At 600 dpi, doing one scan line at a time, I estimate the total processing time for a single sided 3.2 in. X 4.0 in. board, to be approximately 10 hours and 40 minutes.
EDIT: Of course that is just with one laser diode being utilized for processing. Time could be reduced, by using multiple diodes.
Gosh, even a 1985 Gerber Photoplotter would do that board in less than an hour.
At that amount of time, does this make sense compared to simpler optical means?
Gosh, even a 1985 Gerber Photoplotter would do that board in less than an hour.
As I mentioned, that is with one laser diode. If you had an array of say 12 diodes, processing would be much faster. Baby steps
Seems to me, producing masks and a good process and fixture for registration would be faster and easier.
I agree whole heartedly, BUT the problem is obtaining high quality masks to produce nice boards. If I could still obtain PMTs locally, I would go that route, but PMTs are a thing of the past. Then you have photoplots, but there is nothing available locally, and they are fairly expensive, and who do you trust with your money, to obtain a 1:1 ratio. Then you have printers, which are not truly opaque, and therefore make it very difficult to obtain quality exposures, to produce a quality etched board. I have the process, but the positives or the negatives are the catch.
As for milling, your spaces between traces are limited to the smallest mill available, as well as inside radii. Additionally, the milled boards that I have seen, didn't look all that great to me.
Each of those methods are perfectly fine, if you have the right tools and equipment.
The laser power is adjustable... I think it might just work...
I think I've also seen laser engravers used with painted PCBs.
Pass one burns off the paint on the to-be-etched areas, and then they optionally repaint and burn off pads for solder-mask effect.
No details on how the residue cleans, or fumes, or how well the resist works in practice....
One thing you absolutely have to do when plotting a raster image is to provide for an acceleration zone before the beginning of each line. This will guarantee that in the part being exposed, the laser velocity will be constant. Otherwise, the beginnings of each line will be overexposed.
At 600 dpi, doing one scan line at a time, I estimate the total processing time for a single sided 3.2 in. X 4.0 in. board, to be approximately 10 hours and 40 minutes.
Gosh, even a 1985 Gerber Photoplotter would do that board in less than an hour.
That's a good litmus test reference.
Seems rather than a pixel based raster, with tiniest dots, it may be faster to become somewhat more like a a photo plotter.
Run some tests on under-and over-exposure limits of varying DOT sizes from the laser.
Choose the largest dot size, and draw like a photo plotter/pen plotter, with PCB traces as vectors, flashes as filled spirals.
Drilled pads can stop the spiral early, to give the drill dot, but would also need trace-end pull-back to unmask the hole.
One thing you absolutely have to do when plotting a raster image is to provide for an acceleration zone before the beginning of each line. This will guarantee that in the part being exposed, the laser velocity will be constant. Otherwise, the beginnings of each line will be overexposed.
I am really not sure what you are saying, but I do understand overexposing the beginning of each scan line.
Even though I am uncertain what you are saying, please consider this....
1. The laser will move at a constant speed with no ramping.
2. Every board will have a boarder around the perimeter of at least 1/32" with no copper, meaning that during that first 1/32" there will be no exposure
One thing you absolutely have to do when plotting a raster image is to provide for an acceleration zone before the beginning of each line. This will guarantee that in the part being exposed, the laser velocity will be constant. Otherwise, the beginnings of each line will be overexposed.
It depends on how much control over the laser you exert - If you can PWM the beam strength, you can ramp it from off to your desired power level at the same rate as you ramp from 0 to your desired speed. Laser cutters do this when doing vector-style cuts, though when engraving they usually do provide a ramp-up range like you suggest. It's possible the results are more consistent that way.
At 600 dpi, doing one scan line at a time, I estimate the total processing time for a single sided 3.2 in. X 4.0 in. board, to be approximately 10 hours and 40 minutes.
Gosh, even a 1985 Gerber Photoplotter would do that board in less than an hour.
That's a good litmus test reference.
Seems rather than a pixel based raster, with tiniest dots, it may be faster to become somewhat more like a a photo plotter.
Run some tests on under-and over-exposure limits of varying DOT sizes from the laser.
Choose the largest dot size, and draw like a photo plotter/pen plotter, with PCB traces as vectors, flashes as filled spirals.
Drilled pads can stop the spiral early, to give the drill dot, but would also need trace-end pull-back to unmask the hole.
Back then, when I worker for Gerber,we had to vary the output of the Osram lamp with the velocity of the bed. The head was stable and the bed moved, stepper driven.
Later, we used zeon tubes on our higher speed plotters (42 Drum Plotters) and varied the pulse based on linear velocity.
I came across another test of a 405nm 20mW laser performing exposure. In this instance, the traces were solid copper, but they were also jagged, do to loose mechanics and wobbly structure.
He was basically exposing at a feed rate of 12in/min.
I came across another test of a 405nm 20mW laser performing exposure. In this instance, the traces were solid copper, but they were also jagged, do to loose mechanics and wobbly structure.
He was basically exposing at a feed rate of 12in/min.
Did they mention what dot-size, for that feed rate ~ 5mm/sec.
Well..... I certainly hope that I will be a bit more successful than some of the other folks who have tried. Just the gears alone, to achieve 600, 1200, 2400, etc.... is $70. This experiment will be a little costly. Good thing I am just shooting for a very small machine
Actually the film that I intend to use is negative acting, so the first 1/32 would be exposed. I keep thinking of positive masks
If you use negative acting file, one approach would be to take engrave-pcb design flows, which are Gerber -> trace-outline-polyline
derived for the router paths, and then apply that pathway to your diode.
Result is an etched PCB, that looks like an outline-engraved one. Plot times should be much less than a full raster scan image.
Okay It will step and pause equal amounts according to the accuracy provided by a Propeller counter
Why pause? Just start traversing the line and fire the laser on the fly. You really don't have to be concerned with discrete "dots." Step-and-pause is gonna be really slow and will prematurely wear out your mechanics.
If you use negative acting file, one approach would be to take engrave-pcb design flows, which are
Gerber -> trace-outline-polyline
derived for the router paths, and then apply that pathway to your diode.
Result is an etched PCB, that looks like an outline-engraved one. Plot times should be much less than a full raster scan image.
There is no code support for that in conjunction with the use of a Propeller as the processor. I wish there was.
If you use negative acting file, one approach would be to take engrave-pcb design flows, which are
Gerber -> trace-outline-polyline
derived for the router paths, and then apply that pathway to your diode.
Result is an etched PCB, that looks like an outline-engraved one. Plot times should be much less than a full raster scan image.
There is no code support for that in conjunction with the use of a Propeller as the processor. I wish there was.
I'd expect a PC host to manage the (non-trivial)
Gerber -> trace-outline-polyline
and then the resultant created simple path polylines can feed to the Prop.
Given I'd call engrave-pcb design flows mature, there must be PC based code somewhere out there ?
jmg, are you advocating for plotting vector paths with the laser? If so, I would advise against it, since exposure control is almost impossible in this scenario. Coming directly from Gerber data, there is simply too much overlap for this to be successful.
I am not certain, but I would imagine that a large portion of board house that use LDI, probably also use masks. Place a mask over the board, and just scan the entire board with the laser on.
jmg, are you advocating for plotting vector paths with the laser? If so, I would advise against it, since exposure control is almost impossible in this scenario. Coming directly from Gerber data, there is simply too much overlap for this to be successful.
There must be some film tolerance for at least double-exposure, or Gerber draw/flash would never have worked ?
I was thinking mainly for engraver type paths, but you do raise a good point : that the engraver software probably does not care/check about multiple passes, (aside from wasted time) and so could possibly have > 2 overlays.
Most cases I can think of would result in either 1 or 2 passes, but there might be corner cases > 2?
I guess some checking code could verify both Gerber and engraver tool paths for exposure risks ?
I am not certain, but I would imagine that a large portion of board house that use LDI, probably also use masks. Place a mask over the board, and just scan the entire board with the laser on.
There would be no point in using a scanning laser in this scenario. Just flood the masked board with UV.
There must be some film tolerance for at least double-exposure,
In fact, they recommend double exposure to reduce bad board count
I'm talking about correctly exposing with either one-pass, or two. ie tolerating one unit, or two units of (mJ/cm2), without suffering over-exposure effects.
I guess with more SW work, you could take explicitly single-exposure paths, and either double run those, or double the power.
Comments
Gosh, even a 1985 Gerber Photoplotter would do that board in less than an hour.
As I mentioned, that is with one laser diode. If you had an array of say 12 diodes, processing would be much faster. Baby steps
I agree whole heartedly, BUT the problem is obtaining high quality masks to produce nice boards. If I could still obtain PMTs locally, I would go that route, but PMTs are a thing of the past. Then you have photoplots, but there is nothing available locally, and they are fairly expensive, and who do you trust with your money, to obtain a 1:1 ratio. Then you have printers, which are not truly opaque, and therefore make it very difficult to obtain quality exposures, to produce a quality etched board. I have the process, but the positives or the negatives are the catch.
As for milling, your spaces between traces are limited to the smallest mill available, as well as inside radii. Additionally, the milled boards that I have seen, didn't look all that great to me.
Each of those methods are perfectly fine, if you have the right tools and equipment.
I think I've also seen laser engravers used with painted PCBs.
Pass one burns off the paint on the to-be-etched areas, and then they optionally repaint and burn off pads for solder-mask effect.
No details on how the residue cleans, or fumes, or how well the resist works in practice....
-Phil
That's a good litmus test reference.
Seems rather than a pixel based raster, with tiniest dots, it may be faster to become somewhat more like a a photo plotter.
Run some tests on under-and over-exposure limits of varying DOT sizes from the laser.
Choose the largest dot size, and draw like a photo plotter/pen plotter, with PCB traces as vectors, flashes as filled spirals.
Drilled pads can stop the spiral early, to give the drill dot, but would also need trace-end pull-back to unmask the hole.
I am really not sure what you are saying, but I do understand overexposing the beginning of each scan line.
Even though I am uncertain what you are saying, please consider this....
1. The laser will move at a constant speed with no ramping.
2. Every board will have a boarder around the perimeter of at least 1/32" with no copper, meaning that during that first 1/32" there will be no exposure
It depends on how much control over the laser you exert - If you can PWM the beam strength, you can ramp it from off to your desired power level at the same rate as you ramp from 0 to your desired speed. Laser cutters do this when doing vector-style cuts, though when engraving they usually do provide a ramp-up range like you suggest. It's possible the results are more consistent that way.
Back then, when I worker for Gerber,we had to vary the output of the Osram lamp with the velocity of the bed. The head was stable and the bed moved, stepper driven.
Later, we used zeon tubes on our higher speed plotters (42 Drum Plotters) and varied the pulse based on linear velocity.
I came across another test of a 405nm 20mW laser performing exposure. In this instance, the traces were solid copper, but they were also jagged, do to loose mechanics and wobbly structure.
He was basically exposing at a feed rate of 12in/min.
http://www.jameco.com/z/21-334-GC-Electronics-Positive-Pre-Sensitized-Single-Sided-Copper-Clad-Circuit-Board-4-x-6-Inch_616251.html
4 x 6 Inch $4.39 ea, Can be exposed with common fluorescent lamps, Use with GC Developer Part Nos. 22-225 or 22-226
and they also have
http://www.jameco.com/z/14-046-Datak-LKG-Industries-Premier-Single-Sided-Pre-Sensitized-Circuit-Boards-4x6-Inch_2161326.html
4x6 Inch $7.95 ea, May be exposed with any ordinary 100W lightbulb, Use with positive-acting developer P/N 156734 ($21.49 ea)
maybe the more expensive option, can accept a higher plot speed ?
If I can find a decent supplier, I intend to perform my tests with this film:
dupont.com/content/dam/dupont/products-and-services/electronic-and-electrical-materials/printed-circuit-board-materials/documents/DEC-LDI8000-DataSheet.pdf
-Phil
Okay
Gerber -> trace-outline-polyline
derived for the router paths, and then apply that pathway to your diode.
Result is an etched PCB, that looks like an outline-engraved one. Plot times should be much less than a full raster scan image.
-Phil
Because stepper drivers need a pause in between steps
If you read the bitmap thread, you will see that I changed from dots to lines.
There is no code support for that in conjunction with the use of a Propeller as the processor. I wish there was.
I'd expect a PC host to manage the (non-trivial)
Gerber -> trace-outline-polyline
and then the resultant created simple path polylines can feed to the Prop.
Given I'd call engrave-pcb design flows mature, there must be PC based code somewhere out there ?
-Phil
There must be some film tolerance for at least double-exposure, or Gerber draw/flash would never have worked ?
I was thinking mainly for engraver type paths, but you do raise a good point : that the engraver software probably does not care/check about multiple passes, (aside from wasted time) and so could possibly have > 2 overlays.
Most cases I can think of would result in either 1 or 2 passes, but there might be corner cases > 2?
I guess some checking code could verify both Gerber and engraver tool paths for exposure risks ?
-Phil
In fact, they recommend double exposure to reduce bad board count
Phil
duh
I guess with more SW work, you could take explicitly single-exposure paths, and either double run those, or double the power.