In this research I have tried to find the limits of a toner transfer method, find out the minimum width of spaces (gaps) and tracks, and choose the best paper for toner transfer. The practical goal of this research is to make a PCB with track as small as 0.1mm. Some people on the Net claim that they have done it and can repeat it easily.
All photos are really huge and from 300KB to 4MB in size. It is done so all small details can be seen. Photos are inserted as links.
I will not get into the theory of toner transfer. I will describer how and with what I do it at home.
I use Sprint Layout 5 software to make the PCB layout, old HP LajerJet 1200 (600dpi max) printer with refilled cartridge with unknown cheep toner, 1.5mm one sided PCB laminate, small travel iron. I etch with Ferric Chloride. I clean the toner with very strong solvent (called "solvent 646" here in Russia). Solvent does not really matter. You can use acetone.
Before I transfer the layout to PCB I sand the copper with very find sand paper (1000), wash it with water and the clean it with the solvent.
Here are the criteria to choose the right paper:
- good toner coverage and good toner binding to paper
- good transfer to copper from paper
- easy paper detachment from copper after toner transfer
- toner coverage (melting with) by some kind of pore closing layer
The last one needs to be explained. As you know Toner Transfer exists only because laser printer and any laser printer uses toner to print. However, any toner which can be used in TT method has a big disadvantage: pores. Because of the pores etcher can get into the copper areas seemingly covered by the toner. This results in pitting of copper filled areas and undercutting of tracks. Photopaper theoretically can be a solution to the problem due to its "photo layer". This is not a real photo layer like in good old photo paper. It is a thin plastic like layer which absorbs ink quickly. During toner transfer this layer melts and covers (or mixes with) the toner. This plastic covering cannot be etched and strongly protects the copper parts during the etching process. However, according to the experience of others, this layer frequently attached itself to the PCB very strongly and hard to get rid where needed. Because of that small spaces become very hard (or impossible) to make clean. There is a commercial product for sealing the pores of toner called GreenTRF (basically foil covered with something). Some people have reported great success with it, like track as thin as 0.15mm. However, GreenTRF must be ordered over internet and not cheap.
At first I tried usual office paper (80g/m2). It sucks. It does not release toner.
Then I used paper from glossy magazines. I can say it works. However, you must stay your PCB in the water for quite long time; toner is not fully transferred on to the copper, toner flows too much under the iron.
Then I tried local Russian glossy (240g/m2) and mat paper (170g/m2). It was utter failure. The mat paper does not release toner at all. The glossy one sticks to the PCB so strong that it is impossible to detach it without damaging the PCB layout picture.
Then I tried glossy HP Premium Photo Paper 240 g/m2. Expensive. The print quality is amazing, but the paper stick to the PCB too much too. Cannot detach without damaging the layout image.
HP mat paper for ink printers does not provide protective layer on top of the toner. Toner is not transferred completely at some places. No good.
Semi-glossy HP photopaper. Even after 30 minutes in warm water I could not detach the paper from PCB. Eventually managed it, but all toner stayed on the paper.
Thermotranfser paper for ink printers. I did not print on it, but wanted to use it after toner transfer to seal the pores. No good. Did not work at all.
At this point I felt really disappointed. I saw no exit in this situation. I could think only about two options: buy more different papers (more money and time waste) or get myself a new couple of hand.
However, here what Thomas Edison said:
"Many of life's failures are people who did not realize how close they were to success when they gave up. "
He was so right!
After all the failures I went to read about GreenTRF, commercial toner transfer systems, and how they work. When I was reading about press'n'peel and noticed that they say that the temperature must not be high (like for polyester). And I though, maybe I used too high temperature and overheated everything?
I set my iron between 1 and 2 dots and used the mentioned HP Premium paper, which were dead sticking to copper. I heated for 3 minutes, went to the bathroom, turn the cold tap, poor the cold water for a second or two and pull the paper. It detached without any resistance! 90% of PCB layout image was on the copper. The toner was VERY black and there were almost not toner on the paper.
After 3 more tries I figured that the best temperature is one small "click" higher that 2 dots (my iron has many small "clicks" positions between dots). The back side of the paper starts to melt at this point, so, I put one layer of usual office paper between it and the iron. I have done all tests on a very small PCB (like 5x5 cm) and there were no need to move the iron. I pressed the iron a bit and sometimes moved the iron with average press all over the perimeter of the PCB. I did it for 3.5-4 minutes.
After that I take the paper, go to the bathroom, poor cold water on the pcb for 3 seconds and pull the paper with one move. Here is the result:
The paper becomes bluish where the tones was. The transferred toner of the pcb feels like plastic. It is shiny and glossy. No photo layer notices in the spaces.
Here is how I measure the sizes: digital camera with 8Mp resolution and a ruler. I put a ruler on top of the pcb and make a very close macro photo. That I count number of pixels in 1 mm, count number of pixels when I want to measure the distance and divide. This way I get the size in mm.
For those who doubt this method I made some tests. I have a chip from TI psp54310pwp. According to its datasheet the pin should be 0.19mm and space between pins must be 0.65. I measured it with my method and got 0.20mm and 0.62mm. So, as you see, the method works.
Testing tracks width and gaps between tracks
I have prepared a hard ferric chloride solution and etched for 15 minutes. Here the result:
Let's clean the toner:
Looks very good. All the tracks are intact (even 0.05mm), 0.1mm gap did not work out at all. 0.2mm gap seems to okay.
Multimeter test showed that all tracks are conductive and electrically intact. Gap test showed that 0.2mm gap is conductive! Here is the place where there is a connection bridge:
(outlined in red)
This was fixed with one easy move with a needle.
But what are the REAL sizes?
As before, I have put a ruler and made macrophoto.
According to the photo: 1 mm is 30 pixels.
- 0.05m 5px 0.17mm
- 0.10mm 7px 0.23mm
- 0.15mm 9px 0.30mm
- 0.20mm 10px 0.33mm
This is sad. The paper is good, the hand are straight, but the tracks cannot be less that 0.17. I have set the printer to use max toner thickness when printing (for my printer it is setting to print on a transparent medium). Maybe if I set the printer to use less toner the resulting tracks will be less in width, because the toner will flow less under the iron.
- 0.2mm 6px 0.2mm
- 0.3mm 9px 0.3mm
- 0.4mm 13px 0.43mm
Oh! For some unknown reason the gaps are ideal!
So, everything turns out pretty good, but, alas, the tracks are off size.
On a side note about laser printers. My HP LJ 1200 can do max 600dpi. This means 236 lines per 1 cm. Which is 24 lines per 1mm. I think that error is huge at such sizes. Especially with a refilled cartridge.
Does anybody know how much toner flows during printing? Let's do a simple experiment. Lets' print the pcb layout on the paper and measure the size on the paper BEFORE the transfer.
Let's do 4 experiments (let's use vernier calipers as a ruler to get more precision):
- rich toner flow print on photo paper
- usual print on a photo paper
- usual print on a usual A4 paper (after rolling the sheet though the print get it termally shrunk)
- measure the just made pcb
I'll measure only the 0.1mm track.
- rich toner print on phto paper
per 1 mm - 33 px.
track 0.1mm - 6 px - 0.17 mm
- usual print on a photo paper
per 1 mm - 35 px.
track 0.1mm - 5 px - 0.15mm
- usual print on a usual A4 printer (thermally preshrunk)
per 1 mm - 35px
track 0.1mm - 5 px - 0.15mm
- just made pcb
per 1 mm - 35 px
track 0.1mm - 7px. - 0.2mm
So, how can we even talk about 0.1mm if right after printing on the paper the track is wider? And then we will melt it with an iron, so it will get a bit wider again. Hoping that it will be etched on the edges and will get thinner is stupid.
Maybe it is my printer's problem. Maybe 600dpi is just not enough. I have HP LJ 3055 at the office which can do 1200dpi. I have done a couple tests on it with usual office (80g/m2) paper and on HP Premium Photo paper.
Here is the result for 0.1mm track:
Photo paper: 0.125 mm
Usual office paper: 0.11 mm
Not bad! A lot better than my home 600dpi. However, the toner is very different. It does not hold on the photopaper good enough. The tracks have very rough edges. The 0.05 mm tracks are not intact (the size is 0.09mm). Apparently, the new printer aggressively tries to save toner. I did not find any way to make it spend even more toner.
Lets transfer it to the copper.
On the copper the 0.1mm track is 0.17mm
10 minutes of etching in a hard ferric chloride solution and here is what I got:
(As you see the quality is a lot worse than with home 600dpi prints)
The resulting size of 0.1mm track is 0.18 for one and 0.2mm for the other.
The track width for 0.05mm track is 0.15mm. But it is very very rough and broken in the middle.
No 0.05mm track has passed the electric test. 0.2mm gap was conducting as before. I cleaned the 0.2mm gap with a paper knife and it passed the electric test.
The result: 1200dpi does not help, the toner/printer is not very suitable for the task.
Then I decided to make a new test pattern and test repeatability.
Here is the patern:
The places where toner did not stick to the paper are marked. I don't know why it happens sometimes, but it is related to the way I pure water on the paper and when I pull the paper. I have notice that it is better to pure the water right between the paper and the board.
The new pattern has SMD pads.
One has width of 0.44mm and gap of 0.17mm in the layout.
The other has pad width of 0.63mm and the gap of 0.62mm.
In the top right connect you can notice SMD 0402 capacitor pad with silk screen.
I took the HP Premium paper and use my home 600dpi printer.
Size on the paper:
- 0.63 mm - 0.63 mm
- 0.44 mm - 0.43 mm
- 0.1 mm- 0.14
The resulting sizes on the board after etching:
- 0.63 mm - 0.65 mm
- 0.44 mm - 0.43 mm
- 0.1 mm - 0.22 mm
- 0.05 mm - 0.18 mm
I have no idea why 0.43 did not change at all. Maybe it is some special size which printer can print right and some kind of electrochemical forces hold the toner together. Or maybe it is the size when a bunch of errors at different stages compensate each other.
The tracks are smooth, the edges are smooth.
Electric test showed that all tracks are conductive; 0.2mm gap has a bridge. 0.2mm passed after I deleted the bridge with a paper knife.
4 of out 4 cases 0.2mm gap has a bridge witch can be fixed easily with a paper knife. Maybe it is better to do before etching.
SMD pads test showed that they have no short circuits at all. The pads are of correct size then the gaps are also correct. Which means that we have 0.17mm gaps without shorting. Don't ask why. I don't know.
- HP Premium Photo Paper (Q199HF) rules!
- If you want to use a laminator for toner transfer then you should get one with temperature control.
- Etch fast! The longer you etch the more undercutting you get.
- Iron temperature must be selected individually for each combination of iron, paper, toner and, maybe, a board.
- The limit of TT with 600 dpi printer with refilled cartridge with cheep noname toner is 0.18mm.
- In order to get a track 0.18mm-0.2mm wide you need to draw 0.05-0.1mm track in the pcb layout software. This is really important and it sucks.
I think the topic of 0.1mm with TT method can be closed. It is simply impossible to do with toner transfer! However, what you can do is enough to do TQPF144 for AVR32 without any problems (TQPF144 is 0.23 pad with 0.24 gap).
If someone thinks he can reliably and repeatably do 0.1mm track with TT I will be happy to see a 3200dpi scan or a good macro photo with a ruler. But I doubt that will ever happen.
Glossy paper definitely absorbs better that transparent film. Transparent film is used in photo method above a photo resist layer. So, the width of a track on a transparent film cannot be any smaller than on a glossy paper. So, the minimum width for the transparent film is 0.15mm. Then during the UV lighting some shadowing and parasitic lighting happens. Because of this the track can be wider or narrower. Sometimes the deviation is really big. For example, in my tests for 0.15mm the resulting track was anywhere from 0.1mm to 0.2mm.
Experiment with 0.09 track showed that it is easily etched off in many places. So, I think the limit for photoresist method is 0.15mm. If someone thinks he can do reliably and repeatably tracks smaller that 0.15mm using photoresit then I am waiting for a macro photo or a 3200dpi scan.
ACHTUNG! WARNING! ACHTUNG! WARNING!
Later I found out that the above mentioned HP Premium Photopaper absulutelly does not work for PCB's larger than 3x5 cm because when I pull the paper, it pulls our pieces of toner from the pcb. Read the next part for information about the right paper.
ACHTUNG! WARNING! ACHTUNG! WARNING!