Miscellaneous
Information
Just a bunch of interesting links and documents that I found were related to this project.
http://laser.shows.org/scanners.htm
http://sound.westhost.com/amp_design.htm
http://www.motionsystemdesign.com/archives/14_4329.pdf
http://www.motionvillage.com/training/handbook/resources/2000_08_pid_pos_loops.pdf
http://repairfaq.ece.drexel.edu/sam/laserdps.htm#dpsadl3
http://www.geocities.com/CapeCanaveral/Lab/3931/laserdps.htm
http://www.laserfx.com/Backstage.LaserFX.com/Systems/Scanning3.html
http://perso.wanadoo.fr/redlum.xohp/electronics/315M/Compass315M.html
May 9, 2005: Added various closed loop schematics I found on the Web. They are not very clear, both visually and conceptually (!) but still, they may help understanding how feedback loops work, and see how different engineering teams have chosen to implement it.
Of course, Chan's version - the one I am working with:
Updated Oct 5, 2006 - now using version 2 of his design (without feed-forward).
Updated Oct 5, 2006 - Optical PD details
My replacement circuit of the RF PD with Optical PD (OPD)
Principle of operation
A textual description would be nice to have here, I know.
Other schematics of galvo drivers...
A660
Unknown
Tropel
GS-A102 (open loop, but w/current feedback)
Interfacing to a PC's parallel port (LPT)
Also, a simplified schematic of my own version of a dual channel 8 bits DAC. The one you see on this site.
Updated January 31, 2006
Here is another popular DAC circuit that many people from this site asked me to support within my SW. So I did.
DAC0832
Here is a PDF document that shows how to interface the
amplifier to the PC's parallel port.
May 25, 2005: Answer to that frequently asked question: Where do you get your small mirrors?
Here is a nice source of mirrors for your galvos. I discovered it by pure luck, as I was searching for flash circuitry, long time ago. You need what is called a first surface mirror, i.e. where the chrome (or most likely aluminum) plating is applied to glass, but left exposed to air and not covered by protective paint like ordinary mirrors you'll find in stores. Anyways, the mirrors you'll find in stores are way to thick and heavy for galvo application. You need speed, remember? The good ol'Polaroid camera is there to save the day and many bucks too. They are dirt cheap in garage sales all year round, or in thrift stores. You can usually buy them for a few bucks. The model Sun600 was very popular in the 70s and 80s. If you see one, or a cousin model, jump on it. Actually, there's a mirror in there almost for sure, whatever the model.
Here's the Sun600, a classic:
What is a giveaway for sure that our mirror is there, is the slanted back at 45 degrees.
What a funny guy I can be! Look at those dirty hands manipulating such a nice mirror.
Nothing like fresh motocross grease for my galvos...
You can see the mirror is you open the bottom.
Not really visible here, but it's only to show you where to look.
That's really thin. Just what we need. About 1mm.
Normally, you'd pay quite a few bucks for these, if you bought them already made. But the tradeoff here is the work you have to do to get them the way you want. I mean you have to cut the big mirror into smaller pieces. The challenge is to end up with two very small mirrors, and the same size too. If they are not exactly the same size and shape, some phase error may occur. Your circles will not be perfect, but more elliptic because speeds will be different for each axis. I'm not sure to what extent this is critical, but just try your best when cut them. Actually, I have never been able to cut these. Every time, I end up performing what I call "controlled-smashing-of-delicate-optics" (!). Then I pick up the pieces that look roughly the right size and grind them with the Dremel until I'm close enough. Even grinding with a fine wheel is very difficult. I glue them using a drop of superglue or epoxy for the very final version.
Updated Oct 5, 2006 - (BAD IDEA!!!! - do not waste time with this)
I almost removed this crap, but I leave
it there so others see what I've tried.
May 5, 2005: Another attempt at obtaining position feedback : going optical
After spending too much time fixing the ground on the shaft, and getting shaky readings, I decided to try a different method of getting the position of the shaft. I've read that today's galvo use some kind of "optical" method, but nowhere I can find exactly how this is done. I decided to try my own variation, using a transparency gradient on a thin disc, through which light would be passing and being detected on the opposite side, then converted to a voltage. In other words, vary the light's intensity with a rotating disc. Of course, the big challenge here is linearity. If it is not perfect, the graphics will be distorted. And while creating a radial gradient that is linear is supposed to be simple, the final output still depends on so many variables that it will likely be hard to create such a disc. I decided to give it a try, and here are my results.
The base image is created from Photoshop. I tried two versions, 60 and 90 degrees.
These would fit on a 8 X 10 sheet, in case I want to try printout snapshots later.
This is a 60 degrees radial gradient.
This is a 90 degrees radial gradient.
More experiments:
With these two half moons, maybe I'll try to see if they could be used to create and "angular shutter", like scissors. One is fixed, the other rotates and creates a widening gap through which light passes.
A 360 degrees radial gradient
Original Photoshop files:
These files are mathematically perfect, but transferring them into our real world will screw all that up! We can only try to do our best to keep it as is. First, I tried the obvious : take pictures of the CRT monitor of my PC. I took care to calibrate in order for the gradient to show all its nuances, including deep black and bright white. Then, using a tripod for stability, and a 35mm camera I took pictures of the screen in full darkness. As a first test, I decided to leave the camera in auto mode because I had no idea where to start in terms of shutter speed of aperture. The film was a 200 ISO, Fuji, color. This is what I get when I cut the negatives:
Note that this is a 360 degrees version that I tried because I really wanted to see the full effect of the gradient, and better evaluate the linearity.
When I combine two discs, but invert one of them, in order to see what the complement would be, I get uniform transparency as I expected. This is good news, because it means that linearity is good.
This is the opposite test, where the two discs are "in phase", i.e. enforcing each other. This doubles the effect of the gradient.
An optical switch such as this one will be used to monitor the output.
When I tried this, I could see the saw tooth I wanted, but the amplitude was very small, even with the combined discs. The reason is simple. My film has been underexposed and is not completely black where it should.
Also, another more or less expected problem I has on the negative was the interference between my monitor's refresh rate and the camera's shutter speed. This left artifacts on the film that looked like angled stripes as shown here:
In conclusion
This seems to be a solution that would work, but the film requires more contrast to get the maximum amplitude. I tried to print the files to a laser printer at 300 DPI on a 8.5 X 11 sheet, and the results were very good. I used the JPGs included above, without "fitting to page" as many drivers offer. They are already calculated for that paper size, so I didn't give the drivers a chance to screw up all the quality I put in them. In the next days, I will take pictures of these images under good lighting conditions, and with different "F stops" (apertures) using the manual mode of my camera. This way I will get over and under exposures and I will be able to test which is the best. I will note the type of lamp, wattage, distance to sheet, lens type, etc. All to make a recipe where I will be able to redo it again without all the research I am doing right now.
May 5, 2005: Power supply problems
During the weekend, I worked on the project with my cousin, and we realized at the end of the day that we were defeated by some very strange phenomenon. Whatever we did, as soon as we drove the galvo a bit, we lost it's control totally. We checked everything dozens of times, desperately trying to figure out what the hell was going on. I found later the next day, that when some power was being pumped in the coils, the power supply was unable to keep up with the demand, and became a too low. When the regulators are fed with a voltage below 19Vdc (15V + 4V min differential), this has the effect of creating wild fluctuations at the outputs of the +15 and -15 volts regulators. These provide power to the logic of the circuit, so it explained why the behavior changed so drastically under these conditions. It became very obvious as I decided to put the scope on the regulator's output!
Solution: Build a second power supply, just to the power output opamp. The first one will remain as is for the logic and both will share their ground connection. I decided that I might as well build a "real man's" power supply. Here it is:
40000uF !!! Twice, please. With a 40Vac / 10 amps transformer (w/center tap)
Poor little one in background can barely fill daddy's shadow.
But since I added this big guy, my galvo has gone crazier than ever, even at low power. Maybe I need some regulators here too. Will test this later.
Update (May 9) : was because my PD was too close to the heat sinks and was influenced by stray fields (see Version 5C page).
Here is a technique I have found to easily create and secure the coils.
Of course, this is just sample with a few turns only, to show you. The idea is that the 4 pins are soldered underneath, and when the winding is over, I will bend them in a U (but inverted) shape and solder the other end to secure the coil in place. The pins are placed in such a way that they barely clear the magnet, so they get as close as can be.
Updated June 6, 2005
An easy oscillator circuit to replace the hard to find EX03-16M
Note that using this circuit allows tuning to the PD optimal frequency, by swapping the crystal. Those are much cheaper. Output can reach an amplitude of more than 13 volts when chip is powered from 15V. Ok to use any fast cmos 4093 equivalent.
See the output here, just before it enters in PD.
Note that I modified the filter at the PD's output. This removes completely all the residual 10Mhz I had before mixed in my PD signal. This caused my signal to look more like a thick band instead of a clean line on the scope. Of course, I always use the 20MHz BW limiting feature on the scope's channel input, but even then...
So, by moving the 220uH inductance "sooner" in the filter, i.e. closer to the PD, I realized that it removed much more RF, and and also prevented the gate from being loaded. Now, when I connect the oscillator to the filter, the output's amplitude of the gate doesn't go down at all.
In case you need it, here's the older version
