Analog Galvo Driver

[laserist]

Typically in the bad old days up to three quadrature osc would be summed - you could do more but it tends to be more garbage than value. Each osc has a sine/cos and often a inverse sine output. (Substituting the inverse sine for the sine reverses the direction the "circle" is drawn and so for example changes a triangle (made by two quadrature osc at the right relative frequency and gain) to a cycloid with a single inside loop.) You could run VCOs in series for a frequency modulation effect, but you could do that with a summing amp on the input and just routing the output of one of the paralleled quadrature osc to the VCO input - or in the case of a the phase rom mentioned - you could use a unipolar or bipolar (or other shape) version of the input clock with a divider to get it into the frequency range of the output. The osc can also be amplitude modulated for some interesting effects. Next an analog multiplier on the output of those summing amps will support spirals and diamond spirals. (or a place to inject audio mod.) Inject an offset and you get swept spirals. Four analog multipliers and two summing amps can perform z axis rotation. At some point you'll want to switch sources either mechanically or using analog switches and this is where you need to consider just how you're going to deal with gain and offsets generated by your different signal paths…

[SaltyRobot]

Ahh OK so the ratio between frequencies comes when two osc are summed. So any individual osc will have a sin and cos at the exact same freq but when summed with another osc at a different freq provides the ability to make a star, triangle, etc. becomes available.

When you mention inject offset. That is just adding a DC part to the analog signal correct? So is this also how a joystick could move the pattern all around the screen correct? Is there a De Facto Standard for DC added to a signal by a joystick? Like do people typically use +/- 5VDC?

Any recommendations on dealing with gain and Offsets generated by the different signal paths. How do they do it in a live setting? Are the shows half per-recorded?

[laserist]

The Laserium projectors originally had four sets of scanners. There was a master gain and master symmetry for all and individual gains for each scanner. The individual osc each had gain, symmetry, frequency, polarity, AM, and FM controls. Typically the gain of the individual oscillators were all set so that even with all of them on at "10" and the masters at ten it wouldn't overdrive the output and blow scan amp fuses. More modern projectors just clipped the signal - so blown fuses became pretty unusual, but I still ride the master gain a lot - clipped signals are ugly.

Laserium's joysticks had a master gain also. There were several modes, and overrides for each scanner. One joystick was typically for offset and one for rotation, but you could and did configure to put both functions to one joystick at times.

The offset I mentioned was to be applied before the spiral multipliers this results in what we called a swept spiral like below:
100039.jpg


Say you have an image signal that goes through A, B, C, D analog "processors", and you grab the signal after A and route it through C and then E - if you send them to different scanners (and especially if there's a spiral involved) any offset between the images becomes glaring a times. Consequently you need to consider your various signal paths and provide for gain/offset trims where needed - exactly where depends on the design and the quality of the analog processing.

[mixedgas]

I bought the design rights to a line of analog consoles from a estate sale of a noted console designer. who once lived near Seattle. I also have my own stable and temperature compensated quad oscillator design. It took me 15 years of playing around to make one I really like. I HAVE BOTH DIGITAL AND ANALOG OSCILLATORS. Each has its place.

This was to move off the unstable, shaky, wiggly cantankerous 4423* and the horrid VCCS state variable filter oscillators with limited frequency range. As we have learned from DZ's threads, these are expensive to build. Is there really enough market for me to consider this at about 100$ per oscillator channel, out the door price? I also have the mother of all color modulation boards, and the log-sin-lin ramp generator for "spirality". Both cost between 150 and 200$ to do right. Those costs are raw boards without a panel and the multiple power supplies.

No, I'm not going to publish the schematics, this was bloody expensive to develop and the design rights were not cheap. The color sequences are breathtaking, but once upon a time were copyrighted. If I sell them, you'll find markings removed from certain parts, I have to do that per the terms of sale, for various reasons.


*Note the 4423's charm is its personality, no two shows will ever be A. Stable, or B, Exactly Alike.
This does add a unique quality to a show that makes it look different, and the thermal noise is not easily copied in software.
The 4423's unique ability to stall and stop oscillating at low and mid frequencies is not fun. The circuit needs a reset button to restart your show. It can take 10-15 seconds to restart when crashed. You actually watch it slowly build oscillation amplitude when this happens. It happens frequently.

The need for specialized capacitors with the "right" range of ESR and Q factor makes cloning the 4423 possible, but very, very expensive. It also needs dual precision potentiometers. The cost of making a 4423 clone is hideous. This is why the one mil-spec clone on the market is 150$ each. Certain obsolete sonar systems use the 4423, otherwise its long gone.

Steve

[SaltyRobot]

Thank you for your input. Although I was inquiring about the NDA schematic you have I really don't want the answer to the solution. I want to learn how to make the Quad oscillator the best that it can be. Like the OP said back in Feb. there really isn't alot of information that focuses the Quad Osc to what we do with lasers. Actually quite a few articles say there is mostly no use for the Cosine achieved from this type of circuit which I find humorous. Most of the information that is useful is from you and posted on PL, since you have 15+ years experience in these designs. That is why I figured I would revive this post and see what other information I can gather. From the sounds of it I will need to do like you and hack away at it for many years to perfect my own design, which is fine.

If I have any specific questions during my quest I will shoot them your way.

I bought the design rights to a line of analog consoles from a estate sale of a noted console designer. who once lived near Seattle. I also have my own stable and temperature compensated quad oscillator design. It took me 15 years of playing around to make one I really like. I HAVE BOTH DIGITAL AND ANALOG OSCILLATORS. Each has its place.

This was to move off the unstable, shaky, wiggly cantankerous 4423* and the horrid VCCS state variable filter oscillators with limited frequency range. As we have learned from DZ's threads, these are expensive to build. Is there really enough market for me to consider this at about 100$ per oscillator channel, out the door price? I also have the mother of all color modulation boards, and the log-sin-lin ramp generator for "spirality". Both cost between 150 and 200$ to do right. Those costs are raw boards without a panel and the multiple power supplies.

No, I'm not going to publish the schematics, this was bloody expensive to develop and the design rights were not cheap. The color sequences are breathtaking, but once upon a time were copyrighted. If I sell them, you'll find markings removed from certain parts, I have to do that per the terms of sale, for various reasons.


*Note the 4423's charm is its personality, no two shows will ever be A. Stable, or B, Exactly Alike.
This does add a unique quality to a show that makes it look different, and the thermal noise is not easily copied in software.
The 4423's unique ability to stall and stop oscillating at low and mid frequencies is not fun. The circuit needs a reset button to restart your show. It can take 10-15 seconds to restart when crashed. You actually watch it slowly build oscillation amplitude when this happens. It happens frequently.

The need for specialized capacitors with the "right" range of ESR and Q factor makes cloning the 4423 possible, but very, very expensive. It also needs dual precision potentiometers. The cost of making a 4423 clone is hideous. This is why the one mil-spec clone on the market is 150$ each. Certain obsolete sonar systems use the 4423, otherwise its long gone.

Steve

[laserist]

At Laser Images our philosophical basis slipped at times – at times it slipped a lot, but it was the guy in front of the audience that helped define what we did. There was exactly one BurrBrown chip in the Mark 6 and it was added pretty late in the early game as a spiral ramp shaper. I have to say I never really liked the effect much. Our analog image generator dates back to the Mark 4 (our 1st production system), and was quite simple - 3 sin/cos quadrature oscillators with switchable inverse sin. It was implemented with a variation of the OpAmp circuit found in innumerable data sheets and application notes. The “digital” image generator was implemented with 555 VCOs driving a 32 step Johnson counter. The Johnson counters outputs were connected to summing amps via 1% resistors to build a “Steppy” quadrature sine or triangle wave. The 555 timers would couple through the power supply to frequency lock leading to some interesting artifacts that were never intended but looked good – you cold get rid of the lock with cmos 555s, but nobody really liked the smoother look. The “digital” osc also had AM and FM which got used a lot.

Voltaire said, “The best is the enemy of the good.” I guess you could argue one side or the other, but the best that never sees the light of night just doesn’t matter at all.

[SaltyRobot]

How precision were your Pots back during the Mark 4? It seems like it would have been really expensive for 5% and under Pots back then. I mean hell precision Pots now days start at $25 for 5% and go up from there.

The more I read it sounds like I am just going to have to bite the bullet and have a Quad Osc made and play with it. Then make another one.

At Laser Images our philosophical basis slipped at times – at times it slipped a lot, but it was the guy in front of the audience that helped define what we did. There was exactly one BurrBrown chip in the Mark 6 and it was added pretty late in the early game as a spiral ramp shaper. I have to say I never really liked the effect much. Our analog image generator dates back to the Mark 4 (our 1st production system), and was quite simple - 3 sin/cos quadrature oscillators with switchable inverse sin. It was implemented with a variation of the OpAmp circuit found in innumerable data sheets and application notes. The “digital” image generator was implemented with 555 VCOs driving a 32 step Johnson counter. The Johnson counters outputs were connected to summing amps via 1% resistors to build a “Steppy” quadrature sine or triangle wave. The 555 timers would couple through the power supply to frequency lock leading to some interesting artifacts that were never intended but looked good – you cold get rid of the lock with cmos 555s, but nobody really liked the smoother look. The “digital” osc also had AM and FM which got used a lot.

Voltaire said, “The best is the enemy of the good.” I guess you could argue one side or the other, but the best that never sees the light of night just doesn’t matter at all.

[laserist]

How precision were your Pots back during the Mark 4? It seems like it would have been really expensive for 5% and under Pots back then. I mean hell precision Pots now days start at $25 for 5% and go up from there.

The more I read it sounds like I am just going to have to bite the bullet and have a Quad Osc made and play with it. Then make another one.

Short answer - not very precise - there's a limit to how precise a single turn 1/2 inch form factor will allow...

Laserium's analog image generator used Allen Bradley Mod Pots for frequency control. The coarse frequency control was a dual pot and the fine frequency control was a single pot and introduced a bit of phase shift. Later the fine frequency pot was replaced with a conductive plastic version and was noticeably smoother. I did an implementation here in St. Louis for my reboot experiment using 20 position dual switches for coarse frequency and a dual 10 turn for fine frequency control. I like the 10 turn fine frequency control, but the 20 position switches suck from a performance point of view - IMHO. I'll tear them out and go back to some decent dual pots at some point. The analog osc had dual log/reverse log pots for symmetry so the phase shifts could be adjusted away if it mattered. The digital osc used a single pot with the wiper grounded - so turning off center would ground that osc's X or Y signal to the summing amp.

I suspect Jon would sell you either the CYGN-A or CYGN-B if he has one lying around. Alternatively while wire wrap is as ancient as I am - it does allow for incremental development and experimentation. These circuits are just not that complex…

Well wirewrappin the CYGN-B would be an exercise - 32 resistors per axis - so 256 just for the resistors - I'd just buy it...

[SaltyRobot]

I think I am going to start out by developing a few boards in Eagle and have them made in China. If I hurry up I can have them by X-mas. LOL. Maybe I'll pay for the express shipping for the Quad Osc so I can get it by end of Nov. Damn customs.

I gave up wire wrap. It has its place just not on my work bench. I have 10K+ hours of soldering experience from previous jobs and much prefer soldering SOIC and QFP with 0603 (0402 is doable but not as reliable with my eyes) passive components by hand to the wire wrapping PDIP. Goes so much faster. Of course that's once you get the boards from the slow boat.

[pschlosser]

And I'm following this thread with interest.

I don't have much to add, other than I have several BB 4423s I'd like to put to use.