Making printed circuit boards

 

After my source for PCB making pretty much dried up last year, I decided to adventure into this route myself. 

I remember having done this in long past days, knife cutting the circuit in a thin wax layer on top of the board and using concentrated hydrochloric acid to etch, but I decided that I should have a bit more professional approach.

 

The artwork:

I am using CIRCAD98, an unsophisticated freeware program that, although not so user friendly, produces adequate results for my purposes.  I use 20mil tracks and 80mil pads.  The drawback of these dimensions is of course that one can't lay a track between a 100mil spaced rows of pins.   Going to 15mils and 60mils respectively is still doable, but it stretches the technology a bit, so beware.

Printing can be done on a good quality laser printer, preferably one with a 600dpi resolution and relatively new toner cartridge.  It can be done either directly on a polyester film (be sure it is one fit for laser printers) or first on plain paper and then photocopied on a polyester film at the copy shop in the village. 
Recently, I used an inkjet printer directly on polyester films, the ones that are specially coated for inkjet printers.  In spite of my initial fears that the "black" inkjet ink would not block the UV-rays as would the laser printer toner, it actually works perfectly, with no noticeable difference between both printing technology.

 

Transferring the image on the PCB:

I use pre-sensitised material from Bungard (a German manufacturer), which is more or less readily available in some shops here in town.  It is not too expensive and comes in single and dual layer material of several dimensions, including the ever-popular 160 x 100 mm.

For the illumination, I was quite lucky to be able to buy for almost nothing a second hand facial tanner (Philips, Home Solaria HB172) equipped with 4 CLEO 15W lamps.  It illuminates a surface of roughly 200 x 300 mm with harmful UV-B rays (as is written on the box), which is however perfect for our purpose.   I made a small rig with some scrap wood, so that the facial tanner could be positioned parallel to the PCB at a distance (to the lamps) of 120mm (not critical).  

For the actual illumination the protective film is removed from the pre sensitised PCB, the artwork is aligned on top of it (beware of the mirror image syndrome!) and the whole is covered by a sheet of 2mm thick float glass (salvaged from a picture frame).   The latter adds some weigh to the polyester film and ensures it makes contact with the PCB.

Although the illumination time is not very critical, a calibration is absolutely mandatory: the pre-sensitised material ages a bit, as do the UV lamps.  Use a strip of pre-sensitised material and cover it with a piece of cardboard, uncovering it by say 10mm every 30s during the illumination.   I came to an optimal timing of 4.5min, though 3...6min would have been OK as well.  After the illumination a faint latent image is visible (especially in fluorescent light, one does not see it well in incandescent light). 
Note that it is better to over-illuminate a bit than the contrary, as lots of problems are to be expected during etching when there is even the slightest photo resist present where it should not be.

 

Developing the image:

Very easy with a 10g per litre solution of NaOH (the dry white powder, otherwise used in the home as “sink cleaner”) at room temperature. 
The NaOH solution should be fresh, absolutely fresh, i.e. made from dry powder only minutes before using it, as it neutralises at the contact with the air.  The few problems I had when developing (slow and incomplete removal of the photo resist) were due to this.

The PCB and the solution go in a small open plastic vessel  (the ones that are used to develop photos), which is gently rocked during the process, the illuminated parts disappear in a mere 15...20s.  After that, the PCB should be thoroughly rinsed with much water and dried for at least a few hours.  If there is still a bit of the photo resist remaining (one can feel it with the fingers), one should consider an extra 1...2 min of illumination, realigning the artwork is of course critical in this case.

 

Etching:

At first, I use a 250g per litre solution of Fe₃Cl at room temperature (it takes a while to dissolve).  Once again the PCB and the solution go in the open plastic vessel, which is gently rocked during the entire etching process.  One needs just enough solution to cover the PCB by a few mm.
At room temperature the etching speed is rather slow; it takes some 30min to etch all the copper between the tracks and even longer if the solution is already partially saturated from e.g. previous etching. 
In order to accelerate the process one can heat the Fe₃Cl solution to 40…50°C, only 6min are needed to etch the copper away.  I use a second open vessel of larger format and put the vessel with the Fe₃Cl solution inside it, filling the room between both vessels with very warm water (almost boiling).  Waiting a while to let the temperature stabilises before starting the actual etching.
In my experience so far, the under-etching is very much reduced if the etching time is shortened, hence I strongly recommend the hot method for all the fine line printing.
A word of caution Fe₃Cl is reputed to be a very nasty chemical.  Beware, it stains everything (even stainless steel) and the stains are almost impossible to remove.

Recently, I used the "other" etchant: a mixture of HCl (Hydrogen Cloride, also used as "sink cleaner") and H₂O₂ (Hydrogen Peroxyde, used as desinfectant and bleaching agent).  Both chemicals are readily available in moderately concentrated solutions at the local drug store for little money.  For 250ml etchant, I use 120ml 30% solution HCl, 30ml 30% solution H₂O₂ and 100ml water.
The etching is performed at room temperature and takes only a few minutes, and if anything the limits of the resolution are even pushed further down using this etchant.

After the etching, the PCB should be thoroughly rinsed with much water and dried.  With some acetone, one can remove the remaining photo resist.

 

Some further considerations about fine line printing:

For long I was persuaded that SMD techniques were beyond the homebrew technology as described here.  Then came the opportunity to make a circuit involving a 80pin LQFP (low profile quad flat pack) with a lead width of 0.35mm and a lead pitch of 0.65mm.  Quite a challenge indeed.  Although I would not recommend using this density for an entire circuit, it is entirely feasible to have a few such circuits in a design.
I would now recommend using SMD for all projects: SOIC semiconductors (1.25mm lead pitch) and 1206 sized passive components.  Not to have to drill holes (or at least not so much) being the added benefit.

As it was my first project using SMD on such a scale, the design was split into several sub modules, so that if one part would have to be re-done, the entire circuit would not be lost.  All of the fine line printing was concentrated on a small (approx. 100mm by 50mm) two layer board, the bottom layer being the ground plane and un-etched.

I used the same CIRCAD98 artwork package as it contained the required SMD templates, care was taken not to overdo it:  using only 0.35mm lines when needed and 0.65mm lines everywhere else.   The final artwork was then printed on a 1200 dpi resolution laser printer.   The rest of the process was very much identical as the one described above.  There is absolutely no problem with the transfer of the image and the developing, as the limitations are optical and well beyond SMD sizes.  I used the hot etching process and had no problems with under-etching whatsoever, even not with the 0.35mm lines.  My feeling is that the limit of the etching technology would be lines of 0.15mm or less.

Soldering the standard SMD components by hand is a relatively simple matter providing that one is equipped with a fine tipped soldering iron (15W version with a tip of 1mm), a pair of tweezers and a good magnifying glass or better still a head band mounted pair of magnifying glasses.  
A small drop of solder on one pad, “gluing” the component therein, gently pressing the component down while heating the solder on the first pad again, and finally soldering the other pad.  It is easier and faster than mounting a through hole component, since the board need not to be flipped continuously.

Soldering the LQFP component is an entirely different matter however, one thing is for sure: it involves lots of patience.
The pads were tinned first using the soldering iron and some solder. The excess solder had to be removed, which is a dangerous operation as the pads and the track are easily torn off from the PCB.  Using a fine meshed solder wick is a good idea, but beware of applying too much force when taking the wick away.  Using a hand held vacuum pump works also providing to keep the nozzle a bit farther away.
After this operation the circuit was carefully checked for shorts and interruptions.  Then, the LQFD component was carefully positioned over the pads, as this is critical so, I took the time for this part of the operation.  I used a bit of contact glue to keep the chip in place, before the glue sets one can move the chip around a bit to have it exactly aligned with the pads.
The last operation, soldering, is fairly easy as it involves pressing with the dry (no added solder!) tip of the soldering iron on each of the leads, one by one applying a very gentle pressure until the solder under the lead melts.  I had to sharpen the tip of the soldering iron somewhat (less than 0.5mm).  Otherwise this operation went very well and it took me less than half an hour to solder all 80 leads of the chip.
Again the circuit was carefully checked for shorts and interruptions, this time with a multi-meter, a needle mounted on one of the probes.  Checking especially for open circuits between the lead and the pad (failed soldering) and for shorts between adjacent leads (solder bridges). 
Whether I had luck or not, I don’t care to know, but I had none of the latter and only one of the former on the whole 80 leads.  It took me nevertheless the next full hour to repair the circuit, and eventually I had to solder a fine wire (one strand of a braid) from the lead to somewhere further in the circuit as the pad was torn away.