The Mitochondrion 3.5 – An Excessive Glowstaff (6 of 8)

Tips for making circuit boards – (tiny, dense, double-sided boards in your kitchen):

Like a fool, I tried to pack far too much into far too small a space. Hence the PCBs had to be equally dense and tiny. But then again, fitting electronics into a 25 mm tube is never going to be easy.

(For those of you who don’t make circuit boards, the process is: design board layout on computer complete with all the holes for components, pads to solder down the legs on chips,, and tracks to make up the circuit; print that layout onto plastic film using a laser printer; iron the film and a copper-clad board together so that the ink layout transfers from film to board; etch the board and the ink stops the etching, leaving the copper circuit laid out under the ink on your board; scrub off the ink and you’ve got your circuit.)

The solution to fitting everything in was long, thin boards and lots of time spent tweaking circuit layouts to squeeze everything in. Oh, and time spent working out just how far I can push iron-on transfers and etching. That’s a 44 pin QFP package above, 0.8 mm pin spacing. That’s how far I can push it.

So, here’s my tips, for using laser printed masks, Jaycar press n peel plastic film, and persulphate etch:

1) Different laser printers give different performance. I’ve a Brother at home and I’ve never got a good result with it. Masks printed on the Kyocera at work come out great.

2) Just because you can make 0.25 mm wide tracks, doesn’t mean you should. I found the error rate went up harshly when I tried to do anything finer than 0.5 mm.

3) Cleanliness matters. Finger grease was my main cause of failed board making. Don’t touch the ink side of the transfer film ever. Scrub that copper-clad board hard with detergent and steel wool till shiny all over. Dry with paper towels and hold by the edges only.

4) Keep everything flat. When ironing designs from film to board, you need strong, even pressure all over. This requires a working surface where you can put all your weight into the iron, and not have the surface bow under you.

5) Lining up masks for double-sided boards is a swine. I would put the two masks face-to-face, no board between them, then drill two 0.8 mm holes to locate them. Then drill corresponding holes in the board blank. Clean all the swarf away, then assemble boards and masks with drills through the holes to act as locating pins. I had a hard time getting the holes perpendicular enough, but I was drilling these by hand. Don’t drill them by hand.

6) Transfer film stretches. I only found this to be a problem with boards over 200 mm in length, but it’s just a pain in the arse. When ironing the boards into place, try to keep them flat but not under tension.

7) Wait until the board and film is cold before peeling off the film. Run it under a cold tap, for longer than you think you need.

8) Touching up masks: Despite all the pedantic preparations, I’d often get drop-outs and gaps in the masks after transferring them to the copper, especially around holes. So, etch-resistant pen and delicate hand work. For fixing a gap in a narrow track on the mask, surrounded by other tracks, I found it easier to just chuck a blob of ink on there, and then re-cut the gaps between the tracks using a blunt knife to scrape away the excess ink. Doing this by hand is possible, but it’s much easier with guides. I’ve a set of metric drills in 0.5 mm increments and the shanks make great parallel standoffs. Hold the board in a vice and scrape lines parallel to the vice jaws.

9) Touching up etched copper to get rid of shorts: Use a sharp knife and don’t try to do it freehand. Use drills as stand-offs, as in the previous point. If you can design your boards with traces in parallel, then scrapping off shorts between tracks is much easier. But really, it’s easy to have traces farther apart, if you’ve got the space.

10) Tin pads straight away, then lacquer straight away. Clean copper takes solder much more easily than even week-old stuff. And the lacquer just keeps it shiny.

11) Don’t put vias under chips. Especially if you’re making the vias by sticking some wire through a hole, soldering the ends and filing flat. This is time-consuming to make and a nightmare to debug.

12) Removing surface mount chips from boards – use thin nichrome resistance wire. Remove what solder you can with wick (some kinds are better than others, I had good luck with Goot). Run some fine nichrome wire through the gap between board, legs and chips, behind the legs. Clamp one end of the wire between the board edge and vice jaws. For the other end, wrap it around something you can pull on and gently pull the wire at 90 degrees to the line of legs. Apply soldering iron to the one leg that the wire is pulling against and when the solder melts, just that leg should lift. Repeat for all legs on a side. Using this technique, I had pretty good results removing SMT chips from homemade boards, without lifting the pads and screwing the boards. It still doesn’t work every time though.

13) File boards to size after you’ve etched them. I wanted my traces right up to the edges of the boards, within 0.5 mm. I haven’t got a board guillotine, just a hacksaw, so I cut a rough size, made the etch and then filed to get the final size. This was tedious.

14) Stainless steel kitchen tops rock for chemical resistance and ease of cleaning up. They’re anti-static too.

With all of those, I could happily make some pretty detailed boards in the kitchen, suitable for surface mount chips with 0.8 mm spacing. Still, a 40X1 in surface mount is lots of time and hard work.

11 thoughts on “The Mitochondrion 3.5 – An Excessive Glowstaff (6 of 8)”

  1. I wimped out on home-made PCBs, because I really wanted plated holes and solder mask. It’s a shame there is no reasonably priced prototype-grade board options in NZ, the only ones I’ve found are all overseas and shipping kills it. But I suspect the volume is just too low to make it work.

    I do love me some soldermask. SMD is much nicer with it. 🙂

    1. Yeah. Any good suggestions for Oz-based prototype boards? Currently I’m designing the next board to be made in the kitchen (i.e. fat traces, no solder mask…) coz I don’t want to wait & pay for the shipping from UK/US/Bulgaria/wherever.

      1. The boards I’ve been getting have been done in China and shipping was actually very reasonable. However, the minimum order is a single panel, which is a lot of boards (~23 NTP server boards, which are 51mm x 80mm, for your stuff it’d be all the boards in the world), and spikes quite a bit if it’s not single-design for a panel. They are a 1 week turnaround tho which is pretty damn good, and the quality is excellent.

        Nothing I know of in AU. I have seen a few places but they are unwilling to provide any pricing without a gerber, and TBH it’s just wasting my time to have to go thru that process to find out it’s no cheaper than a whole panel from China. And the only reason I don’t rush into doing that is I feel sad I have heaps and heaps of scrap boards as a result of Design Fail!

          1. For bonus complication, I’m after boards 240 mm long so Seeedstudio are out, they’re 200 max. Olimex get really expensive over 160 mm long, so I might end up in the kitchen if I’m in a hurry, or waiting for Gold Phoenix or someone like that.

          2. Kitchen or Gold Phoenix seem like the best plan for anything that long.

            Does it need to be 240mm? I’d have assumed you would have gone with a very repeated design. Tho I guess trying to secure it is the biggest problem and the more segments the harder it will be to not have it fall apart.

            It’s a shame you can’t get tubing with a ~1.7mm channel down both sides, it would make it a bit easier to fix in place.

          3. This is the board that holds and connects up everything. On the board is the Arduino, the sensors, the audio analyser stage, the memory, two DC-DC converters, six status LEDs and five connectors. It can only be 21 mm wide. So yes, it’s going to be long and there’s no way around that. Oh, and because it’s so tight in there, all the components have to be on one side. And there’s two high-current pass-throughs from one end to other, because the two halves of the monster battery pack are at either ends of the tube.

            Sorting out the layout was a serious brain-bender…

          4. Yeah I think you have to make a very clear decision early on how the board will actually be produced, and stick to that method. If you’re getting something professionally made, then you can design it down to 7mil track/space and vias anywhere you like. But that’ll never be done home-brew, at least not without building your own photoetch setup 🙂 (And I do know someone who seems to be doing that)

            I kinda knew I was never going to fit the NTP server into a home-brew board, and once I accepted that, it was a lot easier to make it compact. Tho QFN still give me a lot of grief for layout 🙁 Curse you exposed pad!!

  2. via under chip is a surprisingly difficult habit to break.

    I’m having horrible problems with getting the ink to stick to the copper, no matter how hard I clean it — scrubbing with 3m abrasives, chemical etching, only touching with clean metal tools, and the stencil still won’t stick. says I’m not heating it enough but I’m heating it sufficiently to scorch the paper. Grumble. I’m wondering if it’s an issue with the laser printer. I’ve had luck using a fine-point sharpie as a resist pen.

    My friends who do double-sided print oversize — larger than the board — with really heavy fiducials around the outside, then put the two sides face-to-face on a light table, line them up, tape them around the edges, to make a pocket, and slip the board inside. This might also help with stretch/distortion.

    1. Can you try a different kind of ink? This probably means trying a different brand of printer. I believe the Kyocera printers have a different kind of drum and hence a different kind of ink, so I recommend hunting down one of their printers.

      Also, may well give myropcb a try – have just spent the last four hours sorting out a design that should work and just be manufacturable in the kitchen, but you know? Life is short and I’d rather be working on the house.

Leave a Reply

Your email address will not be published. Required fields are marked *