Best way to cool galvos?

[illuzion]

How are you guys keeping galvos cool? I just started having some clipping troubles with my CTI units (6210s) under heavy workload, when it was clipping, I touched the XY mount, it was HOT. Do you have the XY mount just bolted to your main optical chassis plate, or is there some fancy heatsink/cooling setup you need to run? Maybe a little thermal interface grease is the answer, the XY mount got really hot, but the plate it was mounted to was only warm.

[liteglow]

My DT-40 galvos have heat sink on them..
And I got a tiny fan to cool them down..

But I was told to not cool them down to much, it can reduse the quality of the galvos output...
as they need the correct temprature to work at 110%

[dave]

Heat sink grease definitely wouldn't hurt..

Although i would advise against mounting a fan any ware near the galvos, as you will probably get interference from the fan, and may also get issues with the mirrors getting crap blown over them...

What do you mean by clipping? Perhaps you are overdriving them? If you are referring to clipping as the same as audio clipping, this is very bad, as clipping generates high frequency harmonics, which will damage the galvos, and this may explain why they are getting so hot

[illuzion]

I am not dramatically overdriving them, plus I am using the CTI servos which have reasonable protection against that. What I get is the system scans fine for a period of time, maybe 20 mins or so, then the image starts wandering, no longer stable, lots of jitter, and it starts to shrink vertically, and the galvo mount is HOT, no whining or resonance sounds, just no longer able to mantain a stable scan image. Not doing anything super crazy, I was able to do this with the ILDA 30k pattern projected at like 10-15 degrees. I figured a fan would be a bad idea, noise + dust induction into what should be a clean assembly. I figured the best is a water cooling setup, to get heat off the galvos and move it somewhere else (ie heat exchanger w/fan), but this sounds like overkill for what I am doing, I would like to keep it simple as possible. Should the 1/4" plate it's mounted to be sufficent to cool it passively???

[stanwax]

Im not entirely sure what your problem is but you need to get to the bottom of it rather than masking the symptoms. The galvos and mount should not get hot unless you are really driving them hard (very fast and trying to get big angles). Find out whats wrong and fix it is my best advice.

Rob

[fluff]

One of my customers had a set of 6210's completey seize up.

The only real explanation was overheating, so I made sure the replacements were well secured and put a few dabs of thermal paste on them.

No problems since then ( 3 months ago ) so it seems to have done the trick.


They shouldn't need anything extra under " normal " usage ( imho ), so if it carries on heating up, you not getting enough thermal transfer or your overdriving or your sending a distorted signal etc !! I've , sold about 15 sets all mounted on the metal base as a heatsink with no other cooling. Only one has died .

Fluff

[yaddatrance]

I'd agree with fluff with one addendum. If there's even space for thermal grease between the scanheads and the mount then the mount might be out of tolerance. The holes for the CT mounts we use are reamed out to tolerance so the effect of wear on tooling for making the mounts means that the mount gets even tighter!

The real culprit is that servo gain is set too high. This usually happens when you tune everything to be correct at a wider scanangle than standard. The problem which arises at that point is that (due to the cambridge driver's architecture) you're sending a lot more current into the coils... The CT drivers have a relatively low fixed voltage (+/-15-30V) to work with... The rule of thumb is that the higher the voltage, the quicker you can change the flux of the coils (by effectively pushing the old voltage level out quicker) with the additional advantage of using less current for the same density. And because you can't adjust the stock driver's drive voltage, the only way to make them move "correctly" at wider angles is the boost the gain, which can even double or triple the current used normally if improperly set. If you're using "official" cambridge made drivers then you'll see the scanners compress a bit as there are some sanity checks built in. The fact that you're running into problems makes me think that either you tuned the scanheads to the very edge of the spec or you're using a microamp or similar. If heatsink doesn't work, try turning down the scanangle to see if it cools down, and if so, then retune the 30k pattern while scaling the pattern down to a lower scanangle.

[illuzion]

Okie here is a little update, my main problem was that the XY mount was getting very hot (indicating that there was adequate thermal transfer from the galvos to the mount (so mount tolerences are out of the equation). The problem is the heat is staying contained within the mount. I pulled it and noticed some remnants of a barcode label on the bottom of the mount (probably acting as a thermal insulator), so I cleaned it up with solvent, and went over the aluminum with some scotch brite, and cleaned up all the mating surfaces and put some high temperature coefficient grease on all mating surfaces (XY Mount, riser block to get it to beam height, and baseplate. Powered it up with same test as before, ILDA test pattern 30k, pretty high scan angle, where it lost servo control at 10-20 mins before, it ran fine now, ran it for approx 35 mins, the temperature differential between the Galvo mount and the baseplate was much smaller, there was still localized heat at the point where the galvo coils are, but nowhere near as bad as before. Boy did the servo amps heat up though, and I have them on a massive heatsink. Everything is genuine CTI, servos too, I am using the 678 amps with all appropriate lightshow cti mods. Performance seems to be good, I am not complaining, I am reasonably sure I have these tuned to near max performance. I can display the ILDA30k without problems, and I can scan at very high angles, but when I do the ILDA pattern is still very good, only thing that I lose is the circle in the square gets smaller. Part of the problem is these servo amps do not have separate adjustments for LF damping and HF damping like some of their more modern units do. But from what their head engineering guy told me, that for lightshow, these amps will blow away the newer 671 amps on a 6210 set because they never developed the mods for the 6210 lightshow app on the the newer 671 (or 673 dual driver) driver, and the 671 with the "high output" (discrete amplifier) option would be useless as the coils can't handle the current.

[Astroguy]

I have a small 12vdc fan that blows on the back of my galvo mount. I have been using that way ever since I built my RGB projector. I have not had any problem with performance with this setup. I do have a filter for the air flowing into the projector too.

[buffo]

I can scan at very high angles, but when I do the ILDA pattern is still very good, only thing that I lose is the circle in the square gets smaller.

Darn... I wish Bill Benner wasn't so tied up with this trip to China. I bet he can explain this a whole lot better than I'm about to try to...

The center circle is drawn with the scanners in ballistic mode. This means the points for the circle are already intentionally spaced far enough apart that the galvos will never actually reach the current point before they are sent off towards the next point. Thus, they are moving at absolute maximum speed and are also under constant acceleration. (The test pattern is designed with this in mind.)

If the circle is getting smaller, it means you are scanning too fast and the scanners can't reach the outer boundaries of the square even when traveling at max speed. Which means you need to either slow down your scanning speed or reduce your scan angle, or both. It also means you are overdriving the galvos, which explains the excessive heat. They may be able to survive it, but they certainly weren't designed to operate that way.

Adam