Toner transfer PCB etching in a (bio)physics laboratory

| tagged with
  • pcb
  • etch
  • toner transfer

Toner transfer PCB etching in a (bio)physics laboratory

While with the recent advent of batch PCB services one-off PCB etching has become remarkable inexpensive. That being said, sometimes you need the board yesterday. Also PCB etching can be quite convenient even for simple circuits, where point-to-point prototyping would be possible but labor intensive. Moreover, the use of even large (1206) surface mount parts allow the board to be much smaller than possible with other prototyping methods.

Thankfully, in a laboratory setting one is blessed with a variety of tools and materials that most homes do not have.

One unfortunate limitation of the toner transfer method is the difficulty of producing double-sided boards. While it is in principle possible

Overview

The toner-transfer method exploits the inability of laser printers to bind toner to glossy paper (e.g. photo paper). I have had good luck with all of the photo papers I have tried and have even used magazine paper before (although its light weight made handling and printing tricky). The procedure goes roughly as follows,

  1. Print a mask on glossy paper
  2. Transfer mask to board
  3. Etch
  4. Post-processing

Board material (known as copper-clad board) consists of a sheet of substrate material (usually FR-4 (fiberglass and epoxy) or phenolic plastic) plated on either one or both sides with a thin layer of copper. The board I generally use is available from RadioShack, although this is something that can also be found for quite cheap on eBay. While many of the items on eBay have surface blemishes, this largely doesn’t matter since we will be conducting an extensive surface preparation process anyways.

Producing the Mask

I use the wonderful gEDA EDA package and I would strongly recommend it to anyone looking into a designing a board. While its learning curve is a little steep and integration between tools is sometimes a bit weak, to anyone used to working in a UNIX environment it should be easy to pick up and unmatched in its flexibility.

When using producing a board, I like to give myself wide tolerances to make the process more forgiving. I prefer to use 12 mil clearances on all traces and parts. This can be set by selected all parts on the layout (Alt-A) and typing :MinClearGap(Selected, =12, mil).

Recall when laying down your PCB design that through-hole components should be on the opposite side from your traces.

To produce a mask in PCB, one first exports the board to a Postscript file (File -> Export Layout, select PS, default settings are fine). This will produce a document with each of the layers nice formatted with alignment marks.

One thing which has tricked me numerous times in the past has been mirroring of the image. For this reason, I will always print a “dry-run” copy of the front mask along with the assembly diagram on standard paper. Laminating these together with print facing out should look just like the final board. I verify that parts’ chiralities are correct and that vias line up with this test “board” before printing my final transfer mask.

Surface preparation

Before the mask can be transferred, the surface of the PCB material must be cleaned and prepared. One unfortunate property of copper is that it oxidizes extremely quickly. This layer of copper oxide will not be etched by the etchant and will be unsolderable. Also, for obvious reasons, the board must be free of grease and other dirt before the mask transfer.

  1. Roughen the surface slightly with medium abrasive such as fine grid sandpaper or medium steelwool.

  2. Drop the board in an organic solvent (e.g. acetone, methanol, or even ethanol; purity doesn’t matter) for 5 minutes or so. This will remove any residual grease.

  3. Rinse in water

  4. Drop the board in a acid bath for 10 minutes. I usually use a HCl solution with a pH of about 3, although this shouldn’t be picky. This will remove any copper oxide remaining on the surface.

Printing and transferring the mask

Print the final transfer mask on glossy paper on a laser printer and cut it down to the size of the board. If possible, increase the toner density of the print (on my printer, an HP LaserJet P2015dn, this involved setting the “Image Quality” setting to “Highest”). I generally wait to do this until I have everything prepared for the transfer.

For the transfer, I use a piece of optical breadboard on a Fisher Scientific heat plate set at 200 Celcius. This ensures I have a nice flat surface with large thermal mass. On top of this I place my board and printed mask (with mask on top facing the board). On this I place another sheet of flat aluminum with a heavy catalog for weight. After allowing the board and mask to heat up for five minutes or so, I’ll use a cylindrical roller (e.g. New Focus optical rod) to press the mask on. I’ll then wait another five minutes and repeat. Ensure that every point on the mask has been compressed.

After pressing and waiting several times (usually this takes about 20 minutes), I’ll pull the board off the heater and put it in a bath of warm water, where it soaks for at least 10 minutes. This makes it much easier to remove the paper without pulling off the mask. You should be able to remove the first layers of paper by rubbing with your fingers. After the first layer is remove, allow the board to soak for another 10 minutes before removing the remaining paper.
Paper still remaining between small features (e.g. traces) can be removed with a pointed object but try not to damage the copper.

This part of the procedure is quite sensitive and it’s worth spending time to avoid removing any more of the mask than necessary. That being said, if part of the mask does come off with the paper, this can be easily corrected with a black Sharpie (other colors haven’t been tested). Make sure the area is thoroughly marked with the Sharpie as experience has shown that Sharpie ink doesn’t resist the etchant as well as toner.

Etch

Copper can be etched with a variety of liquid etchants. I use a ferric chloride solution from Radioshack, although a little Googling should demonstrate that any laboratory with some basic chemicals should have what is necessary to produce an etchant solution. If I had not already started using my ferric chloride, I would probably use a copper chloride in HCl solution.

When using ferric chloride be careful to avoid touching the solution with anything you don’t want stained. Ferric chloride can be

After allowing the board to dry, throw the board in a bath of etchant and agitate. I use a rocking platform this which works quite nicely. Keep a careful eye on the etching process and flip over the board a few times while etching (the top usually etches more quickly than the bottom). I find that a good etch usually takes around 10 minutes, although this will heavily depend upon the strength of your etchant.

Post-processing

Removing the toner from the etched board can be accomplished by a combination of solvents (e.g. acetone) and mechanical abrasion (e.g. a Scotch-Brite pad).

To drill you can use any suitably sized drill bits (you’ll almost certainly need more than one for usual via-sized holes) and a drill-press. The board will have a tendency to “jump” after drilling through the bottom of the substrate, so be sure to hold it securely. I use a set of #63 (0.0370“) PCB drills purchased on eBay, although it would be nice if they were a tad smaller. Drill with the traces facing upwards and ensure the bottom of the board is supported with a piece of wood, otherwise the board substrate has a nasty tendency to crack. It never hurts to drill more slowly.

Finally, I like to tin all of the traces of the board right after etching to ensure they don’t oxidize. This makes soldering to the board much easier after it has sat for a while. Also, especially with large holes, getting through-hole parts to wet to the board is sometimes difficult. This is made much easier by pre-wetting the pins of any through-hole parts before soldering.