Problems with DIY PCB UV exposure unit

[brian_mk]

I have been making PCBs at home for years.
Until recently I used Positiv 20 with a two tube DIY UV exposure unit.
The two tube unit works quite well but is only suitable for small PCBs.

A while ago I decided to build a larger UV exposure unit with five tubes using an old flatbed scanner.
I put a lot of time and effort into the project. (see photo).

I am experiencing a problem when trying to use the unit: The traces near the center of the board are thin, distorted and apparently over exposed.

At first I thought it might be too long an exposure time or light seepage / diffraction around the artwork.
I discovered that the PCB is not perfectly flat: It is bowed in the long direction by 1.0->1.5mm
The flimsy plastic scanner lid does not apply sufficient presssure to push the board into contact with the artwork over the entire board area.
So for my next attempt I left the lid off and placed a heavy (3kg) book on top of the PCB to keep it flat.
The problem still exists.

The attached image (from a flatbed scanner) shows an initial small test PCB exposed in 4 strips: 2min, 4min, 6min, 8min.
Even though the Positiv 20 coating on the test board looks mottled and not very even, the traces for the 8min exposure look about right - perhaps slightly under exposed.
I exposed the full size board for 10 min.
You can see the problem in the lower scanned image.
The traces around the edges of the board are not that bad.
Those toward the center are blurred, mis-shapen and over exposed. It looks like light seepage around the artwork.

To eliminate a possible problem with one of the tubes, I tried swapping the center tube for one at the edges.
It made no difference.

The artwork is printed using an Epson inkjet (original Epson dye based inks) and Jeststar film. I have used this in the past with no issues.
To eliminate the possibility of insufficient ink density, I tried reversing the print and making a
contact print onto Rollei Ortho 25 sheet film using a photographic enlarger. That produces very dense blacks.
It made no difference to the result.

I am baffled and have just about run out of ideas. Any suggestions?

[Ian.M]

The edges don't get any light from adjacent tubes past the edge, so of course the UV intensity is higher in the middle.  To fake an infinite array of tubes, add polished aluminium or stainless sheet full height vertical mirrors at both edges, spaced a bit less^* than half the tube to tube spacing from the edge tube.

* To attempt to compensate for the less than 100% reflectivity of the mirrors.

[jmelson]

I use dry film resist with a vacuum frame to expose the PCB material.  The frame is two sheets of thin Plexiglas with a small vinyl tube as a gasket around the edge.  This tube is sold on eBay for protecting fishing hooks, but is sold in several foot lengths.  I glue some small blocks of the Plexiglas to keep the tube from being sucked inward by the vacuum.  This holds the master artwork snug against the sensitized PCB and prevents stray light from spreading under the dark part of the artwork.  I only use a one minute exposure in my rig.
Jon

[brian_mk]

The edges don't get any light from adjacent tubes past the edge, so of course the UV intensity is higher in the middle.  To fake an infinite array of tubes, add polished aluminium or stainless sheet full height vertical mirrors at both edges, spaced a bit less^* than half the tube to tube spacing from the edge tube.

* To attempt to compensate for the less than 100% reflectivity of the mirrors.

I measured the light distribution across the glass bed using an incident light meter. It's probably only sensitive to the visible part of the spectrum but it should still give a good idea of the light fall off around the edges. I only measure a fall off of around 1/2 stop. I could be wrong but I doubt that would be enough to acount for the effect I am seeing. The test strips I carried out covered a range of 2 stops.

The Mega LV204 unit does include reflective edge strips at a roughly 45deg angle.
Many other similar exposure units don't.

I will see if I can find something I can use as an angled reflector and check it using the light meter.

[coppercone2]

I read a single point of high power light is better from a strong LED. I think I did it with a 15W UV led, it was expensive, and I did not play with it much because kinda sick of CAD

Based on the intensity being too strong in the edges, can you elevate the center bulbs? ARCH shape

[jpanhalt]

Try removing the middle tube and see what happens.  I also used a reflector, but I suspect the main problem is simply overexposure in the middle.

Edit:  If you have an older digital camera with a low-pass filter on top of the sensor (e.g., a Nikon D70), remove the filter to make a UV sensitive camera.  It's a very easy mode, but it trashes the camera for other uses.

[brian_mk]

This evening I added angled reflective strips around the outermost tubes (as per the Mega LV204).
The incident light meter indicates it has made a definite improvement.
It now indicates virtually no light falloff around the edges.

Tomorrow I will try a test exposure using a shorter exposure time.
The previous exposure was 10 min. Although the center is overexposed, the edges look reasonable.
If the edges were 1/2 stop down (as indicated by the meter) that would suggest 7.5 min should give the right exposure at the center.
I think I might go a bit furter and go for a 1 stop reduction i.e. 5 min.
I will also use a heavier weight to remove any possibility of a gap between the board and the artwork caused by bowing.

[jpanhalt]

When I did DIY PCB, I sandwiched the artwork and PCB between single pane window glass and used spring clips to hold the assembly together.  Window glass transmits the UV of interest. Some people add a little good water to decrease reflections in the inevitable air gap.  I didn't and never had a problem

[tooki]

Edit:  If you have an older digital camera with a low-pass filter on top of the sensor (e.g., a Nikon D70), remove the filter to make a UV sensitive camera.  It's a very easy mode, but it trashes the camera for other uses.

That’s an IR filter. They don’t filter the sensor for UV because the lenses already do that, due to usually being made of glass that’s not particularly transparent to UV (which is why serious UV photography requires special lenses). To dabble in UV without buying special lenses, you can just get a UV band pass filter to put onto your existing lenses, just with loss of brightness. Not likely to be a problem here, pointing straight at UV tubes.

[jpanhalt]

It's a high frequency "long pass" filter.  Of course, by UV I was referring only to near UVA which glass fluorescent tubes produce and is most important for positive photoresists..

From Nikon: https://www.nikonimgsupport.com/eu/BV_article?articleNo=000006377&configured=1&lang=en_GB#:~:text=The%20filter%20is%20built%20into,high%2Dfrequency%20waves%20in%20images.

An Optical Low Pass Filter (OLPF), is sometimes referred to as an Anti-Aliasing filter. The filter is built into many digital cameras and is located directly in front of the image sensor. The main role of the OLPF is to stop high-frequency image information and reduce the effects of moiré and false colour caused by high-frequency waves in images.

From Wikipedia ("Optical Filter")

Longpass
A longpass (LP) Filter is an optical interference or coloured glass filter that attenuates shorter wavelengths and transmits (passes) longer wavelengths over the active range of the target spectrum (ultraviolet, visible, or infrared). Longpass filters, which can have a very sharp slope (referred to as edge filters), are described by the cut-on wavelength at 50 percent of peak transmission. In fluorescence microscopy, longpass filters are frequently utilized in dichroic mirrors and barrier (emission) filters. Use of the older term 'low pass' to describe longpass filters has become uncommon; filters are usually described in terms of wavelength rather than frequency, and a "low pass filter", without qualification, would be understood to be an electronic filter.

[brian_mk]

Success.

Changes included...

1) Correct the light fall off at the edges by adding reflective strips around the outer tubes.
2) Rotate the PCB 90deg on the platten to minimise the effect of any remaining edge light fall off.
3) Reduce the exposure from 10 min to 6 min (just under 1 stop). That seems about right. I don't think I could have gone any lower without leaving fog in the exposed areas.
4) When I opened up the scanner case, I realised that the way the glass was fixed, the edges were open to the UV light beneath.
Would that act like a light pipe? I masked the edges with black tape just to be safe.
5) I used a heavier 4kg weight concentrated at the center of the PCB to remove the bowing and keep if flat against the artwork.

I am unsure which of the changes were the most significant.

The PCB will need a bit of touching up with a Sharpie before etching because of problems with uneven Positiv 20 coating, flecks of dust etc. but I think it should be useable.

[jpanhalt]

If that mask is from an inkjet , there seems to be something wrong with the printer or the film you are using. Are you printing in black or color?  Yoo may have better luck with color.

The first image is a typical inkjet mask and the second is a microscopic view of an inkjet without black.  Both work fine. 

[brian_mk]

If that mask is from an inkjet , there seems to be something wrong with the printer or the film you are using. Are you printing in black or color?  Yoo may have better luck with color.

The first image is a typical inkjet mask and the second is a microscopic view of an inkjet without black.  Both work fine.

I haven't uploaded a picture of the inkjet mask or of the high density contact printed film version (using now very expensive Rollei Ortho 25 8" x 10" sheet lith)!
The images I uploaded are of the surface of the PCB after coating with Positiv 20, exposing to UV and developing in NaOH solution.
Some images were using a flatbed scanner, others using a camera.

Anyway, I etched the board this morning and am happy with the result.
I just need to drill, tin the copper and assemble the components.
I just hope my design works as expected.

[jpanhalt]

I misinterpreted what that original image was. I used presensitized boards from DigiKey and after development,* the remaining resist was homogeneous and clear.  Its color was sufficiently darker than exposed copper so finding defects was not a problem.

*I used 0.4M KOH.  I chose that as it seemed less sensitive to development time than NaOH.  In fact, etch time was more critical then development time.  I also used a hot air "gun" after development to slightly remelt the resist edges. 

[brian_mk]

I should add thanks for eveyone's suggestions.
In particular Ian.M for his comments regarding adding reflectors to reduce the light fall off around the edges.

[brian_mk]

I should add thanks for eveyone's suggestions.
In particular Ian.M for his comments regarding adding reflectors to reduce the light fall off around the edges.

My only reason for using Positiv 20 is that, as a retired engineer, I have an existing stock of old FP4 board.
Most of it is double sided, which has the advantage of less bowing but it's more difficult to use for DIY at home when you don't have the luxury of 'plated through'.
It also uses more etching solution.

Positiv 20 is a pain to use - it's hard to get an even coat without it looking like hammerite. It's also almost impossible to avoid dust unless you have a clean room at home.

I initially used pre-coated board when testing the new exposure unit but found the coating to be very thin and a lot more sensitive to UV.
That completely threw my existing Positiv 20 exposure timings.

[tooki]

It's a high frequency "long pass" filter.  Of course, by UV I was referring only to near UVA which glass fluorescent tubes produce and is most important for positive photoresists..

From Nikon: https://www.nikonimgsupport.com/eu/BV_article?articleNo=000006377&configured=1&lang=en_GB#:~:text=The%20filter%20is%20built%20into,high%2Dfrequency%20waves%20in%20images.

An Optical Low Pass Filter (OLPF), is sometimes referred to as an Anti-Aliasing filter. The filter is built into many digital cameras and is located directly in front of the image sensor. The main role of the OLPF is to stop high-frequency image information and reduce the effects of moiré and false colour caused by high-frequency waves in images.

From Wikipedia ("Optical Filter")

Longpass
A longpass (LP) Filter is an optical interference or coloured glass filter that attenuates shorter wavelengths and transmits (passes) longer wavelengths over the active range of the target spectrum (ultraviolet, visible, or infrared). Longpass filters, which can have a very sharp slope (referred to as edge filters), are described by the cut-on wavelength at 50 percent of peak transmission. In fluorescence microscopy, longpass filters are frequently utilized in dichroic mirrors and barrier (emission) filters. Use of the older term 'low pass' to describe longpass filters has become uncommon; filters are usually described in terms of wavelength rather than frequency, and a "low pass filter", without qualification, would be understood to be an electronic filter.

Those are two very different things. By “high frequency”, Nikon doesn’t mean the wavelength of the light, but the spatial sampling frequency. These are basically a slight blur filter to prevent aliasing (moiré). It doesn’t have anything to do with the wavelengths of light involved. It is NOT a UV filter. (The optical low pass filter usually integrates the IR-blocking filter, but they do not have to be combined.)

People who do astrophotography and want maximum resolution above all else will sometimes modify their cameras to remove the OLPF, but leave the IR filter in place (or install a pure IR filter, or just use an IR filter on the lens). Nikon and Canon even sold special astrophotography versions of some cameras without an OLPF, with the warning that colorful moiré patterns can appear when using them to shoot everyday things.

Yes, there is some overlap in filter terminology, and that’s unfortunate. But you’re decidedly mistaken here.

https://petapixel.com/what-is-a-low-pass-filter/ says:

Unlike an infrared filter, a low-pass filter operates on high-frequency spatial information, whereas an infrared filter removes spectral information. To put it simply, the optical low pass filter slightly blurs out the image before it reaches the silicon sensor.

http://www.optics-online.com/lpf.asp?sort=Thickness says:

An OLPF cuts off the lens MTF above the sampling frequency of the imager resulting an overall MTF curve that approximates a step function in spatial domain.

https://photographylife.com/what-is-low-pass-filter

See how they are NOT talking about blocking UV? A “long pass filter” is NOT what is used in DSLRs.

[jonpaul]

bonjour à tous
We made many exposure sources similaires, and used a translucent matte diffusion mylar film spaced 1..5 cm from the lamps, to even out the light over the surface.

Easy to try,

Jon

[nigelwright7557]

Why not bite the bullet and get your pcb's manufactured ?
JLCPCB are very cheap, good quality and fairly quick depending on postage option.
I have had over 250 designs through them and rarely get a problem.
Its 40 years since I last made a DIY pcb.

[brian_mk]

JLCPCB are very cheap, good quality and fairly quick depending on postage option.

There's a JLCPCB ad "1-8 Layers PCBs for $2" on this very page. Sometimes saving time (getting your PCB in a couple hours rather than 2 weeks) is more valuable than money (saving $50). Sometimes there are things JLCPCB can't do, like ultra-thick copper.

If you're doing an HDI PCB with BGAs, an external manufacturer is the obvious choice, but that's far from the only board type out there.

How expensive is it for customers based in the UK?
We usually get screwed in the UK compared to prices in the USA when buying any electronic components.

For example 100mm x 160mm double sided FR4?
The prices on their website appear to be in USD.
What about delivery and import charges?

I am retired and only do this for a hobby, so unless it's dirt cheap it's probably not worth it.
I use Designspark PCB software from RS Components.
The user interface is really frustrating to use. It also crashes quite often, but it's free.
Is JLCPCB able to use the files Designspark generates?

[tooki]

Given that you can change the currency to GBP, I think it’s fair to assume they have plenty of UK customers.

It took me two minutes to quote your board: £8.28 for the boards (5 pieces, the minimum), standard 2 day production time. £2.97 for the cheapest 2-week shipping, faster options from about £13.

Import duty is your responsibility, so it’s whatever the UK does.

JLCPCB accepts (only) industry-standard Gerber files. If your program can generate those, then you’re fine.

[Siwastaja]

Perfect contact with the artwork and PCB is critical. This is difficult to achieve by sandwiching with glass, even if you add a lot of weight. Using a thin plastic film instead and adding a vacuum pump which sucks air under that sheet, pulling the sandwich together is a good, well tested idea. Unevenness of light is another possible source of problems.

[tooki]

JLCPCB is dirt cheap, but doing it yourself is also dirt cheap if you don't put a price on your time. It's all a matter of whose labor you want to enjoy - your own, or some robot's in China (and I mean literal robot).

There is no practical method to make real plated through-holes, vias and slots at home. (By “real” I mean not those goofy swaged eyelets.)

And in all truth, JLCPCB charges less for finished PCBs than western suppliers of PCB materials charge for copper-clad board.

[Siwastaja]

We did do through hole plating in a home-like environment but yeah, it's a lot of work, experimenting, yield problems, etc. While possible, I realize it's not practical for most. (It's also a personality question how much time you are willing to spend in playing around, if you just want the results, it's not worth it. But if you enjoy the journey itself - then sure.)

We used the conductive ink from Think&Tinker Ltd., see https://www.thinktink.com/stack/volumes/volvi/condink.htm for holewall activation, then electroplating in a DIY tank and a DC power supply. Yield was usually around 99% which is actually quite crap - with a board of say 200 holes, you'd typically have 1-2 of those that fail to conduct.

The reason to do the boards yourself is not cost savings, but very quick turn-around. It enabled innovative working patterns. Sometimes I got an idea, draw the layout in PCB CAD and make the PCBs, and 4-5 hours later I have a working prototype. Something with difficult IC packages or just tens and tens of tracks, so nearly impossible or just very cumbersome to prototype deadbug way (or breadboard).

One example would be an adapter board for a connecting several FPGA development boards together using wide parallel buses. Hundreds of pins and wires with 0.050" pitch connectors, would have taken full day soldering manually using wires. PCB done in less than 2 hours and connectors simply soldered in. It takes a few seconds to draw a track in PCB CAD, compare that to actually cutting, stripping, soldering a wire!