peel) and found the former to be superior for a number of reasons. I've lost all enthusiasm for etching boards at home but will try ferric chloride next time.

Anybody have any thoughts on how sodium persulphate compares to ammonium persulphate?  I've been using the former, because that's what came in the cheap 'so you want to etch a board?' kit I picked up locally on a whim.   It's given me good results so far with press'n'peel transfer but it does take a fair while, even sitting in a warm water bath.  Mind you, being transparent and not staining everything in sight seem to be handy advantages over ferric chloride for my kitchen-sink adventures!

[RCMR]

Do you need to find a completely blank page, or does it not matter what's on the page? Have you ever wound up with text or pictures from the page on the PCB after ironing?

I avoid pages with lots of dark ink beause I found that large areas of black can be a bit of an issue.  Lighter colors don't seem to matter at all though and don't stick to the PCB during the transfer process.

[IanB]

Anybody have any thoughts on how sodium persulphate compares to ammonium persulphate?

I don't suspect there is any difference. I believe it is the persulphate ion that does the etching and it will make little difference whether the other ion in solution is sodium or ammonium. From my reading it works best with gentle agitation at about 40°C, and a typical etch will take up to half an hour at this temperature.

I recall from long ago when I did some etching with ferric chloride that the ferric chloride worked quite fast, but it might still have taken 15 minutes. On the other hand ferric chloride is messy stuff to deal with and you really don't want to have it anywhere indoors like a kitchen or bathroom where it can stain things.

I'm amused today how people say copper sulphate (the result of etching) is poisonous when as a child I used to play with big containers of it to make crystals and do other experiments. How times change...

[Psi]

Psi- That's a good work. I haven't try those press-n-peel before, after seeing your finished board I think I might try it.

Thanks
Just be aware that there is an art to it. It takes practice to get the hang of it and find a method that works for you. (Time under the iron, pressure, movement etc.)
Also my pic there is a bit out of focus, if it was in focus you would see that those fine tracks are pretty wonky. They vary in thickness by 50% long their length.
So yeah, tracks that thin push the technology quite a bit.

Have you tried ferric chloride? The Ultrakeet site compared ferric chloride with ammonium persulphate using PulsarProFX (a variant of press and peel) and found the former to be superior for a number of reasons. I've lost all enthusiasm for etching boards at home but will try ferric chloride next time.

I only tried ammonium persulphate once and i couldn't get it to work. I had to make the solution so hot to get it to etch that the toner on the thin tracks melted off.

So yeah, i always use ferric chloride. It works well, ya just have to watch you don't splash it anywhere as it destroys clothing and stains most surfaces.