PCBs without leaving the lab

For photolith

PCBs without leaving the lab

Postby Doug Coulter » Sun Jul 18, 2010 4:09 pm

I use the very old and dead-simple Protel tools, Traxedit or Easytrax for layouts when possible. Though there are some other free or shareware programs out there that will do, I just like these because of how dumb they are. The newer Protel tools (very expensive!) for example, won't place a pad for you unless it is in some complex component library you basically have to get from them, and are so complex that unless all you do is lay out boards all day every day, you'll never get facile with them. The advantage of the dumb version is that for it, a 16 pin dip is a 16 pin dip -- it doesn't have some obsessive schematic capture insisting that you tell it the part number, than asking for money sent to the mothership-corporation to allow you to place the darn thing. The old stuff will eat or put out netlists, however, which is useful. If you type in a netlist, you can then have it show the connections as straight lines between the components after you place them, then re-place and rotate them to make the fewest lines cross before placing the real tracks. Handy. Easytrax and Traxedit are dos-only but do run fine in any dos emulator -- like dosemu in linux. They have a very mnemonic set of commands, and key shortcuts galore (all the function keys plus either shift, alt, or ctrl) but what I do is just use the first letters of the commands in the menu structure. So if the command is going to be "file, save, yes" to confirm the save, the shortcut is fsy. I printed a cheat sheet to lay over the function keys, but that's too much fishing to find the magic key when "pc" means "place->component" and so on.

I will try to find a way to get a copy of the tools I use up here so you can download it and use it, if I get any requests. Both will go straight to a laser or inkget printer transparency, with some nice options, so you can use the readily obtainable positive photoresist pre-sensitized boards from Marlin P. Jones and others. They will, with some hassle, also make gerber plots for sending off to APCircuits or other board houses. We find that the transparencies are a little imperfect in how black the black is (and for the UV types of boards, red works better), so the exposure time has to be found for your system experimentally for best results. We do that here by using a strip of aluminum foil on a strip of scrap board, initially covering all but a little bit of the board, then pulled back one increment about every 30 seconds or so while under the lamp -- the lowest exposure that develops easily is the one to use for best results. This depends a lot on your light source, so I can't give a number for other than my setup, for which it's about 4 minutes under a pretty bright 250w mercury vapor lamp (streetlamp with outer phosphor envelope broken off so the UV can come out). The developer for these is simply a dilute lye solution -- they sell a concentrated, fairly impure version that you dilute, but you can make a better performing one at home with plain old lye (make it saturated) and then diluting it as recommended in the board instructions. You can make it a little stronger for under exposed boards too, but if you overexpose them, you'd better watch closely while developing and get them out of the developer before you lose all your tracks or get pinholes. Sometimes you can fix a gap with a sharpie (the stinky ones resist most etchants), but that's pushing your luck a bit, since the solvent also dissolves the resist that was there, and no way it looks nice once done. Note that lye solution will become sodium carbonate solution after some air exposure...so keep it bottled up and don't mix a lot more than you'll soon be needing.

The way to see if you've developed your board correctly is to use some Ferric Chloride (ugh) on a paper towel, as it will instantly corrode and change color the exposed copper parts. For etching, you can use that messy stuff (permanent stains on all it touches -- even pyrex) or some other, tamer etchant. To tell the truth, there's not much that will do this job that you want to have get on you -- ammonioum persulphate works at boiling temperatures with a mercury activator -- also ugh, but at least you can see through it while etching. We keep this kind of thing in old 2 liter pop bottles between uses.
Note that if you buy solid ferric chloride someplace and do your own mixing -- first, do it soon after you get the stuff. The second is that this stuff is one of the more hygroscopic things on the planet, and exotherms when mixed -- you have to add it really slowly to the water or it will boil explosively. It may be even worse than diluting pure sulfuric acid! Took me an hour to make a gallon, and starting with cold water, it was on the threshold of boiling the entire time as I added it a spoonful every couple minutes.

A nice trick, if you're going to drill your boards with a dremel drill-press (recommended) is to tell Trax to make guide holes in all the pads of about 15-20 mils. These act like centerpunches to guide the drill bit in, using the slop in the drill press to get that happening. Makes things go lots faster.

So, a step by step here:

Lay out the board. During layout, especially if you're working with some new parts (things like connectors are all wildly different and usually you have to get out calipers and try to make the right footprint), you print some 1::1 check plots to make sure your parts fit on the paper representation. That will save you some grief for sure. Also, you can check things like 90 degree connectors to make sure you got the pinout right with how they fold the wiring inside -- backwards connectors are all too easy to wind up with. Also, put some text on every board layer so you can later be sure you put the artwork on right side up during exposure -- I don't want to admit how many times I've made boards the wrong way when I didn't do this!

Make your artwork. I like laser printer art a little better than what I get from an inkjet, but it's truly a question of one marginal thing or another. I've known people to try and make two copies and tape them together in registration to get blacker blacks, for example, but I never do that, just do the exposure just so -- easier for me. I wrote a little perl/tk script that has a timer with the right number as the default, that shows a shrinking pie chart and beeps when timing out -- makes it easy.

Expose your board. If doing double-sided, you can do just one side, then drill a couple of holes to get good registration for the other side, or you can pre register your artwork by taping both pieces to a board scrap along one edge, then insert the board, and expose both sides before developing. That trick avoids slippage -- if you just tape them together with no spacer, they won't line up with a board thickness between them -- and you'll not be able to notice until too late. Here, I use an exposure frame I made from two pieces of glass, one glued to an iron angle, the other hinged to that, so when closed they are both flat, as the hinge is setup to have the 1/16" spacing between them. I clamp these together with cheap plastic spring clamps for exposure so nothing moves.

Develop your board - mix the developer according to directions for the first try -- you may want to adjust that once you get some experience. I put them in a flat tray, and rock the tray for agitation that makes it go faster and more uniformly, but also have been known to just rub on trouble spots with a fingertip or brush -- the manufacture isn't all that uniform, but what saves you here is that the photo sensitive layer is thick, and "linear". In other words, light exposure merely affects the rate of dissolution of the resist in the devloper -- even a non exposed board will eventually lose all the resist left in there long enough.
So, you can correct a little for exposure in the developer (none of this will be news to any film developer, it's about the same set of issues). I check development with Ferric Chloride etchant, it immediately shows any little bad spots. For the worst of these (as in, sometimes there's a thick spot or a bit of dirt in the resist) I'll just whip out an exacto knife under the microscope/magnifier and fix that.

Once you have that all working, you etch the board. Warm etchant works a lot faster than room temperature, and I've been known to preheat it in the microwave oven when I get impatient. Agitation is very important here. The low tech way is just to have a hook of baling wire, and lift and drop the board in the etchant tray, but the resist is very fragile, so you have to be careful not to scratch it. I've also built tall, wide, but thin tanks out of plexiglass and put an airstone (from the fish store) in there - but with ferric chloride, 'ware the fizz -- that stuff ruins almost everything it comes in contact with. In general, unless I've overexposed or overdeveloped the board a good bit, I'll leave it in the etchant for a few minutes after it looks done -- a thin film of Cu you can't see easily will really mess things up in some high impedance circuits later on. As I'm not doing plated through holes, I just make vias and then solder a bit of WW wire through them, or solder some components on both sides. Inelegant, but it works!

What I have found here is that this is by far the quickest way to get some medium complex thing going on and tested. A PCB is one heck of a lot easier to modify if you have some minor mistake than a perf board is, and you're not really going to put into service something on one of those plug in kludge boards. But if I need more than about 2-3 boards (unless they are tiny and surface mount with few holes) I go out to APCircuits for more -- they do such a nice job, cheap and fast, that I mainly do this for one-offs. Nowadays, some parts are only available in SMD with some oddball footprint, and you really can't practically prototype them with dead bug techniques at all. So for some things, you begin with a PCB, not end with it.

You can choose whether to leave the resist on there or not when done. It's easy to solder right through if left on there, but will come off easily too with rosin cleaner, acetone, or a lot of other things. Not really a solder mask material at all. If you're going to strip it -- do it before you drill the holes, as the little burrs will catch your solvent soaked paper towel and make a mess. If I'm not going to build the board right away, I leave it on to keep the copper underneath very clean and ready to solder.

I have tried the electroless tin plating solutions. The word is -- avoid. They don't work well and more often than not make soldering harder -- the opposite of the claim for them. If you need fancy boards, see the APCircuits link. I have found nothing that beats Glyptal from Caswell plating for coating boards that need low leakage and corona so far. If you get this and are using it in tiny amounts, put it in a canning jar, as the metal can seal will fail after a few uses. As this stuff isn't cheap, it matters a lot to keep your stock from getting hard on you.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: PCBs without leaving the lab

Postby Doug Coulter » Fri Sep 24, 2010 2:04 pm

Ok, since I've just done this this morning, here's a step by step with pictures. I realize most of the members here don't need this level of detailed instruction, but there are those lurkers too, and this is worth putting out there -- it makes life so much easier.

Due to our limit of 7 pix/attachments per post (10 mb total max), this is going to take a couple posts to get them all in.

I often find that even if I'm not completely sure of a circuit, one or two wires or cut tracks on a PCB - even having to add a component to the back side, still amounts to less work than a perfboard, and a lot more maintainable later, even for one-offs as described here. The board I made is shown on the computer screen here. Now we'll take it the rest of the way. I printed it out on my laser printer, using paper at first to solve the inevitable margin and orientation problems before sticking in that expensive film. I generated the print file with holes in the pads, as they make great guides for a sloppy dremel drill press later.

We are working with fairly new positive resist pre-sensitized board, available from a bunch of sources. The newer stuff says you can do it with a regular 100w bulb, but I use a rig I built by getting a 250W street-lamp kit (mercury vapor) and breaking off the outer envelope to get rid of the phosphors and UV block.

To just make sure everything is scaled right, I test the components against a paper printout, and make sure I cut the board to size correctly, using the real art and the print frame I built.
Mockup_printframe.jpg
Printframe

A good contact print frame is essential. This one is made from thin glass (so it doesn't stop much UV) from the hardware store, and a piano hinge and silicone goop. The trick is setting it so when a 1/16" thick board is in there, the top is flat -- so I made the hinge height adjustable. You can see the board and art here, with the light protective film still on the board so you can take your time and fix anything that needs it at this point. Now it's time to fire up that exposure setup and let it warm up. Merc vapor lamps change output in both amount and spectrum for the first few minutes, so I let that happen while I put the real printing setup together. you want to do that well away from sunshine -- dim room light at most, as this stuff is hyper sensitive to sunlight (about 30 seconds exposes it on a bright day). So right after I have it all setup in there, I put something like a magazine over the print frame so I don't get overexposed from room light alone.
Exposure_setup.jpg

I used my general purpose environment chamber for this. You don't want all that short UV getting into your eyes much, so the door is nice, and it stays a little cleaner in there that way. This is just a big box I built that I can heat or cool the insides of, or make it real humid in there for rust blueing -- or just hanging freshly finished knife handles so dust doesn't get in the finish.
So, while that's warming up, we get the real exposure set up.
Mockup_realboard.jpg

The blue-green is the photo sensitive resist.

Exposure time would be pretty non critical if everything else were right, but that's almost never the case. For this source I use 4.2 minutes in that rig. I wrote a little perl/Tk script for this computer that runs a set time and beeps me, like an egg timer. The photoresist is kind of proportional, so you can fudge things a bit in development by varying time, strength of developer, and so on. The developer for these is a 10% lye solution. Even if you think you've got an airtight jar, it will eventually turn to mostly sodium carbonate as it eats CO2 from the air, and that happened to me this run. The carbonate doesn't hurt it, it just doesn't develop, so I added more lye, and overdid it somewhat, causing problems you'll see later on.
Developing.jpg

As you can see, where it was exposed, the blue washes off faster than it does where it wasn't exposed -- as much. No laser artwork has black as perfectly black for one thing, and even a fresh board will develop if left in there long enough. I made a bad spot on this board you can't see yet when I poured in fresh lye right on it while in the "tank" which is a hambuger tray from the grocery store. So, don't do that. FWIW, that expensive developer they sell is just a saturated lye solution, not real pure lye even, that you dilute 10::1. It's a lot cheaper to just get a jug of lye and do that yourself...
Now we test if we developed it well enough, using etchant. You must wash the board off completely first, as the two chemicals don't like one another one bit, and make a precipitate, I use a windex jug with distilled water and a paper towel (but be gentle, the resist film is real fragile and easy to scratch).
Testing_develop.jpg

By just wiping a bit of Ferric Chloride etchant on there with a paper towel, you can see if there are any flaws, and now is the easiest time to fix them. There can be resist so thin you can't normally see it that will still hinder even etching, and this shows it up really well. Most other etchants don't do this so well, and if you don't like Ferric Chloride (because you have some brains and are tired of having it permanently stain all it touches), you'd still keep a little for this step.
As is normal, there were a couple of flaws, which you fix now. Either scrape a bit with an Xacto knife, or add a little dot of sharpie, depending on which way the fix needs to be.
Flaw.jpg

Can you see the flaw here? Actually there were several, which I fixed. There was also some eat through (that dumb move of pouring in a little pure lye right on the board), which I didn't fix because a: I didn't see it, and B: for this I don't care much, it's different with smd and 8 mil tracks, here they are fat more for corona reasons than any other, and I plan to solder coat them anyway.

Getting near the limit on pix, so I'll continue this here.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: PCBs without leaving the lab

Postby Doug Coulter » Fri Sep 24, 2010 2:52 pm

Continuing on, we are going to etch the board now. I'm using cheap Ferric Chloride I got in solid form from Marlin P. Jones. Ammonium persulphate etc are friendlier when you have them.
When they say add the Ferric Chloride to water slowly, they aren't kidding around. It may take up to an hour to do for a full batch without making the solution boil violently, and this is not stuff you want hot and splattering around. It's almost like adding pure sulphuric to water, extremely exothermic. It doesn't seem to go bad in storage, and I use 2 liter soda pop plastic bottles to store it in. When fresh, it's pure brown, it gets some green/brown in use which lets you know it's not going to last that much longer. Adding a spoon or two of HCl helps it last a little longer, if you can hack the fumes from that (you don't use much!).
Etch_progress.jpg

I used the same hamburger tray, washing it out well between the chemicals. I don't usually do boards this big, so I only snatched one out of the garbage for this. I wanted to show this partly etched board to point some things out. Spent etchant collects where there's a lot of exposed copper, and etching stops, even though I was rocking the tub back and forth as hard as I could without the etchant coming out. I have tried other ways, from sprayers to air agitated baths in specially made thin (but wide and tall) tanks, and they do work better if the setup hassle is worth it and you don't mind the amount of this evil stuff they put into the air. This effect is also true in developing, and a very gentle brushing with a fine brush helps that go better. You will note that the etching is fastest near the edges and near the places where the resist is due to this effect, and with thin lines you have to worry about undercutting a little bit. The etchant works better the hotter it is, but to be around it, you don't want it too hot. I heat it up to about "making tea" temperature in the microwave first.
Etched.jpg

Here it is all nicely etched. A little too much, some of the tracks got pinholes. No sweat on this one, as I'm going to coat the tracks with solder to make a radius that cuts corona down anyway.
You can solder right through that resist, but I generally clean it off with acetone just before soldering, which gives a nice pristine surface (at least until you touch it with your hands) that solders easiest.
Now it's time to drill the board, with the setup below. This works better than it looks, especially since I left guide holes in the pad artwork to guide the drill tip.
Drill_setup.jpg

Note I have the quill lever set so it's about horizontal, and in use I grab it between two fingers and it takes almost no motion or effort for each hole. I use light, light, more light, and magnifying goggles and watch for parallax while the drill is real close to where I think I want the hole. With a little practice, this goes pretty fast, but of course if you had one of Jerry's CNC machines, you'd do it with that instead.
Drill_detail.jpg

Obviously, you want to use a drill a little bigger than the component leads here. I mic'd them at 31 mils, and the next larger size solid carbide drill I had in the box I didn't have to search for was 46 mils -- fine for this. If the collet on your Dremel tends to grab like mine, don't put the drill in full depth -- the collet frees easy when you're done if you loosen the nut and then push the drill in -- then pull it out. There has to be some "in" left for the trick to work. Having to pull hard "out" means you break these drills too easily.
Once you've drilled the board, make sure the parts fit. Here it is when them upside down to show that. What I actually did was drill for just one first, in case I was going to have to fudge things because I didn't get the art just so. It happens.
Drilled_FitTest.jpg

Now for the easy and fun parts, stuffing and soldering.
Stuffed.jpg

No need to use the fancy Metcal on this job, the old standby Weller station does fine, and has variable temperature, which I am going to take advantage of here.
I am using real good solder for the pads, and then I'm going to go back over it all with plumbing solder at 100f higher tip temperature -- it's cheaper and it blobs better.
Here's the result.
Soldered.jpg


Now, since I'll be pushing the creep insulation strength of the pcb with 20kv peak between the cap stacks, I'll paint it with glyptal before use. Wait for it:

Here's the family Joules ;)
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: PCBs without leaving the lab

Postby Doug Coulter » Fri Sep 24, 2010 7:41 pm

I plan to put in a few tricks here for getting double sided with littler more difficulty than this. It's not that hard once you get this part hammered.
And there's more than one way that works.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA


Return to PCBs and photolithography

Who is online

Users browsing this forum: No registered users and 2 guests

cron