Any help for figuring out how to tune a WonSung galvo set?

[jsibre]

Hi,

I've been a bit of a lurker here for a few weeks or maybe a few months, but today I had cause to create an account so I could ask if anyone knew what I had on my hands.

Short version
I bought this WonSung 23/30kpps galvo set on eBay (it's also on Amazon UK and Aliexpress).

Here's a good pic (from eBay listing) of the drivers.


You'll note there are no model numbers on the driver PCB (I've removed a heatsink and checked the backside. None there either.)
You may also note that other images from WonSung's site, and the sale listings show different boards. The ones I have look like the image above. One group of 2, one group of 4, and one loner (pots, that is).


The Question
Does anyone either:

• Know where I can find docs for this thing?
• Or, know which pot does what?

The listing describes it as

Potentiometer description

• IS Input scale (adjusted only in factory)
• SG Servo gain (power of the feedback signal for internal PID controller
• LFD Low frequency damping (correct overshoot)
• HFD High frequency damping (correct undershoot)
• LIN Zero offset (electrical offset of the driver, adjusted only in factory)
• PS Position scale (increases or decrease input sensitivity of the computer, DO NOT change it)

But it doesn't indicate which is which, and the astute among you may note that six pots are described, but there are seven in the image. I tried assuming they were left-right, but probably got myself in trouble. (After all, in which mounted orientation???)


Here's the TL;DR version
I bought this WonSung 20kpps galvo set about a year ago (Jan, 2016), thinking it would be fun to mix lasers into my microcontroller hobby. But for various reasons (Kids, mostly?) it sat on the back of my desk till about mid-November, unused. I then hooked it up and started playing. Very cool
I'm not using the ILDA inputs, or even the show card - I'm driving the galvos from an Arduino Due (after pushing the output through an amp to convert the DAC's 0.5-2.7v to a +/-5v, bipolar signal).

At one point, I figured out how to parse an ILD file. Not on the Arduino itself, but on my desktop to generate code that would allow me to (somewhat laboriously) display ILDA files (such as the test pattern) on my rig.

So, of course, I did just that with the 12/30k ILDA test pattern ('A' version).

I was quite impressed at how well tuned it was 'out of the box'. Oh, it was far from perfect, but it was a lot better than I expected.

Then, in anticipation of some new hardware (I ordered a 500mw RGB laser, and had a PCB custom etched to clean up some of my dangling amp circuitry), I remounted all my goodies onto a new 'chassis' (piece of 3/4" plywood). Somewhere in that process, I messed something up. Maybe I banged one of the galvos, or maybe I bumped one of the pots. I'm really not sure, but it just doesn't look like it used to, and tuning it went from "maybe someday, 'cos it ain't awful" to , "Seriously. Needs fixing. It's screwed up, and things that looked cool, now look wonky." At this point, I'm still using the same kit I've had running for several weeks, but on a new piece of wood. (Not using the custom PCB or new laser yet. My laser is a currently a hacked dollar store laser pointer.)

So, I've found some tutorials and videos, and I'm eager to try (er... I've already tried, after making some guesses about which pot MIGHT be which.. didn't go well, but didn't go pear-shaped either.), but they all (quite reasonably!) assume you know which pots are for servo gain, LFD and HFD, at a minimum.



So, that's where I am. Spurious imported product of dubious origin, and no documentation on it, other than what's on the sales listing (which I don't think has changed much in 11 months). Anyone know which pot does what?

I've spent a few hours digging around the web without finding what I'm looking for (but of course, I don't know the scene that well, so don't really know where to look).

Don't be afraid to tell me that what I have is crap, and that I should buy x, y, or z, instead of messing with what I have. I'm willing to consider that - but at this early stage in my hobby, don't want to shell out too many hundreds of dollars on something I may bork fairly naively. Anyway, suggestions are welcome, as well as questions. But, at this point, I'm still keen to see what fun can be had with the microcontroller in the equation (vs running 'normal' software on a computer sending ILDA data to a card that replaces my Arduino's role.

Oh, I'm in the Houston, TX, USA area (I see a lot of posts where the answers seem to depend on the person's location)

Thanks for any pointers,
Jason

[mixedgas]

Can you read schematics? If so I MIGHT be able to help.

Steve

[jsibre]

Switched to desktop site to try and upload picture...


Also, I've switched over to the new custom PCB for amplifying the Arduino output at this point.

[mixedgas]

This link might help. It does need a scope for some portions. Almost all low cost scanners are clones of the Cambridge device shown here. Yours is a reduced parts count version with lots of changes to the power output stage. However I'm sure the position sensor portion and signal input will be copied exactly. That lets you find the scale factor and linearity pots based on the schematics. Your amp has no Polyfuse, fuse holder, or output jumper to lift the drive to the galvo coil, so that makes things tricky. I cant see where they brought out test point one or TP two for that matter, anything to save a few cents.... Section 7.1.2 is what I'm referring too...
~
As you do not know the position sensor scale factor for your galvos, leave that part alone. Changing scale factor more then a few percent will result in a complete retune. Changing it too much can take you into a region where you cant tune well at all.
~
You might also have an issue in needing a "clean" known source of the ILDA test pattern.... Your best hope without a Pangolin or LSX is to get in the PL Facebook page and see who is near you with one. I know one guy in Houston who might lend you some time on a QM system for tuning. I'm trying to find Dave's number, but I'm at work.
~
I don't know how your system adds dwell, guide, or anchor points, if it does... Hence you need to be careful if using the ILDA test pattern.
If your DACs are not double buffered, updating at exactly the same time, you may see some skew in the images, or have real trouble with the circle in the square.
Beginners often try to tune at too wide a scan angle, too.
~
In that case, if you suspect your test pattern is off the reservation, revert to the Laser Media test pattern till you get close, my sequence is Quadrature Square, ILDA, Grid, LM, repeat...
The LFX patterns are fine... Its just your software may not follow some traditional steps when dealing with the output waveforms.
~
Make just small changes, or you might find yourself very unhappy....
`
http://www.skywise711.com/lasers/scanner/scanner.html
`
Skywise is kind enough to host that for me...
`
The above factory initial setup is a laboratory scanner amplifier procedure, it does anything but get you a clean ILDA pattern... In fact it will turn a good ILDA tuning into a very fast rise time tuning that you would use for a lab or factory scanning setup. However it is the best cold start procedure (Well, its the ONLY one!) when your lost and have to come up from zero. Ignore the parts for the old G120 amps, the document is blended...
`
Easiest way to quickly set the linearity is to attach a true RMS voltmeter while projecting a centered, symmetrical test object and adjust for least RMS AC volts at the feedback sensor output.
Its not the best way, but with what you have to work with, its a start. To do it well requires a digitizing scope.
`
If I'm sure I'm using a meter that will not kill opamps, I measure the potentiometer resistances before I make big changes. (system off, of course!) Because they are "In Circuit" , you'll see all sorts of non-linearity on the Ohm Meter.
~
I'm jammed up today, but I can help later in the week. Find Buffo's tuning procedure here on the forum, its not my favorite way, but the "three legged stool" method works well for beginners...
~
Finding Eight Degrees for the ILDA pattern:
https://pangolin.com/userhelp/scanangles.htm
~
Steve

[jsibre]

Gah! I accidentally deleted my previous post while trying to edit it to correct some typos I noticed (beware the mobile version of the site!) - I was able to copy the text on my phone, will re-enter the message for posterity when I get home, by appending to this one.

Here's the message that was in between 3 & 4 above
-----------------
Hi Steve,
Thanks for the response.

Yes, I can read schematics. I'm no EE, but I can generally make them out OK.

What do you have in mind? (I don't have schematics for these boards- but maybe you do for some board where the pots are known?)

Also, I have made some progress since my first post. I've tentatively identified the six pots in a row, and while I'm not positive about two of them, I'm reasonably sure about the other four.

I'll try to attach a picture later - I can't figure out how to do it from the mobile interface to this site.

That said, I still have concerns, because even knowing (100% sure on gain and LFD, 80% on HFD) which pots do what, I'm having trouble tuning. I have certainly hosed up some of the other settings, and don't know if they may be related to my issue.

So far, after two tries, I can't get the circle centered in the center square. I can get it round, and just barely touching on the right and bottom, but then the top and left are approx 5% shy of touching. If I tune to get them touching, the other two sides extend beyond the square's sides. (I'm only using a single laser at the moment, so it can't really be an RGB alignment issue.)
It kind of makes me wonder if the test pattern I'm using is not quite right. (I got it from the laserfx web site.)
I don't have a pic of that at the moment. But I can get one later if that helps.

Jason

[jsibre]

Steve,
Thanks for the detailed response!

I'll have to provide a more thoughtful reply later, from home, but quickly:

• I do have a digital 'scope. Nothing too fancy, but seems adequate for this task.
• "I don't know how your system adds dwell, guide, or anchor points, if it does..." If by 'system', you mean my Arduino and the software it's running (I.e., not the galvo drivers, galvos, etc), it doesn't add any of them. I don't know what they are (though the names give me some hints).
  That's sort of a topic to itself: Not knowing how ILDA patterns are meant to be processed, and armed only with a link to a document detailing the file format, I naively just pull out the coordinates and the blanking signal, and turn them into data that I used to manipulate two DACs and one digital output. It attempts to output the points at the correct cadence (I've been using 20kpps, knowing the pattern is intended for 12/30k, but not really sure what that actually implies.) Also naively, when I see the blanking signal is high/true, I 'blank' immediately (it's a dollar store pointer... there's no blank, I just cut power to it), and then move to the coords associated with that state. I'm not sure if that's right (vs going to that blanking state after arrival), but assume from the output I'm seeing that it's correct-ish, at least. That leads nicely to your next point, which was a face-palm moment for me...
• The DACs are NOT buffered. At one point I was considering that I might have been better off buying a separate DAC chip with a sync signal (rather than buying a 'duino with dual 12 bit DACs - true, not PWM) for that very reason. As currently programmed, I'm outputting X first, and then Y (or vice versa - need to double-check the code). I'll just bet that's key to the off-center issue. I'll try reversing the sequence and see if that moves the circle. I'm not sure what the delay is between one being set and the other, but it's on the order of microseconds, but, considering I'm only adding a 29-30uSec delay to each point to achieve 20kpps, I have to assume that a significant part of the remaining 20-21uSec is being spent on each DAC write. Maybe 5-10. So, it could totally skew the circle as I'm seeing if they aren't happening as intended.
• Not sure I completely follow what you're saying about setting the linearity (or where I find the feedback sensor output, but guessing some time spent reviewing the document you linked will help with that.
• Actually, Buffo's doc is what I've been using so far I have an ILDA Test Pattern file, I've faked my own Quadrature file, based on what the image showed (entered coords, ILDA-style, then run through my showIlda() function), but I have no idea if I got it right - I only entered coords for the corners, not sure if there should be intermediate points as well. It also doesn't look perfect, but it seemed to serve the purpose of doing the first bits of the tutorial faster than the ILDA Test Pattern. I haven't got a copy of the Laser Media Test file, and haven't tried to make one.

That's what I have for now - I'll repost the other post later, and read the links you've sent tonight.

Thanks again,

Oh, and an aside: In learning about the file format, I learned that it supports 3D. I originally thought maybe that meant for 3D visualizations, but wasn't sure how that was meant to work. The whole thing kinda made me cock my head to the side. Fundamentally, it's still 2D, even if the artist makes it 'look' 3D... I just learnt yesterday that some galvo systems include facilities for controlling the focus point of the laser as well. Is that what the third dimension is actually used for?

Jason

[mixedgas]

I'll get you definitions of the point types. I need to scan in a document that is not on-line anywhere. While a 3D frame is ultimately projected as a 2D calculated object, storing it in 3D has certain uses, including adding foreshortening, real time manipulation, etc.
!
I'll get you a test frame bank file. I keep one on a USB stick at home.
!
If your scope had at least 1 Mhz of bandwidth, you can see what you need to see. After all, max frequency content of a 30K test frame is a bit over 2.5 KHz... At 30K, you can see subtle errors in the DAC updates, but your close enough for now.
!
Locate and bring out TP1 (Position Sensor Output) if you can, its very useful. TP2 is ground anyways. It also helps to look at the Galvo current, which is at the high side of the 0.1 Ohm output current sense resistor.
!
Linearity can somewhat screw up the Circle in the Square on the ILDA test pattern if its off. Its sort of the dynamic balance of the position sensor, which is two or more photodiode pairs. Normally one does not need to ever touch that potentiometer, if the factory totally screwed it up, it can affect your test pattern. Its a very subtle adjustment. I would not worry about it.
!
Your processing of the frame is "correctish" as you say. Your not that far off....
!
A Quad Square is usually 2-3-4 successive points at the corners of a sharp square wave, goal is to make a 30 to 50 Hz square with no intermediate points for fast risetime to excite ringing in the galvos for tuning. A well tuned galvo will show almost no overshoot or undershoot at the corners of a Quad Square.
:
Some lite reading:
http://www.photonlexicon.com/forums/...ILDA-STANDARDS
!

Steve

[jsibre]

Pictures of my current state of tune.

This is 'original' (i.e., before swapping sequence of writing to DAC - X then Y):


And this is after swapping the sequence to write Y first, then X:


What I find really interesting is that the circle didn't move, but it did squish along the 135º - 315º Axis.

There's also an embarrassing amount of waviness in that X Axis arrow... Hard to see that without taking a photo.

Also, if I didn't make it plain earlier, I DID mess with all six pots early in the process (I left the loner seventh pot alone, since it seemed strange out there all by itself). It wasn't that I was trying to mess with all of them just to shoot myself in the foot, but in the process of trying to figure out which was which, I did just that. To regain something close to what I started with, I've had to adjust IS, SG, LFD, and HFD. I've tried further fiddling with LIN and PS, since, "why not, at this point?", but haven't had any luck.

I'll see if I can find TP1 on my boards (as compared to the schematics) and then go from there. Throwing up a symmetrical pattern to test with is no problem.

P.S., tonight, after re-adding images, I hooked up my scope, and measured the time between X update and Y update: About 3.8μs

[mixedgas]

Invalid Attachment Specified, I can't see your pics.

Steve

[jsibre]

Hmm... Not sure what happened there. I'll try again tonight.

Did you know that the ilda test pattern PDF in the 'lite reading' link is... messed up? It goes from page 5 of revision 003 to page 6 of revision 002, and loses a few paragraphs in the process.
The glossary was very educational, helping to paint the landscape for me.

Oh, and I've started looking at the schematics, but it's slow going - the scans are sometimes hard to read, and the pinout of my connectors connecting galvos to drivers differs from the schematic. My connector only has 7 wires (but I think I figured out which correspond to what, eliminating some possibly-redundant grounds before I called it a night last night.)

Thanks !