I think I've etched one board, ever. After that experience I just use breadboards from vector or rat-shack, or I dead-bug things on an unetched copper groundplane. If I feel a need for an "etched" board I slice islands out of the copper with an x-acto knife and surface-mount through-hole components onto the pads (think of dead bugging with pads).
If I really needed accuracy I'd make the circuit on Eagle and send it out to
formatting link
or another fast-turn house. If you want to be trapped by proprietary software you could use
formatting link
Both of these latter methods, incidentally, work extremely well if you are being paid by the hour -- they can do it much cheaper than you can unless you're getting starvation wages.
The ARRL Handbook covers all of these methods.
--
Tim Wescott
Wescott Design Services
http://www.wescottdesign.com
When I tried the iron-on method, I had to go back and fix up a couple of traces but nowhere near 20-30% of em.
Indeed, direct positive PCB's are preferable. But even then you really do have to check each and every trace because it's a tricky process.
For the past 5 years I've been using
formatting link
for all my stuff and had 100% success for single and double-sided stuff. Single-sided, double-sided, ridiculously high lead pitch surface mount stuff, I ain't ever gonna go back to etching this stuff myself.
Amen. Between each stage of the process is a step called INSPECTION. Obviously, the OP is skipping this.
The first responder (Burridge) mentioned a pen. When I did it, I used fingernail polish for touch-ups and cleaned the board with acetone at the end. . .
Yeah. If this is just for the hands-on experience, that's one thing. Not handing this off to the pros as early as possible is just nuts these days.
I've been using AP Circuits for my manufacturing (quantities ~100), but for one-offs they're pretty pricey. If I know I need just one of something, I generally use Olimex instead. Olimex's quality is not quite as nice but the price is a tiny fraction. (Unfortunately, they have different standard drill racks and different minimum trace sizes, so it's inefficient to proto with Olimex and then manufacture with AP.)
AP's "proto one" service gets you 4 to 6 small boards for something like $60, as I recall. No solder mask or silk, only sheared to size, but that's just like doing it at home. I've done dozens of different boards, 8/8 design rules, and they've been perfect so far.
What's Olimex's minimum lot charge for a 2-side plated board?
You can etch stuff yourself (I did as a kid) with 100% success rate but not plate through. I made some really fine pitch 'circuit board transformers' that way since the commercial ones cost way too much.
An additional issue with that is what to do with all the chemicals afterwards. Some people store that forever and then one day the container rusts out or the cat knocks it over.
Doh! My mistake. My production vendor is Advanced Circuits, not AP Circuits. Yes, I see that AP Circuits advertises "Two 2-layer PCBs for under $50.00" (for a 1.5" x 1.5" board, their minimum size); a good deal.
Olimex's deal is that you get one 100mm x 160mm board for about $40 (full featured: double-sided, silk-screen, solder mask, plated-through holes), including shipping. The nice thing is, if your design is smaller, they'll panelize and cut for free - so for instance, I recently got 25 very small boards for $40. However, it takes a couple of weeks to get the PCB (in the USA, anyhow), because it ships air mail from Bulgaria. You can pay (a lot) more for faster shipping.
I've been using Pulsar film, available from Digikey (182-1003-ND and
182-1021-ND) to do a dozen or so boards so far, including some fine-pitch SMT designs, though all single-sided.
The process is fairly straightforward, but has taken me a lot of experimentation and frustration to get it right. As a result I'm going to write up what I've learned one of these days soon, so others don't have to figure the same stuff out the hard way. Here's the quick version, off the top of my head, with as much detail as I can in this format:
0) Design your board carefully, according to a number of rules that I still need to codify. Some easy ones: use copper pours wherever possible to remove less copper; make sure you set your trace/gap numbers to a comfortable level, I haven't gone below 10mil/10mil yet though I easily could; set the pad annulus to 15+mil so you have some drilling slop; don't let the copper pour or other traces get too close to the
*inside* of any SMT chip pads, you can't wick that stuff away.
1) *Laser* print the *positive* artwork onto the transfer paper, cut to size and taped (Avery laser label bits) to a piece of backing paper (both to feed into the laser, and for positioning).
2) Thoroughly rough up (Scotchbrite pad / steel wool) every sq. mm of the board, then wash very carefully, possibly using acetone as the last step (to remove oils *and* copper dust). Do not touch the copper after this step.
3) Laminate/iron the artwork to the board. Pulsar resells a GBC laminator that has both high heat and pressure, and will accept up to
0.03125" boards (not 0.0625", big deal, IMO) for ~$50.
4) Dump the board paper-side up onto the *surface* of a bowl of water and wait (tick tock) for the paper to literally fall off (takes longer, but much more consistent results, don't rush this step).
5) Carefully wash everything off the toner surface, and make sure there are no oils left either.
INSPECTION: if there are any missing areas, clean the copper off with acetone, go all the way back to step 2 and start over. DO NOT SKIP ANY OF STEP 2, do it all over again.
6) Laminate/iron the green film onto the board, let cool, and carefully peel back the plastic.
INSPECTION: if the film wrinkled and left visible toner on the board, or if any toner is visible on the peeled off plastic, go back to step 2 and start over, as per previous inspection step.
7) Grab some gloves, a sponge (combo glove+sponge works great for me), and a few ounces of FeCl or equivalent. Get some FeCl on the tip of the sponge, and lightly *scrub* the board all over until you can see all the way through the copper-free areas when held against a light.
8) Thoroughly clean the film and toner off with acetone, scrub it down with steel wool, and clean it again. Cleaning is your friend.
9) Dump the board in some Liquid Tin (MG Chemicals 421-500ml) for
5-10min. When removing the board from the liquid tin, do not touch the copper+tin surface until you've washed it off and let it sit for a minute. I kept getting fingerprints on the tin and finally realized it came from *above*, not underneath the plating. The stuff is soft. It's also extremely nasty stuff, you *MUST* take even small amounts to a disposal site. MG says even 50ml diluted into 5 gallons of water is too nasty to dump down the drain. You can use extremely fine-grade steel wool, *very* lightly, to clean up the surface a little bit, but as soon as it shows a little copper tint to it, you're done. Very nice color effect though.
10) Use appropriate drill bits, such as the re-ground ones you can get cheap in sets of 25. Only problem is that there no guarantee of size selection with those. This would be a good time to thank yourself for cranking up the pad annulus and leaving yourself some room to mess up.
11) SMT components can be soldered by hand, but I've had far better luck using solder paste and a toaster oven. Digikey has it (KE1507-ND) but as a rather high price ($42 for 35g), and you'll need to find a syringe tip to use, the smaller the better. I'm still getting a feel for exactly how much paste to use for different parts, and the time/temperature constant of my toaster oven is abysmal, so the process is still evolving, but it's worked fairly well so far.
I've taken a few closeup pics of some of my most recent boards, both good and bad:
formatting link
It's not in a decent gallery because the software I was using didn't put up with an upgrade and I haven't figured out what's broken about it yet.
I haven't done any yet, but I plan on doing some attempts eventually. The process would be to put the toner and film on one side, drill 3+ reference holes, then transfer the other side. The problem may be what the laminator will do to the first side when going through the second time. It may be necessary to use even more extra dimensions on the PCB and transfer films and carefully tack the two pieces of transfer paper down through the drilling process, and laminate the two sides at the same time.
The through-hole plating problem is another one I need to research. Pulsar recommended going to International Eyelets (ask for a catalog, you get it within a few days) who make nothing but "PCB eyelets" designed to do through-holes. They're going to be noticably bigger than your average via, but comparable to a normal through-hole component pad. Absolute smallest one their catalog shows a 0.030" hole diameter,
0.020" finished hole, with a 0.046" flange/pad. That's actually not too bad. I'm going to get some trail stuff pretty soon and see what happens.
I used fingernail polish as etch resist for small jobs that didn't need fine work and it worked great. It can be removed with nail polish remover. I never tried etch resist pens because I didn't trust them.....
On the ground plane side, I needed to clear the areas around the holes and I used a drill bit with some tape around the shank which I rotated by hand to clear it out. The drill bit would center easily in the hole and had a diameter larger than the clearance I wanted so that I could stop when it was right.....
This works great, as long as one doesn't need tiny tracks....
Andy
PS I prefer Ferric Chloride for etching. I never was happy with Ammonium Perchloride (or whatever the other stuff was ). Keep the solution warm, and jiggle the board occasionally with the side to be etched facing down......
About a year ago I bought a Roland SX-8 (vinyl) sign-cutter for some Scout activity signs AND as stencils for glass etching (via sand blaster).
Works just ducky for both of those applications, but I'm tempted to try it as stick-on resist.
Anyone tried such a scheme?
...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| E-mail Address at Website Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food.
I am not a hobbiest, but I do lots of small (prototype) runs of PCB's.
I was using Olimex for a while, but they are no longer on my list of approved vendors. I got 3 runs of different boards in a row that had hair line breaks in traces. When I went to them for some resolution (to a problem of their making), their solution was for me to ship the bad boards back to them, and they would ship me some new boards... with me paying international shipping all 3 ways, and them waiting for the bad boards to arrive before starting the run of replacement boards. Since I use courier shipping to speed things up a bit, that would be $180 in shipping costs, and 3 weeks worth of time.
Other problems with Olimex are: their nonstandard tool rack (you have to figure drill sizes to be the preplating size); wide variability of hole plating thickness; holes that were clogged with copper plated fiberglass swarf; misalignment of drill holes and pads; panelized boards that are sheared, so they have rough fiberglass edges; use of the copper board outline to gage their shearing, leaving 1/2 of the copper outline as part of the board, so your boards are always oversized, and have a conductive trace around the outside edges; minimum trace size of 8 mil, (if you try to use 8 mil traces, you will find them sometimes as small as 5 mil in size); and frequent overetching of the board. Plus, they will only stock 1 oz, 1/16th inch, FR4. If you need something else, tough!
I am done with foreign suppliers! But I am still looking for a good domestic supplier. I sure wish ProtoCircuits was still in town. Their leaving started me on this odyssey of pain.
The Roland came with a program called Stika that can do outlines. I've only used it on BMP's, but maybe I can make it all play through Universal Document Converter.
Plus ExpressPCB can be cajoled into printing "solids", making outlining much easier.
I have an advantage... my sight in inadequate to do less than 50mil traces ;-)
...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| E-mail Address at Website Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food.
Yes, and no. I use an IBC-912 Boardmaker, which is a PCB mill. It engraves an outline around the traces to isolate them from the rest of the ground plane.
You would be doing much the same thing with your Roland, except you would be peeling away the parts you didn't want to etch. The problem is in your PCB cad package. They generally are based on the photoplot plan for making boards, that is that traces and pads are separate entities, and multiple exposures are still black. What will happen, is when you plot onto your Roland, first will come an outline of all of the traces, and next will come an outline of all of the pads. The pads will chop off the ends of the traces, and you will have a stencil that is nothing but tiny pieces.
What you need is a program that acts like a photoplotter, and puts the ultimate silouette image into memory, and then creates the vectors that are necessary to make the true outline. There is a program like that available on MIT's website for the "how to make almost anything" course that the university offers. It is even targeted to the Roland plotter. The author is a real abrasive peacenik character, but his software is top notch.
Alternatively, Christian Zinner has a program called PlatinCNC which does the same job (only much much better). His program is geared towards the CNC routed PCB's, like my IBC makes. I have been using it for quite awhile now, and it is a complete and total marvel! He has it priced for under $200 last I looked.
I still make some myself the classic photographic way, 100% success, takes me about 15 minutes with fresh, well heated etching liquid (the dirty yellow one). There are many good tutorials to be found on the web. The hardest part, I find, is cutting the board and drilling the holes in the right place.
When I was a kid I used to use an etch-resistant ink pen (and a nail and hamer for the holes, in fact I did the holes first) but fine-pitched boards are not easy this way. I remember routing traces in between ic pins (I then had a small electric drill). The pen was rather annoying since you had to press the tip to get the ink out and often this resulted is large blobs on the board.
Generally speaking, SMT parts are actually faster to use than leaded parts. Where I've worked in the past, the production workers preferred the SMT boards!
Those PCB pictures the guy posted are nice... kinda reminiscent of what milled boards look like!
Good point. I tracked down the hazardous material depot in my county. I am going to visit them some time next week to "unload" some used ferric chloride
Once I know how much it costs and how often I can bring it to them, Ill get serious about manufacturing test boards I design at home.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.