Laser "projector" for PCB mask creation

I'm an electronics hobbyist, and would love to find a better way to create one-off and prototype PCB boards. I've done toner transfer, and abhor it. UV sensitive paint transfer works much better, but has its own share of problems. But, some time ago, I found someone who used lasers with copper blanks painted black to create a etching mask (of course, I'd much rather use a pair of glavos than his 2-axis table setup). Now I finally have the time and impetus to construct such a device, I'm looking for some advice on a few topics I know little about.

Laser

• Wavelength? He uses a 100mW 405nm laser. While black paint--by definition--absorbs all visible wavelengths, is there a compelling reason to use blue/violet laser over a red laser? I've read that blue lasers are more readily absorbed than red lasers in most materials, but does this hold true with black paint? Also, supposedly red lasers are "cleaner" (rounder) when slightly out of focus (see footnote *)--is this true in your experience?
• Wattage? I was going to go with 100mW, as that's what he uses. Is there any reason to use a higher wattage laser, other than speed? Any reason to go with a lower wattage, if I can get away with it?

* The ideal distance between the laser and galvo, and galvo and PCB will be crucial. It's something I'll likely have to determine after I have all parts in hand, and even then it will depend on the radius of the circuit being printed. Regardless, I'll still obviously have a slightly varying distance where the laser need to be in focus or very close, given by sqrt(a^2+b^2)-a, where 'a' is the distance between the galvo and board and 'b' is the radius of the circuit design.

Galvos

• Where/What to buy? It seems most galvos at reasonable price are found on eBay and are all manufactured in China, like this. They claim to be closed-loop, which I will want to avoid issues with hysteresis, but are they really? What is your opinion of them? What about the drivers routinely included with them? Lastly, some include a "show card"--I assume this is for running laser shows and is completely unnecessary for my needs?
• Angular resolution & precision? All of the galvos I see are rated on PPS, which is of no use to me--10PPS should suffice! For my application, angular resolution and precision are critical. While I'm sure this depends a lot on the DACs, what can I expect with Chinese "closed loop" galvos?

DACs

While I think I can get by with 16 bit DACs for smaller boards or boards with only larger components, but I'd much rather use 24-bit DACs for more flexibility. What do you recommend?

Safety

• I'm planning on enclosing everything in something the laser can't burn though, like aluminum or steel. However, I'd still like a window to look through to verify operation. Where do you recommend getting acrylic laser-blocking flat panels? I've searched, but most places sell it in larger quantities/size than I need, for wavelengths I don't want to use use (IR), or are extremely expensive.
• Any tips on safely focusing a laser of such power?

... for galvo-scanning you'll need a Theta-optic to receive a horizontal focus plane, what's the most expensiv part.

I'm using diodes with 405nm@200mw, 445nm@1w or IR-diodes with 808nm@1W and 975nm@5W and @9W - here some old infos to my diodes: http://reprap.org/wiki/Laser_Cutter

The vid in your link shows a poor focussing setup - ye shorter the distance from the lens to the focus, the better/smaller the spot.

Most direct used diodes has a 'bar'-shaped beam, so simply refocussing the diode wouldn't result in a round spot - see at the attached images here: http://forums.reprap.org/read.php?27...402#msg-155402

For this I'm mostly using fibercoupled diodes - here the fiber acts as a 'homogenisator' and gives a perfect round beam, that can be refocussed to the diameter of the fiber with a 1:1 optical setup.

When refocussing diodes directly I'm using a pin-hole, but here I'm loosing a part of the power ...

Viktor

It might be cheaper for creating PCBs to just use a x-y motion stage and a microscope objective to focus the laser.
You can then use a UV diode with photo-resist coated boards, or a IR diode to burn the black paint.
No F-theta, and I know you can probably find code from PCB Gerbers to G-code.

Where the Chinese usually skip on quality is the position sensor in the galvo system. You want the 30K amp. Why? Its settling time and bearing precision will be much, much better then a 75$ pair of 10K knockoffs.

Take a look for Cambridge Technology 6450s or 6860s or similar on Ebay, you might like what you find. Too big a mirror for most laserists, and very precise. You will need position sensing amps, unless you use G120DTs or or G330DTs which have torsion bar centering.

Discuss

These are very precise but not that fast. Right up your ally, and they have the amps:

http://www.ebay.com/itm/Cambridge-Te...item1c2f7717c3

Steve

Quote:

---

Originally Posted by VDX

... for galvo-scanning you'll need a Theta-optic to receive a horizontal focus plane, what's the most expensiv part.

---

Yeah, I looked at these and you aren't kidding about the price. My hope is the beam will be narrow enough even when slightly off of perfect focus that it won't be a problem. Most of the boards I make are quite small anyway, and I suppose if I ever do need something large I could burn up to a certain radius, refocus, burn again, etc.

Quote:

---

Originally Posted by VDX

I'm using diodes with 405nm@200mw, 445nm@1w or IR-diodes with 808nm@1W and 975nm@5W and @9W - here some old infos to my diodes: http://reprap.org/wiki/Laser_Cutter

---

Yeah, I'd like to avoid IR for safety reasons, but if you've had good luck with blue diodes as well I'll go with those. They're easier to find in higher wattages than red anyway.

Quote:

---

Originally Posted by VDX

For this I'm mostly using fibercoupled diodes - here the fiber acts as a 'homogenisator' and gives a perfect round beam, that can be refocussed to the diameter of the fiber with a 1:1 optical setup.

---

Definitely something I didn't think about. My only worry is they're harder to find and more expensive, but it looks like they're typically well made.

Quote:

---

Originally Posted by mixedgas

It might be cheaper for creating PCBs to just use a x-y motion stage and a microscope objective to focus the laser.

---

Yeah, which is what he did in the article. But as you can see from this blog post, hysteresis becomes a big problem. While you can get around it to some extent with software and basically eliminate it with pair of linear encoders, it's more expensive and more work, and if I can get accurate closed-loop galvos then I could just avoid it completely.

Quote:

---

Originally Posted by mixedgas

You can then use a UV diode with photo-resist coated boards

---

Hrm, I actually didn't think of this. However the main reasons I want to move away from the UV-sensitive paint are the storage/aging problems I've had with it, and how long it takes to get (have to order it online) and the expense. I can get black paint locally and cheap, and I don't have to worry about storing the prepainted blanks in a dark area, etc.

Quote:

---

Originally Posted by mixedgas

Take a look for Cambridge Technology 6450s or 6860s or similar on Ebay, you might like what you find. Too big a mirror for most laserists, and very precise. You will need position sensing amps, unless you use G120DTs or or G330DTs which have torsion bar centering.

---

Thanks for the suggestion! I'd looked at some more scientific (as opposed to "high speed laser show") galvos and most were several grand. These on the other hand are still in my price range.

I'm not so sure your example blog used the best available cheap technology.

I work with capacitive sensors with nano-meter precision once in a while. its not hard to measure at femtofarad resolution with op-amps and a 10 Khz triangle wave excitation. What I'm saying is hobby feedback schemes never seem to match even the simple stuff that is available cheap. That would let you ditch the belt errors, which are horrid in those photos.

Example: Interpolation chips for encoders are 9-20$ each, and that knocks off the hysteresis and gives you a 16X or 32X boost in resolution, as well as storing the current position, glitch free.

Blue 1 watt laser diodes have horrid spot size, and the spot is oblong. This is true of any diode to a point. You'll need some means of correction for this. You can find single mode ones once in a while, at say 25-75 mW.

Steve

Quote:

---

Originally Posted by Orum

Yeah, which is what he did in the article. But as you can see from this blog post, hysteresis becomes a big problem. While you can get around it to some extent with software and basically eliminate it with pair of linear encoders, it's more expensive and more work, and if I can get accurate closed-loop galvos then I could just avoid it completely.

---

That is not hysteresis, that is backlash and lost motion. Hysteresis you might see in a servo loop, not in this. Backlash is just caused, in his case, by poor mechanical design. Use good linear bearings and precision ground ball screws and you will be good for fractions of a thousandth of an inch. Heck, even rolled ball screws are typically accurate to .003" in a foot. In an light weight system like this you can even use timing belts, but they must be tight, I mean TIGHT. Play music on them tight.

I will come out and say it the mechanical design of a reprap is crap. A reprap was made to be reasonably cheap to make and uses cheap parts. Using hardware store all-thread and acrylic does not make a rigid or accurate machine. Garbage in = Garbage out.

I say ditch the idea of using a laser diode, they suck for this kind of application. Get a cheap chinese CO2 tube and power supply and build a machine around that. If you use blue you need an opaque box to put it in. Most black acrylic is actually transparent to NIR laser diodes so you would have to build the box out of metal with that or find a specific plastic that is opaque to NIR. With CO2 almost everything is opaque to it. You could build the entire box out of acrylic and the beam could not escape (I still wouldn't, though), well, not till it burns though. It takes odd-ball materials to be transparent to 10.6u like germanium, silicon, and GaAs.

A chinese laser tube is cheap! About $120 for a 40w tube and another $120 for a power supply straight from china. And they are easy to run. Dont need to worry about static killing a diode. The power supplies are designed for PWM control too.

Check out the guys over at buildlog.org. A lot of people are building their own laser cutters and he has an open source design for one. Most of use use mach 3 to control it.

I have built and retrofit many CNC machines, some pushing almost a couple tons and use 1kw+ servos on all axes. I have a couple sets of galvos that I want to use for marking so I am in the same boat as you. But I dont think you want galvos for this project. To get even a decent sized board you are going to need a lot of deflection or distance between the head and the work which means it is going to be way out of focus at the edges. You wont have this problem with a well built x-y platform. Galvos are for speed which you wont be able to use with a little 1 watt blu-ray diode.

This is my solution when I need boards: http://oshpark.com The boards are incredible. Made in the USA, double layer, gold plates, purple conformal, silkscreen for both sides. $5 a square inch and you get three boards for that price.

... I was building and refining/refurbishing differnt types of CNC-mills and laserplotters, have one of this chinese CO2-laserengravers sitting on my workbench, waiting for an alternative electronics/controller.

For laser-engraving/cutting I'm actually using my mill with different laserheads, but have plans to equip a really solid/rigid flatbed plotter with lasers - attached some images of the plotter and a 445nm-diode and the optical head of an IR-fiberlaser.

Attached too images showing the absorption rates for different materials and wavelengths ...

VDX, is that a diode pumped fiber or just a regular fiber coupled diode?

Thanks for posting those charts. They are what I have been looking for a while. I can see why my q-switched 532 yag is used for marking gold and even though my 532 is more powerful than my 355nm laser the 355nm might be actually better for steel and stainless.

But in the end, even 9W CW is not much to do anything even with the extra absorption for some materials at the shorter wavelengths.

... the bigger 'head' is the end of a fiberlaser pumped with around 160Watts @975nm and emitting up to 50Watts @1070nm --- it's focussable to <10 microns, so good enough for cutting thin steel sheets or engraving all sorts of materials ;-)

Have fibercoupled 25Watt-diodes too, but with a 0.1mm-fiber only focussable to 0.1mm best :-/

Viktor

Quote:

---

Originally Posted by VDX

... the bigger 'head' is the end of a fiberlaser pumped with around 160Watts @975nm and emitting up to 50Watts @1070nm --- it's focussable to <10 microns, so good enough for cutting thin steel sheets or engraving all sorts of materials ;-)

Have fibercoupled 25Watt-diodes too, but with a 0.1mm-fiber only focussable to 0.1mm best :-/

Viktor

---

Now that sounds like fun! That must have cost a small fortune.

I have a bunch of the 50w FAP modules with 800um fibers. I am hoping I can use them to drive a Spectra Physics T70 head when I can find the right cables.

Quote:

---

Originally Posted by macona

Now that sounds like fun! That must have cost a small fortune.

---

... the OEM price was something like 30-40k€ then, but got it in exchange for some spare parts for PCB-generating and microtech I had in my basement.

I'm in some different groups, where we're exchanging spare tools and bits someone has sitting around ;-)

Most of my 'more valuable' parts were sourced for project developments I've made from 1990 on --- and they often remained after the project was finished as 'sold prototypes' ... and I've got them for free or had built them on my own for the special project - for example a microassembly stage with nano-XYZ-stage, gripper and stereo-cams which I've made by 'refurbishing' a SMT-pick-n-placer (see attached image)

Others, like my CNC-mill, were bought for cheap (300€) as scrap glue-dispenser from a customer of Isel and then reworked/refurbished by me to a working mill ...

I've got recently an older dental-scanner and all the inner parts of a second for testing it's usability or for reworking it to something usefull --- so some more fun ahead ;-)

Viktor

I know what you mean, I get a lot of my stuff the same way. The two DPSS lasers I have were something like $50k new. I traded time for them. I did swapping for my CNC mill and for a whole lot of other stuff.

Looks like that laser could be really useful!

... my greatest problem is lack of spare time :-/

Have all the guts of an old Stereolitho-3D-printer from 3DS without the (then used but outgassed) 30mW-HeCD-laser ... but here I've made successfull collimation tests with a BluRay-diode with 405nm@200mW and a blue diode with 445nm@1W, what's more than enough power to 3D-print or even micro-material-processinq! As the optics and mirrors aren' t coated, it could be even possible to use the 1070nm@50W fiberlaser with the scanner and optics ...

Another actual project is modifying DLP-beamers for use with the 405nm-diode or with a TAE-laser pulsing with 30Hz, 371nm@400kW and 200ps pulse-duration ... maybe something usable for microetching?

And some old projects around 3D-printing/rapid-prototyping with higher accuracies than common - laser-spots <10µ with the IR-fiberlaser, or refocussing the UV-pulse-laser over the 'naked' DLP/DMD-chip into an 'image' of some mm size ...

Viktor

VDX, was this a liconix HeCd laser? I'm looking for a dual cathode supply from liconix for tubes like the 4030, desperately

.. don't know it - a friend got two of this printers with removed lasers and I've grabbed only the transportable essential parts - attached the complete part before disarming ...

Neat. I have been thinking of using my laser cutter to try stereolithography. I will have to work top-down like a traditional STL machine which means I need more expensive resin. Of course I think 6 watts average might be a bit overkill:

http://www.youtube.com/watch?v=W071IQhpCFA

Quote:

---

Originally Posted by macona

I say ditch the idea of using a laser diode, they suck for this kind of application. Get a cheap chinese CO2 tube and power supply and build a machine around that.

---

I don't want to use CO2 for numerous reasons, but mainly safety. In addition I imagine it's harder to focus, and it just provides more power than I need. I'm sure they're useful if I wanted to driectly burn the copper off, but I'd rather mask with paint and then etch.

Quote:

---

Originally Posted by macona

This is my solution when I need boards: http://oshpark.com The boards are incredible. Made in the USA, double layer, gold plates, purple conformal, silkscreen for both sides. $5 a square inch and you get three boards for that price.

---

Yeah, I use Advanced Circuits when I need multiple copies. The purpose of building this was two-fold; to reduce cost and latency when I'm making a one-off board, and more importantly, to reduce cost and especially latency between when I finish designing a board and when I can actually test a prototype. Designing circuits is like programming or anything else of sufficient complexity--put it down for too long and it takes longer to get going when you start working on it again.

I've been looking at the Cambridge Tech galvos, and while they're pricy they seem to be designed for my application unlike many of the alternatives. The only problem is, I can't find much documentation on them. Many of them come with the driving boards, but how do I interface with it? It looks like it has some proprietary (or perhaps just unknown to me) connector. Their website seems to be dedicated to newer hardware, and it's hard to find out much about the older stuff that pops up on eBay.

Quote:

---

Originally Posted by Orum

I don't want to use CO2 for numerous reasons, but mainly safety. In addition I imagine it's harder to focus, and it just provides more power than I need. I'm sure they're useful if I wanted to driectly burn the copper off, but I'd rather mask with paint and then etch.


Yeah, I use Advanced Circuits when I need multiple copies. The purpose of building this was two-fold; to reduce cost and latency when I'm making a one-off board, and more importantly, to reduce cost and especially latency between when I finish designing a board and when I can actually test a prototype. Designing circuits is like programming or anything else of sufficient complexity--put it down for too long and it takes longer to get going when you start working on it again.

I've been looking at the Cambridge Tech galvos, and while they're pricy they seem to be designed for my application unlike many of the alternatives. The only problem is, I can't find much documentation on them. Many of them come with the driving boards, but how do I interface with it? It looks like it has some proprietary (or perhaps just unknown to me) connector. Their website seems to be dedicated to newer hardware, and it's hard to find out much about the older stuff that pops up on eBay.

---

I would say CO2 is safer than anything else. Like I said you can use just about anything to shield from it. A generic face mask or safety goggles are all that is needed for eye protection. Also it is virtually impossible to get retinal eye damage for a CO2, the wavelength can not penetrate into your eye, think of the beam as a beam of heat, not light.

You state you want a window on the machine, this is pretty much impossible on a visible laser with most standard materials. You would have to have a window that it either made from a plastic specifically formulated for the wavelength you are using or coated to filter it out.

CO2 is very easy to focus. Even the cheap chinese tubes had decent beam quality and make a nice small spot. Typical kerf (Line width) on a co2 laser cutter is about .002". A simple GaAs plano-convex lens is all that is used to focus. There are chinese made scanlenses for CO2 on ebay too. A diode is just going to have a crummy beam.

We had a 45 watt epilog where I was working and there were a couple guys doing the spray paint resist technique with the laser cutter. The power definitely came in handy to make sure you got a clean cut in the paint. It is very difficult to cut copper with a CO2 laser since it reflects so much of the light. Heck, they make mirrors for CO2 lasers from polished copper.

On the other hand I can burn holes in copper with my 532nm laser since it has a shorter wavelength that is absorbed by it.

Which CT galvos are you looking at? The standard CT galvo will have a 9 pin D connector on it. There are some custom connectored CT galvos that were made custom for ESI that use a 9 pin D for the coil and a 9 pin micro D for the feedback. ESI uses a separate control box that drove external servo amps to drive the galvos and the box itself closed the loop.

Typically most galvo drivers use an analog interface, either +/-5v or +/-10v. Voltage is deflection and polarity is direction. There are a few different controllers out there. CTI makes one and I have seen them at $1k on ebay. There is a chinese company that makes a board called the ETH6608, it is a standalone ethernet board that will control two galvos. Comes with software to interface with it for $650. You can also use the Etherdream with the http://www.beamconstruct.com software, though the support is discontinued and no new features will be added for it in future revisions.

There are some scanner drivers that are digital, those need one of ScanLabs RTC boards, they sell at well over 1k on ebay.

Have you though about just doing isolation routing with a dremel/rotary spindle mounted on a cnc x/y stage? That's what a lot of the in-house prototype machines do.

Orum, I have a lot of the older Cambridge documentation. If you have the actual article in hand, and give them the serial numbers, support usually flows.
Support will be in the form of hookup data and the tuning/testing manual. Manuals are usually about 20-30 pages and are very intensive/ hard core. Once in a while schematics do come, but not often. Once in a while something shows up on Ebay that they will not support as its a custom driver. However you will find much data in the legacy portion of their website.

Galvo amps are not hard to hook up, you just start at the power output stage and work back.

If your a beginner, buy the whole set, amps and galvos. Building your own galvo amp has some interesting pitfalls. That one I linked down below for 125$ will not be hard to reverse engineer the hook-ups. Laser show guys use high speed galvos, and that set is a high precision positioning set, not known for its speed. That is why its there cheap, the majority of PLers would thumb their noses at a 4K PPS scanner vs a 30K pps scanner. Its perfect for you, its accurate. Odds are good it will have a "position valid" output as well, telling you when you can fire.


Steve

Quote:

---

Originally Posted by macona

I would say CO2 is safer than anything else. Like I said you can use just about anything to shield from it. A generic face mask or safety goggles are all that is needed for eye protection. Also it is virtually impossible to get retinal eye damage for a CO2, the wavelength can not penetrate into your eye, think of the beam as a beam of heat, not light.

---

Well, that's precisely my worry--I can't see it, and at >= 200mW even if it doesn't penetrate the eye it could burn/ignite things without me realizing.

Quote:

---

Originally Posted by macona

You state you want a window on the machine, this is pretty much impossible on a visible laser with most standard materials. You would have to have a window that it either made from a plastic specifically formulated for the wavelength you are using or coated to filter it out.

---

I was planning on getting a window specific to the wavelength of the laser I buy. Problem is, these windows are hard to find, and typically for longer (IR) wavelengths. When I do find them for visible lasers, they're fairly pricy. But I haven't looked as much for these as I have at the other components yet.

Quote:

---

Originally Posted by macona

A diode is just going to have a crummy beam.

---

Yeah, I realize they're a bit of a compromise. The smallest etched areas I get down to are TQFP/LQFP packages, with a spacing between pads of ~0.24mm. I think that will be achievable with a diode laser, but if it doesn't work out, they're not that expensive. I'm sure I'll find other uses for a high power diode laser elsewhere.

Quote:

---

Originally Posted by macona

Which CT galvos are you looking at? The standard CT galvo will have a 9 pin D connector on it.

---

Really anything that's accurate and in my price range (< $200 USD for a single galvo + amp), but mostly 6450s or 6860s as Steve recommended. While the galvo itself has a DB9, the amp they're sold with has the odd connector (bottom center of this auction's image. It also has a white wire pair coming off--is that for the analog input? If so, that would be a lot easier to use, but I'm still curious what the other plug is for. While digital galvos are nice as you said the price rules them out.

Quote:

---

Originally Posted by macona

Have you though about just doing isolation routing with a dremel/rotary spindle mounted on a cnc x/y stage? That's what a lot of the in-house prototype machines do.

---

Definitely. However, hysteresis--I know you called it backlash, but I would define hysteresis as the property of a system's state being dependent on the previous (historical) states or knowledge of the internal state, and backlash as the specific cause of the majority of that hysteresis in this case--concerns me, especially when you have physical contact between a dremel bit and the board. The drill has to be very securely mounted, as does the copper blank. Lastly, while it lets you skip the chemical etching, it's a double edged sword; there's no way to proof your copper blank before you do something irrevocable to it.

Quote:

---

Originally Posted by mixedgas

If your a beginner, buy the whole set, amps and galvos. Building your own galvo amp has some interesting pitfalls. That one I linked down below for 125$ will not be hard to reverse engineer the hook-ups.

---

Oh yeah, definitely looking for the galvo & amp package. Some people go all out and even build their own galvos as well, but that's well beyond my ability for the accuracy I need, and just isn't practical.

I finally found the "Product manual archive" on their site, which I had glossed over before.

Looking at the auction pics, I see the three white wires are the power leads for the amp. That little daughter card is actually a breakout board, looks like it might include the IR laser and certainly includes a laser driver. The two lenses pictured probably make a reasonable first order flat field correction, taking the place of a F-Theta. My bet is that is a lens with a LD under the U shaped folded piece in the silver can. Either that or its one wierd power connector with only two wires.

The amp pictured is easy to reverse, the three VRs will be 7815, 7915, and usually a 7805 or 7808. The Apex chip is the power amp to drive the galvo. The potentiometers are labeled, I just can't make them out in those pictures. So they are probably servo gain, velocity, and damping, which is all a marker system needs.

Piece of cake from there.

Steve

Dont worry about the beam being invisible. There are dozens of people that have built their own CO2 laser cutter and have had not problems aligning it even though the beam is invisible. I personally use small pieces of acrylic to find the beam, others use thermal paper. There are even phosphorescent sticks that glow in the beam. And once you are wearing high OD glasses for the diode you wont be seeing much of the beam anyway. I have set up and aligned lasers in both UV and IR and it is not hard.

Calling backlash something it is not does not make it right. Anyway, backlash can be compensated for, all the major CNC packages (LinuxCNC, Mach3) have backlash comp built in.

I built a little board router for a friend once. Uses pretty generic plastic nut anti-backlash leadscrews. The table is a piece of MDF that I mounted rare earth magnets under and use other magnets to hold the PCB down. It does not move, the actual cutting forces involved are very low.