How is it possible to get faint green and disappearing tracks when stripping photoresist?

Question

I begin to wonder now what is going on.

When I make circuits, I use a presensitized PCB board from MG chemicals along with their positive developer (sodium hydroxide?)

This is my 2nd attempt of a circuit I attempted to develop (I'm reusing the mask that was printed with toner on vellum paper rated at 105gsm).

On the first attempt, the circuit background turned pure copper which is good but some tracks faded away.

Now the second attempt got worse even though I taken advice.

One advice I took (I believe from kinsten, a chinese PCB supplier) was to move the UV light for exposure to 4 to 6 cm away from the board and expose for a shorter time. For both tests I used roughly a 20 minute exposure time.

For my developer mix, I used about 12 parts of water to 1 part of developer which is weaker than what MG chemicals suggested, and I deliberately did weaker to see the action.

I then shut my room light off and the brightness in my room was at about 10% since outside was dusk. It was still hard to see fine print without light shining on it.

Once I shut my light off, I go and use an ultra-bright red LED as my safelight when handling and developing the PCB since the presensitized PCB is sensitive to light.

I then sandwiched the circuit board and the artwork together (toner side to the circuit board)

After exposure, I briefly examined the board and noticed the pattern. Then I put my board in developer for about two minutes. Once I'm done, I rinsed it and thought my circuit is perfect, but once I turn the normal light back on, I was dismayed to see that part of the circuit board background was faint green, but what's worse is that one track was fading away into the background, yet my original artwork didn't call for this.

What could I be doing wrong? Could the vellum be the culprit? am I forced to just use transparencies? I've wasted at least $20 to unintentionally make failing boards and I don't want to make anymore.

Someone suggested using sodium silicate instead of sodium hydroxide but I think there's more to it.

Any ideas?

UPDATE

As requested by someone, I needed to mention in my question that I don't have a good quality camera to take the pictures with.

The track size thats problematic is 0.3mm wide.

The UV source is this cfl-like light: spencers.scene7.com/is/image/Spencers/00912691-a?$Thumbnail$

Answer 1

I recommend you approach the problem in a different way.

You have two different parameters to control for, exposure time, and developer time, so it may take several runs to find out what works well.

Rather than try to make a working PCB, instead try to calibrate your process. You could be as much as an order of magnitude wrong on the exposure and developing times.

Don't waste an entire piece of PCB on an experiment, use small strips.

You may need to calibrate the exposure time. The process for that is very similar to making a 'dark room test strip', and is described here https://electronics.stackexchange.com/a/254412/9006

Then, you need to refine the development process. Make the developer up exactly as specified by the manufacturer.

IIRC, we 'wash' the developer across the board with a soft brush until it has clearly changed.

Design a small test PCB (say 1cm square), containing tracks and spaces at the spacing you need, and one 'size' narrower, and one 'size' wider. For example, if you are aiming to make 10mil track and 10mil space, make a pattern with 8mil, 10mil and 12mil track and space so that you can see what is necessary for development and etching. Replicate the test PCB pattern across a strip of material, say 1cm by the width of your pristine PCB 'blank'.

Then go through a range of exposure times for the PCB strip using the 'test strip technique'

Do not just put it into the developer for a fixed time and expect it to 'just work'. You might be developing too long, or not long enough. Try to develop, and keep examining what happens, say every 5 seconds, while keeping track of the time.

The first experiment might fail, but should 'bracket' (i.e. bound on both sides) something which approximately works, then refine by using finer time steps, so that the time for exposure and development is within 10 seconds.

Once you have a process that works, and is repeatable, then it becomes much easier. With a stable, well defined process, you will be able to use finer track and space.

We use 'proper' laser-printable PCB film, but friends say they use tracing paper with good results.

Edit:
We had to futz with the laser printer to get good black and solid tracks with crisp edges. It was an expensive office printer, but its settings were optimised for nice looking photo's, and to make the toner last. So we had to try permutations of printer settings, and examine with a magnifier or microscope. For large areas of black, we found it helped to have a fine hatch (maybe 10-25%); for some reason the areas got much sharper. (We theorised about the charge on the drum leaking, but never got to the bottom of it. Instead we always used fills with a hatch, and they worked fine.)

Also, IIRC the printer claimed higher resolution than we could get. It claimed 2400 dpi, but AFAICT used different droplet sizes. It appeared, on the settings we tried that it got some of its resolution by varying the dot size, and by image processing the image, but couldn't actually put a dot at any of the 2400 positions across an inch. When we looked at fine detail through a microscope the edges of lines were 'fuzzy', though they looked okay to the naked eye. When we developed and etched the board, the 'fuzz' was a low-resistance short circuit. I think we ended up treating it as a 600dpi printer, and I redesigned a few footprints to enhance the space between pads so it was slightly increased. We were doing 10mil track & space.

Summary: making PCBs is a very 'analogue process' and all the steps need to be understood and characterised, and that may need quite a low of experimentation. Once the process is well understood, it can be repeatable, but until then it can be frustrating.

Answer 2

The contrast of the mask you use gives you the size of the time window for exposure that will give a good result. If the contrast is too low, that window may have negative size, and you won't get a good result with any setting.

The best result I've had was when I used the same process as for offset printing. The transparent film was prepared in a phototypesetter, which gives excellent contrast, then I used the same settings as for printing plates (4 kW UV-A, two minutes direct exposure followed by two minutes through a diffuser curtain). The large contrast translated directly to a wide difference in exposure, which gave me a large time window for development.

I still had to make sure to leave large copper areas intact as they take a lot of time to dissolve and degrade the chemicals quickly.

The other important point is that the "printed" side needs to be placed on the PCB, or the hard shadow will be smaller due to light coming in at different angles (especially when using a diffuser curtain), and ideally you'd be using a vacuum exposure unit that presses the film against the PCB during exposure as well.