750mW RGB Build with homemade Scanners

[JohnYayas]

That's really awesome. You did basically what I set out to do when I got into this hobby. I ended up building software instead, though.

[Pangolin]

Very impressive work indeed!

[janelo]

I am glad you enjoyed my little project, Greg. Thank you!

What is your coil resistance? Coil Inductance?

The coils have a dc resistance of 1.8 Ohms and measure around 45uH.

not only in the beautiful execution of the Electronics but also the complete mechanics of you project.

Thank you Poul! I must admit that I also somewhat underestimated some of the mechanical aspects. Assembling scanners turned out to be (at least for me) quite a fiddly thing to do. There is definitely a learning curve to it. CAD did of course come in helpful to make sure everything eventually fit together.

You did basically what I set out to do when I got into this hobby

This is exactly what I wanted to build when I started playing around with lasers (I can remember spaghetti already being around). Back then my motivation was the high cost of ready-made scanners. Today this of course doesn't make much sense anymore, as the tooling cost alone exceeds the price of an entry-level set.

Very impressive work indeed!

Thanks! I enjoyed reading your book on lasers scanners and it certainly cleared up some of the confusion. In terms of scanning speed I am unfortunately still off by an order of magnitude.

[mixedgas]

I quote:

"The coils have a dc resistance of 1.8 Ohms and measure around 45uH".

The reason why I asked is L/R ~ = Available Slew Rate from the amplfier/ galvo combo.

Ie Cambridge 6800 is 120 uH and 4.5 Ohms..

Saturn 5 is 135 uH and 1.8 Ohms...

Many China products are around 2.4 - 3.2 Ohms and finer wire..

Was hoping I could find some obvious way to help you speed up, so can you try a few more turns and finer wire?

Are you running bridged on your output stage?

Either way, very nice.

Steve

[janelo]

Hi Steve,

thank you for your explanation! If I understand you correctly, increasing inductance should slow down current slew (and ultimately torque bandwidth) assuming voltage slew from the amplifier remains constant. My motors having little inductance would suggest low torque generation resulting in mechanical motor dynamics dominating system response.

I have some 0.15mm cooper wire that I could experiment with. Assuming L~N^2 this would roughly match the CT6800 with 0.15mm wire. I am currently taking a break from building motor coils though.

The motor drivers use single channel power amps with an additional current sense amplifier. I had to slow down the amplfiers because of additional phase being introduced by my CSA setup. I calculated the driver for ~30kHz@-3dB closed loop. I am not sure how this compares to other controllers, though judging by some photos a larger shunt resistor directly feeding into the power amp input seem to be more common (and is probably faster).

[mixedgas]

Others, more skilled in mathematics then myself, have worked out the "Circle in the Square" portion of the ILDA test pattern to correspond to between 2.4 Khz and 3.2 Khz of bandwidth at 30Kpps / 8 Degrees for one of the two ILDA patterns, So 30K response may not be unreasonable. The ILDA test pattern was not designed to be a measure of Galvo response, but a means to tune Galvo Amplifiers for image interchange between companies. It has, however, became a rough measurement of performance that both Artists and Manufacturer have agreed on.

The traditional factory test would be a swept sine or noise impulse response on a Dynamic Signal Analyzer, and pulse width to some percentage of settling at something on the order of eight degrees commanded jump. While I have a DSA, I have no idea how factory would have conducted standard tests in their lab.

Mr. Benner at Pangolin / Scannermax would, however. I look forward to his correcting me.

My next concern would be the sleeves around the rotor as extra inertia, , plus calculating rotor and mirror inertia. Dr. Montague in one of his papers cites a desired range for the ratio for load to rotor inertia. I also have has his original chapter in an earlier version of Laser Beam Scanning which was more detailed.

His more mathematical partner was Dr. Pierre Brosens, who has a few papers out on "Scan speed" in terms of step response.

I'm a bench technician, so my knowledge of "controls" is partially self taught book learning with some hands on experience. I've looked for obvious issues, based on experience with commercial hardware, more so then analysis. I have, in my career, and as a hobbyist, had my hands on nearly everything in the market at one time or another.

Steve

[janelo]

Scanning the 15k test pattern should theoretically require 1.25kHz of bandwidth. This is because the diagonal of the square is sqrt(2)*side length (= 20log(1/sqrt2) = -3dB of damping) and one cycle being drawn using 12 points. This roughly matches with ~300us rise time that I measured - so I'm guessing the current setup has a overall closed loop bandwidth of 1.25kHz?

The 30kHz bandwidth I mentioned earlier only refer to the output stage current driver, that dominates my driver's open loop bandwidth. Also ~5us of dead time (considering ADC sampling, data transfer and controller computation) are added. My rotors have ~0.065g/cm^2 of inertia (not considering the bearings), which is a lot compared to the scanners you mentioned. I must admit that I did not numerically calculate the motor's theoretical bandwidth. I also did not consider other effects that would, as you mentioned, likely be verified with recording a amplitude/frequency response using sine sweep/impulse input. This was primarily under the (perhaps seemingly naive) assumption that electrical effects and possible oscillations would only matter substancially at much higher frequencies.

I remember reading one of Mr. Brosens papers that formulated the "BLIND"-formula for aproximating motor torque (that stuck pretty well ). I can also recall looking for the derivation of this forumula that the was supposedly Optical Scanning Handbook. The "newly" published version being shortened does explain it being removed from the appendix.

[jian725]

Perfect man
1.Change 3x20mm magnets
2.Change 0.13mm coil,get about 3.5~4 Ohm,100uH.
3.currently set to 5A pk enough

[sbk]

Wow man, you have just taken laser DIY to a whole new level, congrats !!

Very impressed, especially on the home-made galvos and digi amp.

Sebastien

[absolom7691]

I might inquiry if you have any plans to market the diode drivers. I'd be keen to purchase, depending on the price.