Analogue Protection Filtering for ILDA XY/Galvos

Hello Photonlexicon community!

This is my first post. I design synthesizer modules for video synthesis through my company LZX Industries. I'm about to release a EuroRack format synthesizer module for ILDA output. My prototypes have been going great and working well, but I am a bit stuck on the technical requirements of high frequency protection filters for various laser galvos. I'm already limiting voltage ranges to within the ILDA specs using Zener diodes.

My current design is a 2-pole butterworth filter with a cutoff around 1KHz, separately for X & Y differential outputs. It seems to be working well for the few individuals I've built custom modules for, but before I do a larger product release, I want to make it variable, with settings for 10KPPS/20KPPS/40KPPS scanners. Here are some initial questions:

(1) Is there a direct correlation between "Points Per Second" and analogue bandwidth? I have heard "divide PPS by 12", but no justification for that. I'm looking for an in depth technical explanation that will aid me in designing these filters.
(2) Are there possible repercussions to filtering X & Y separately, and creating harmonics that would damage the galvos, or does it make sense that limiting each would protect the galvos?

Worst case scenario, I will use a microcontroller with an ADC/DAC to limit the bandwidth, but I'd like to avoid it if at all possible, and keep the circuit simple.

I'm very grateful for any help,
Lars Larsen
LZX Industries

It's a slew rate limiter, not a low pass filter that would be required for protecting galvos. Designing a DC -2400 HZ with an option for Dc-3600 HZ filter that would not interfere in creating the abstract would be a monumental task. Images are sensitive to small amounts of phase shift, you'd need less then 1-2' phase shift over your low pass bandwidth as a design target.
~
You can ignore 12K... Basically systems scanning abstracts with 12K scanners, are a few museum pieces kept lovingly running, by a few skilled people.
~
Have you actually tried your circuit with your OWN well tuned pair of modern 30k galvos? You might have noticed the roll off... This industry is intolerant of expensive mistakes, and we'd prefer to know you have your own projector, tuned to the 30K standard for graphics, and have ran tests. Much can be learned with a signal generator, and monitoring galvo amp test test point one (feedback signal) and TP3 (Galvo Current) with a scope.
~
Formal Protection is done via slew rate limiting, and on professional amps, a coil temperature calculator built into the GALVO position feedback circuit. The Asian clone scanner amps have a bare SOIC pad where the expensive CTC chip goes, so protection for most hobby users is in good design. However only expensive Galvo amplifiers have slew rate limiters, so protection is done by designing the control waveforms to be non-hazardous in most cases... Which is not simply using a low-pass...
~
The ILDA test pattern and ILDA Standard is not defined in terms of bandwidth, because Galvos have a very complex step response that is dependent on what the Galvo was doing in the past, and the angle of the new commanded step. The ILDA test pattern was designed for image interchangeability, and accounts for the complex response. It really does not measure bandwidth. Although there is one portion of the test image that measures a "3 dB down" small step angle under one very limited condition. One very good engineer came up with "divide by twelve" for very small angles, but it is only a approximation based on his analysis of the pattern using a scope. Yes, Galvos have a roll-off, but there are many, many, variables that account for how they behave. Including nasty shaft resonances..
~
It was not the intent of the Standards Committee to make the ultimate technical test pattern for engineers, the goal was to make a pattern that allowed maximal use of the scanner performance while allowing interchangeability of images between users around the world. Most people actually use between four and six different patterns to measure the performance of their scanner... Visually...

~
Hard zener clipping has a definite impact on the image, it may or may not be bad for the Galvo depending on repetition frequency, position of the clip in the scan field, and scan angle of the over all image. Unlike music synthesis, you can "see" your waveform mistakes and they show up in annoying detail.
~
If you limit your console's oscillator Fmax, you will be better off...
~

S.

Thanks a lot for the reply.

Quote:

---

It's a slew rate limiter, not a low pass filter that would be required for protecting galvos.

---

They're electrically quite similar circuits, no? Is it the response curve that is the difference you mean? A slew rate limiter would be an easier circuit than the switched capacitor circuit I was previously using though, so it's a good suggestion.
Here's a simple circuit for linear response I've used before: http://electro-music.com/forum/phpbb...ar_lag_113.gif

Quote:

---

Designing a DC -2400 HZ with an option for Dc-3600 HZ filter that would not interfere in creating the abstract would be a monumental task.

---

Inevitably, it will create some limitations on the frequency ranges and types of signals the circuit can pass -- but that's okay. This is intended to be a generic converter for using any type of signal to control laser galvos via ILDA standards, so I need to assume worst case scenarios in terms of source frequency ranges. A slew rate limiter or lpf with selectable cutoffs is certainly something I am confident in designing. I plan to include a bypass switch for users who really know what they're doing or may be using dedicated consoles with fixed frequency ranges.

Does the "divide KPPS by 12" calculation seem correct to you, when it comes to maximum frequency? Any other tips or advice for designing this protection circuit? Temperature calculation isn't really an option, unfortunately.

Instantaneous change in motion (galvos going from a dead stop to a constant rate of rotation) will require a signal that has an infinite series sum of harmonics.

Consider scanning a simple square. You need X and Y waveforms that have elements that look similar to a triangle wave; ramps that go up and down and dead stops and starts (obvious angles). This is perfectly fine for your scanners.

A low pass filter would mess that all up.

James.

No, A slew rate limiter on a modern galvo amp is not a low pass filter. It looks at input DV/DT (rate of change) and makes a yes or no decision to remove the input signal until the rate is below what the Galvo can handle. It drives an analog switch or removes drive from the motor coil. And it is angle dependent, it will ignore small changes. Modern ones use DSP.
~
Lars, you lived until now, in the audio world, a world of highly symmetrical signals, where the DC offset is removed. You have much to learn when you extend the bandwidth down to DC, which effectively is a signal of infinite bandwidth. Small changes in the phase angle with respect to frequency of a laser show control signal show up, and are unsightly to the viewer. This gets integrated across the whole small signal bandwidth, as James so elegantly expressed. It then goes on to excite resonances in the control system, which are well beyond the scope of this post.
~
The transition region from DC to about 20 Hertz is very important in vector laser displays.
~
The protection units in the past based on a low pass topology, get bypassed, in Laser Show use due to the distortion. Why?, Because the command to jump to a DC offset position, which is very important in laser graphics, has a infinite series of harmonics which pass thru the filter as a spike. The viewer then sees a distracting flash on the screen. The low and mid frequency portion of the image passes thru, while you watch the capacitor in the integrator charge up with the DC portion, and half of your image slowly slews to its new position while the rest catches up. It looks like hell to the customer, and even worse to a viewer. While this concept is just fine in protecting the galvos of a laser engraving system, it looks like garbage to a human viewer, especially in a beam or abstract show..
~
Believe me, I have one set of amps that does that... For a good reason, because they drive large 25 mm mirrors that would torque themselves out of the mount if overdriven in Slew... I might shoot you a video on the weekend, and it will clearly show how you'd be wasting your time... It will even scan the ILDA test pattern at 24K with the large mirrors, which is unheard of in the show community. Yet if issue a command for the test pattern image to move across the screen with a DC offset, the image will tear apart for about one second until the new steady state is reached.
~
I'll pull up a post of a galvo transfer function from the past tomorrow... You might be shocked with what your dealing with... Please get a pair of closed loop, Dragon Tiger or Cambridge or PT Scanners, and learn what your dealing with. This is a very complex system...
~
And NO, as I have explained multiple times in the past on this forum and in others, there is NO solid relation between KPPS and frequency response. The divide by 12 portion for the "circle in the square" portion of the test pattern is an approximation for a fast one to two degree jump of the beam position in the image. Scanner manufacturers specify a small angle or step bandwidth and a large signal bandwidth. The divide by 12 is a guess at the small step bandwidth in one test condition, and has little to no relation to the large step bandwidth. I know the skilled engineer who developed that approximation, and he'd be the first to tell you that your dealing with a much more complex transfer function then that..
~
Scanner Speed during an image is dependent on the instantaneous commanded angle in a way that requires a complex polynomial to describe.
~

Steve

I think I am understanding better. Thank you for your talking this through with me.

Quote:

---

Instantaneous change in motion (galvos going from a dead stop to a constant rate of rotation) will require a signal that has an infinite series sum of harmonics.

---

OK. Is it correct to say the goal of a protection circuit is to limit the speed at which the galvos can move from one end to the other of their scan angle in a linear fashion?

Here is a slew rate limiter circuit below, with a bandwidth of about 500Hz at +/-5V following by an LPF that gently rounds off any transients. The input is a 100Hz square wave, so you can see how the edges of the square are slewed to limit the speed of the galvos.

Attachment 49774

Does this look correct to you?

Thanks Steve. I appreciate the responses, your time, knowledge, and the opportunity to talk to some folks who know their business.

My background is in designing DC-coupled circuits for voltage controlled video processing -- as applied to synthesizer modules which can interact with the audio world, so I'm familiar with the importance of what you're talking about -- in an AC coupled video signal, all the information on relative brightness of the display disappears. But you're right, I have no experience in protecting a mechanical device such as a scanner.

Another approach I could take is by adding a microprocessor which would sample the X & Y waveforms at a high sampling frequency -- detect rate of change in firmware via a galvo transfer function -- and use the micro to control an analogue switch to cut the signals when they go out of range -- this seems closer to what you're saying a protection circuit in a galvo does.

If you were designing this circuit, which route would you take?

Quote:

---

And NO, as I have explained multiple times in the past on this forum and in others, there is NO solid relation between KPPS and frequency response

---

What is the technical definition of KPPS? I get "thousands of points per second", but I am confused by what constitutes a "point." Is this based on a measurement of a percentage of the scan angle, or is it something else?

A point is relative to a DAC with a clock.

It's a sample.

In analog there are no points.

It might help if you think about how a scanner actually works.

There is a shaft with a permanent magnet on it and it's suspended to rotate inside of a stator or coils of wire.

If the coils are fully charged in one polarity the rotor will rotate so that the magnet aligns with the charged coils. If the coils are fully charged in the opposite direction the rotor with flip the other way.

There is a rotational angle indicator on the shaft of the rotor and the rotational position of the rotor is sent back to the scanner's amp.

So to get the rotor to a specific location or angle of rotation, the coils fully charge to move the rotor from wherever it is to where it needs to be and before it gets there, the coils charge in the opposite direction to slow it down and stop it at the desired location.

The rotor has mass so it has inertia. It has to accelerate and decelerate.

So the coils are only charged when the rotor is not where the control signal tells it to be.

Charging the coils makes them hot!