Introduction: Killer PCBs

This instructable demonstrates the process for making printed circuit boards with features as small as 0.005suitable for LQFP or QFN ICs using negative dry film photoresist. This will enable you to handle just about any kind of integrated circuit available--even ball grid array! Pictured are boards with a TSSOP-14, QFN-40 packages using a .65mm pitch and zero insertion force flex sockets with .5mm pitch.

Step 1: Background

After experimenting with home PCB fabbing for a while, I've finally worked out a process that produces reasonably consistent results that actually look pretty good. I spent lots of time trying to use the toner transfer method with varying degrees of success (OK, varying degrees of failure might be more accurate). I also tried Philmore/Datak negative photo resist spray with consistently horrible results (the stuff eventually melted the spray nozzle that came with it and leaked all over the place). Not Green & not recommended. Now I could have purchased presensitized boards and saved a lot of trouble, but I find the material to be too costly for the volume of boards I'm producing. I eventually tried dry film photo resist and I won't be going back! I'm not going to go into the intricacies of schematic capture, or etching since those subjects are well covered by other instructables. No volatile compounds are used--only simple bases which can be rendered environmentally safe by filtering solids and neutralizing with HCl (see manufacturer instructions for proper disposal procedures). This process, when combined with a Peroxide/Cupric Chloride etching process forms an environmentally responsible, Green PCB development process.

If you haven't tried the toner transfer method, do so. Unless you are blessed with magical toner and/or paper, the dry film resist method will yield better results, but the process is a bit more involved. If you are satisfied with the toner-transfer results, by all means, stick with that method. Naturally the standard warnings apply: PCB etching and dry film processing involve caustic materials--be sure to use protective equipment and have an eye-wash station handy (or at least a bucket of water). Also note that dry film developing and stripping involve strong bases--keep them far away from your etching chemicals, or they may react violently.

Thus far, I've used three types of dry film resist, all of which performed well:
--MG Chemicals 416DFR Dry Film Resist About $20.00 for 12" by 5 feet at Frys, Altex and online. MG refused to quote larger quantities, and will not divulge the manufacturer of their film.

--Dupont Riston M115 available at Think & Tinker Excellent resist, much more economical than MG if you want larger quantities (12"x50ft for $96.75, 12"x100ft for $116.26). Outstanding outfit, very helpful, friendly people and lots of great info. Terrific site!

--Kolon Dry Film Resist Korean manufacturer sells for somewhat less than Think & Tinker's Riston, but with a minimum of 500ft cases.

What you will need
- Laser Printer
- Home/Office Laminator
- Laser Printer Transparencies
- Spray Adhesive
- Negative Dry Film Photo Resist
- Resist Developer (sodium carbonate)
- Resist Stripper (sodium hydroxide)
- Glass Sheets
- Clear Tape
- Yellow Bug Light
- Light-Safe Area

- Vacuum Bag or Vacuum Frame
- Collimated UV Exposure Source
- Rotary paper trimmer
- 21 step Stouffer Sensitivity Guide for Calibration

Step 2: Create Artwork

I'm assuming you've captured your schematic and laid out your PCB in something like Eagle--if you don't know what I'm talking about, you'd better start off by learning schematic capture, and PCB layout. Once you've got the PCB laid out, you will need to create negative photo masks. A negative photo mask is a transparency that is clear where you want copper and black where you want to remove copper. One hint: if you make a habit of filling your PCB layers with ground or power planes, you will save a lot of toner. Also, note that you can tile your PCB layout in Eagle using the group copy command if there is not an associated schematic file (otherwise it will complain you have to do this in the schematic"). I found the process of inverting the colors in your output to be surprisingly difficult, but eventually I settled on this fairly simple approach using the open source vector drawing software Inkscape:

1. Make sure you've selected just the layers necessary for etching (e.g. Top/Bottom, Pads, Vias).

2. Use Eagle's print function to output to a PDF file. Even though we're printing on transparencies, you'll still want to mirror the top layer so as to place the printed side closest to the board and prevent bleeding.

3. If you don't already have Inkscape installed, download it and install it now.

4. Open the PDF in Inkscape (accept the default loading parameters).

5. Open the Layers pane (Menu Layer:Layers).

6. Click the + button to add a layer. Name it whatever you like (I named mine "b").

7. Click the down arrow to move the new layer to the bottom.

8. Select the rectangle tool from the tool pane.

9. Right-click the blue color swatch at the bottom of the screen and select Set stroke.

10. Right-click the grey color swatch and select Set fill.

11. Click and drag to draw an rectangle around your board.

12. Select the arrow tool, then select and drag each side of your rectangle right to the edge of your board. The display should look something like this:

13. Right click the black swatch at the bottom of the screen and select Set Fill. Right click the black swatch again and select Set stroke.

14. Click the eye icon in the layers palette to turn off the b layer.

15. Click and drag to select your board (or click the select all visible icon on the toolbar).

16. Negate the image (Menu Effects:Color:Negative)

17. Click the eye icon for the b layer again. You should now have something like this:

18. Click Save-As, then change the file type to PDF via Cairo, and append _out to the filename (e.g. test_out.pdf) so as not to overwrite the original.

19. Click Save-As again to save as the default SVG format.

Step 3: Make Transparencies

This part was a bit of a challenge too. The problem is that most laser printers don't make really dark prints, so some light leaks through the black areas. I tried lots of approaches to selectively darken the toner: dry-erase markers, stamp-pad ink, crayons, charcoal, graphite, softening/heating + additional toner application, etc.. None of it worked. Here's what works consistently for me:

1. Start with small artwork--I use a rotary paper trimmer to cut transparency sheets into quarters (4.25 x 5.5) 2-3 sheets at a time. Small artwork is better because heat-related distortions in the transparency material will be reduced.
2. Open the PDF from the last step with Adobe Reader, and print to a laser printer loaded with your transparency sheets. For my printer (Brother HL-5250DN) I use the following settings: User defined paper size (4.25 x 5.5), no duplex, manual feed, 1200 dpi, Darkest Density. Big hint here: You can have multiple copies of the same printer installed in the Windows, so add a new printer called PCB_Laser as a duplicate of your existing laser printer, then right-click and modify the defaults as needed for PCB transparency printing.
3. Print top and bottom artwork. Hold the artwork up to the light: Do you see any light seeping through the black areas? If your results are similar to mine, you will have enough seepage to cause problems with your resist. Note that you really need to have some larger black areas to accurately judge toner density. If you are blessed with super dense toner, then skip to step 8, otherwise, go ahead and print a second copy of each of your transparencies.
4. Next you need a makeshift light table to align the artwork. This can be as simple as a piece of paper taped on a sunlit window, or a shallow tray containing a hockey-puck-sized light covered by a sheet of paper and a pane of glass. The backlight from a scanner makes a great light table. Just run a scan in transparency/negative mode--chances are it will leave the backlight on for several minutes after scanning then remove the lid and flip it over. The use of a head-mounted magnifying lens will help significantly in aligning your artwork.
5. Next we will bond the transparency pairs (2x top, 2x bottom) to double the toner density.  To do this, take one copy of each transparency and apply spray adhesive to the toner side. You can tell which the toner side is by observing the reflection of light off the surface of the transparency; the toner appears dull on the toner side.
6. Place the non-adhesive-coated transparency toner-side down on the light table--you may want to loosely tape the corners of this sheet to the light table (fold the tape over at the end so you can easily peal it up afterward).
7. Carefully align (register) the adhesive-coated transparency with the uncoated transparency. Once aligned, press firmly to adhere the two sheets.
8. Pass the aligned transparencies through the laminator on the coldest setting to permanently bond the layers together.
9. Align the top and bottom artwork (toner-side in) and tape securely leaving enough room to slide the PCB between.  Alternately, if you have a border that is at least an inch around the board, you can apply spray adhesive to a 1/2" strip along two edges by covering the rest of the mask with the corner of a piece of paper--just be sure to no less than 1/4" between the adhesive strip and the edge of the board.

Step 4: Prepare Copper-Clad

Carefully clean the copper-clad material with a scrubbing pad and a bleach-containing scouring compound (e.g. Soft Scrub w/Bleach). Sand with 320 or 400 grit wet/dry sandpaper to roughen the surface and ensure proper resist adhesion. Dry the board being careful not to touch the board without gloves (to prevent transferring skin oils to the board).

Step 5: Laminate

There are a few different possible approaches to laminating the PCB. MG Chemicals suggests a folder over and laminate approach. Others suggest using either heat or water (or both) to adhere the resist material to the PCB prior to lamination. I've had good luck with this approach:
1. Make your work area light-safe: Turn on the bug light and turn off any fluorescent, or >40 Watt incandescents
2. Cut laminate material 1/2 inch larger than board (double the length if you are doing a double-sided board)
3. Carefully peel just back just the first half inch of the inner film (always on the inside of the curl)
4. Carefully align the laminate ensuring that the laminate covers the board completely on both sides (if double-sided).
5. Press the first half inch of exposed laminate to the board.
6. Carefully pull the remaining inner layer downward a half-inch at a time, while simultaneously pressing the exposed laminate to the board. Be careful not to introduce any wrinkles. Continue to the back side in a similar manner if necessary.
8. Pass the board through the laminator (once the laminator is fully up to temp). Flip over and pass through the laminator again.

Step 6: Expose

Place the laminated PCB between the pre-registered transparencies and tape securely in place. Tape a Stouffer 21-step sensitivity guide over an open section of PCB if you plan to calibrate your exposure process. You want to press the artwork tightly to the PCB to prevent light from leaking under your traces. You can do this with two sheets of glass, or, preferably, you can use a vacuum bag, or vacuum frame. You can then place this assembly in bright sunlight for about 5-8 minutes per side, or use another UV source of your choosing. Note that, despite the tight fit of PCB & artwork, collimation (making light rays parallel) is important to achieve fine traces--refer here for instructions for constructing a collimated UV light source.

Step 7: Develop

Follow the manufacturer's directions for preparing developer solution (usually Sodium Carbonate or Potassium Carbonate) and developing your board (developing removes resist not exposed to UV light).

Step 8: Etch

Use your favorite etchant--just make sure you give some thought to future disposal and consider a green process such as Peroxide/Cupric Chloride etching process (it is literally green, too). And, if you want to dig into the nitty-gritty details of the chemistry involved, this page is for you.

Step 9: Strip

You'll need sodium hydroxide (NaOH) for this--this is the stuff that Brad Pitt uses to burn Edward Norton's hand in Fight Club it dissolves fat, and your skin is composed largely of fat, so be careful. Check local chemical supply houses--I lucked out: my neighbor makes soap, so I just popped over and asked if I could borrow a cup of lye!

Step 10: Solder Mask & Silk Screen

Apply solder mask by scrubbing and re-laminating etched board with a new layer of resist, then remove the protective outer film and apply a second layer of resist to double the thickness. Now expose with stop and silkscreen layers and develop as before (exposure may take slightly longer due to extra thickness of resist). Bake in toaster at 200-220 degrees F for about 10 minutes (being sure to prop up the board so that the resist won't stick to the toaster). Remove the board from the toaster and use a white crayon to fill in the silkscreen text while the board is still warm. Wipe any excess crayon wax off with a towel and use a toothpick to remove excess wax from grooves and traces.

Step 11: Solder Paste Stencil

If you've gotten this far, you must have pulled a few late nights and consumed copious quantities of carbonated (and caffeinated) beverages. If so, grab one of those soda cans and cut the top and bottom off with a pair of scissors or light shears. Try to cut it neatly without ragged edges. Next cut lengthwise to produce a strip of aluminum about 8" by 4". This strip will be very curly, but you can't "unbend" it without making wrinkles. For the following steps, you can either deal with the curl, or you can anneal the metal by heating it in a toaster oven to about 450 degrees F then let it cool slowly back down. Next you must sand both the inside and outside of the can with 220 to 320 grit wet/dry sandpaper to remove the paint (note that there is clear paint on the inside of the can--otherwise the soda would eat through it). You can use some heavy-duty paint stripper for this, but sandpaper seems a bit greener. Once you've gotten all the paint off, scrub well with SoftScrub with Bleach to remove every trace of oil. Next laminate photoresist on both sides just like you did with the PCB. Make your artwork using the tcream layer as you did before, but don't make negatives this time--the output of Eagle is already a negative! Be sure to make two transparencies each for front and back (and be sure to mirror the front so the toner will be right up against the resist). Develop as before and etch in diluted HCl. I diluted the HCl to about 50% (pour acid into water, not the other way around). Etching will start slowly until the oxide layer has been removed, then speed up significantly. Don't etch too fast, or the board will heat up and the resist will come off. If you leave the resist in place you'll have a ~5 mil stencil, or you can strip the resist to get a ~2 mil stencil--but beware, NaOH will attack the aluminum with just a bit less tenacity than the HCl (depending on concentrations). Next use your failed boards (you should have a few of these by now) to rig up a solder paste jig. Apply the solder paste with a paint scraper, or similar, place your parts, and get on with the reflowing...

Step 12: Reflow

Pretty simple stuff here: Heat until solder paste melts, then cool. Be sure to use a fairly stout copper or aluminum plate to evenly spread the heat across the board. Once all the solder melts take the metal plate off the hotplate, and place it on a heat sink to bring the temp back down quickly--cement garage floors work great--just make sure you use oven mits (but don't bake cookies with them afterwards...). You may need to clean up some bridged connections after reflowing. For this, just use solder wick and lots of flux (I like Orange Crush). Apply power and fire it up! BTW, I accidentally ran the temp up too high on the first board and got the cool gradient color effect on the solder mask as a result (see the intro pic). I think the components were still in spec, but I didn't have a good heat spreader under it at the time, so I can't be sure the probe was reading the same temp the board was seeing. The board seems OK so far though...

Well that's it--easy eh? Be sure to watch for my upcoming web site Now have fun and go save the world!
Epilog Challenge

Participated in the
Epilog Challenge