Motorized Fader Signal Question

[Isra]

Hello
I have here some Penny & Giles motorized Faders and would like to use them for a Project.

The Idea is to program a Microcontroller where I can store and recall some Fader positions.
If I look to some Videos on Youtube about flying Faders, I see that they can move pretty slowly or very fast.
My Question is how does the Signal looks like, that goes to the Motor of the Fader. It can't just be driven by
more or less Voltage.
Is anyone out here who could give me some more Infos about that?

Thanks !!!

 

[JohnRoberts]

generally they are driven by small DC motors with servo positional feedback track from the fader to control motor .

http://manquen.net/audio/docs/P&G%20fader%20families%205-20-03.pdf

JR

 

[PRR]

> It can't just be driven by more or less Voltage.

Why not??

My car windshield wipers run slower when the battery is going dead, faster when I rev the engine (generator).

Electric street-cars had a switch to put the motors series for 200V each to go slow or parallel 400V each to go fast.

DC motors have been speed-controlled by voltage for 140 years.

But as you suspect, setting a voltage isn't quite what you want. Low voltages stall from friction. Once you overcome that, the slider moves to an unknown position. As JR says, there is a servo pot, and a negative feedback system to compare desired position to actual position and get an error voltage which drives the motor. If the slider should move fast, the error gets large and the motor GOES. If the slider is to move slow, the error increases until the slider just moves, then stays small to track the slow moton.

 

[abbey road d enfer]

I see that they can move pretty slowly or very fast.

As PRR and JR have already mentioned, speed is essentially governed by the voltage that is applied to the motor.
This voltage is the difference between the control voltage (defined by the set point) and the voltage coming from the control track. When a large move is required, the voltage is high, which prevents stalling. When a small change is wanted, the voltage may be too small to fight friction.
My experience is that you can get about 1mm accuracy/resolution with a very basic servo arrangement, which corresponds to less than 1dB in the fader's comfort zone.
If you want better response/accuracy, you need a PID servo. You may want to implement that in the MCU.

 

[sahib]

...My experience is that you can get about 1mm accuracy/resolution with a very basic servo arrangement, which corresponds to less than 1dB in the fader's comfort zone.
If you want better response/accuracy, you need a PID servo. You may want to implement that in the MCU.

Abbey,

What was the fader length (electrical travel) in your application?

 

[Newmarket]

generally they are driven by small DC motors with servo positional feedback track from the fader to control motor .

+1
The P&G 'model' circuit implements a PID control loop with gain expansion at low error.
The gain expansion is important there as at 'rest' it keeps the string tensioned (but not moving) one way or the other which gives smoother motion in one direction.
The PID control gives good accuracy and damping.

As I recall from a technical investigation this implementation was less 'harsh' than an SSL version which gave more fader wear.
I don't know about the NEVE flying faders. Those faders were assembled by MartinSound using P&G faders but adding their own motors and stuff.

fwiw I did have a project implementing moving fader control by microcontroller.
Didn't work out - fine with single long movements but a fader track destroyer in real usage ! Live and Learn.
Might be better with faster micros now...

 

[abbey road d enfer]

Abbey,

What was the fader length (electrical travel) in your application?

100mm.

 

[abbey road d enfer]

The P&G 'model' circuit implements a PID control loop with gain expansion at low error.

I searched the P&G website (Curtis-Wright now); can't find this document. Would you have a copy?

The gain expansion is important there as at 'rest' it keeps the string tensioned (but not moving) one way or the other which gives smoother motion in one direction.

My understanding is that it gives increased smoothness when moving in opposite direction to the last move. How does it feel in practice?

The PID control gives good accuracy and damping.

No doubt about it. However, mechanical wear may change the characteristics enough to disturb the PID. Seems a good idea to implement it in a MCU with auto-learn.

As I recall from a technical investigation this implementation was less 'harsh' than an SSL version which gave more fader wear.

Why would that be? That is somewhat counterintuitive. I would think that better damping would prevent unwanted bouncing...? Was the SSL implementation digital?

fwiw I did have a project implementing moving fader control by microcontroller.
Didn't work out - fine with single long movements but a fader track destroyer in real usage !

Same puzzlement here.

Might be better with faster micros now...

How would a faster MCU help with this? Do you suggest that the slower MCU was not fast enough, introducing lag in the NFB loop, resulting in unstability?

 

[PRR]

> you need a PID servo

In my day we used Lead-Lag feedback. A slow loop gave long-term stability. A fast loop controlled sudden lurches. Indeed in recent times this is a PID. Much basic info here:
https://en.wikipedia.org/wiki/PID_controller

While we used to do it with dashpots and trim-screws, then capacitors and trim pots, these days all that is usually done in a computer algorithm, and I bet there is open-source code for your CPU.

> with auto-learn

Or that (not real practical with mechanical or analog controllers). Set goals instead of trying to prescribe the method. I can see good and bad points to auto-learn. That first smoke-test may be stop-banging. Once parameters are found, do you leave auto-learn on or off? On, it will compensate for dusty faders, even to the point of burning a motor. Off, it won't compensate. If you set a range around the learnt parameters, it will work fine until the day the dust exceeds the range, probably your most-booked day.

 

[abbey road d enfer]

> you need a PID servo

In my day we used Lead-Lag feedback.

Unless proved wrong, seems to me it's two different names for the same animal. Even the IEEE seems to think the same.

> with auto-learn

Or that (not real practical with mechanical or analog controllers). Set goals instead of trying to prescribe the method. I can see good and bad points to auto-learn. That first smoke-test may be stop-banging. Once parameters are found, do you leave auto-learn on or off? On, it will compensate for dusty faders, even to the point of burning a motor. Off, it won't compensate. If you set a range around the learnt parameters, it will work fine until the day the dust exceeds the range, probably your most-booked day.

My Tascam mixer does it every time it wakes up and leaves it alone till the next turn-off, unless it fails and sends a fault message (never happened).

 

[sahib]

....Might be better with faster micros now...

You don't need a faster micro. You need more bits.

For a smooth  motion and minimal overshoot particularly in shorter travels you have to be able to accelerate/decelerate. I am not sure how many bits Abbey used (I should have asked) but say 8 bits over 100mm you have a theoretical resolution of 0.392mm. If Abbey achieved 1mm at 8 bits then this is the result of the wear and tear and the backlash that he mentioned. 

 

[abbey road d enfer]

You don't need a faster micro. You need more bits.

For a smooth  motion and minimal overshoot particularly in shorter travels you have to be able to accelerate/decelerate. I am not sure how many bits Abbey used (I should have asked) but say 8 bits over 100mm you have a theoretical resolution of 0.392mm. If Abbey achieved 1mm at 8 bits then this is the result of the wear and tear and the backlash that he mentioned.

My solution was purely analog; the 1mm thing is the result of stiction, not lack of resolution.
I think we may attach too much importance to the visual/mechanical aspect of resolution; when writing level, the resolution is that of the A/D, because stiction does not apply.
When reading automation, there are several cases:
If the fader has an audio track, yes, the actual resolution/accuracy of the mechanical position is important, but
If the control voltage/data actually drives a VCA/DCA, the fader is just a visual display.
It seems the latter case is the most frequent today.

 

[zamproject]

Hello

There is various solution, PID anolog or digital, motor power PWM or variable DC
I worked hard the past few years about this subject and can tell that analogue PID and variable DC is the way for great result.
But beside the motor driver there is lot of other things to manage for an efficient and usable system.
You need touch detection and various IO (button and led visual feedback) to handle basic (and extended) functions...
Also proper software handling to talk with "standard" protocol and DAW
Then metalwork !
Everything is in the details ....

This is a 2+ years R&D, bullet proof since one year  :
https://www.youtube.com/watch?v=Z9vLD8GdzgM

Best
Zam

 

[abbey road d enfer]

First video, first test with PMW controller

Do you mean PWM? If it's the case, it's just the way to drive the motor. I notice a lot of bouncing, but I would think it's not the drive method that's at cause, it's the servo loop...

 

[zamproject]

Hello

Yes, first video show noticeable fast oscillations.
I don't get your differentiation regarding PID and PWM, you need both anyway...
with PWM, PID is at soft side !
If you plane to use this in an analogue desk (focusrite ?) you better check asap in situ how PWM leak noise affect audio path.
You may (or not) have bad surprise...

Best
Zam

 

[abbey road d enfer]

I don't get your differentiation regarding PID and PWM, you need both anyway...

I would say you need a motor drive, that could be linear or PWM and a servo, that could be just proportional, or PI, PD or PID. Many choices.

 

[zamproject]

I would say you need a motor drive, that could be linear or PWM and a servo, that could be just proportional, or PI, PD or PID. Many choices.

Hey Abbey

Yes I agree about that, it's in my previous post.
So what do we see from OP video ? first PWM with P only and second with PID ?
I guess both use PWM ...

Best
Zam

 

[Newmarket]

Thanks for the input about unwanted noise in the audio path. Of course we will check this point permanently. Should there be unsolvable problems, we will replace the mono fader with a stereo fader. So this point would be no problem anymore.

Best regards

I may be misunderstanding what you mean but I don't see a link between possible noise induced into the audio signal and whether the fader is mono or stereo ???

 

[abbey road d enfer]

First we started the faders with 12V and a PWM of 70%. Less than 70% does not work because the faders do not move anymore.

Do you mean you powered the motor with  a fixed 70% of 12V? So why use a PWM?

Thereafter, the speed was made dependent on the distance to the target

Does it mean that before, the voltage was not proportional? Like a switch? That would be similar to the regulation used in central heating; not very accurate...

with a digital PID controller.

Indeed PID controllers prodide some improvement on this type of regulation by providing some kind of PWM control.

Should there be unsolvable problems, we will replace the mono fader with a stereo fader.

I don't understand how a stereo fader would improve noise issues...

 

[abbey road d enfer]

The idea at the end is to use a digital potentiometer, which then regulates the audio signal. The mono-fader now leads no audio signal, but only gives information where the controller is. therefore I have to control the audio signal via this digital potentiometer in 128 steps.

It seems every time you give an explanation it makes things even more obscure.

Am I understanding correctly that in the end, the audio signal will be going through a digital pot with 128 steps and the fader be just a visual indicator when reading automation?
Then, if the fader is used as a visual indication of the automation data, you don't need a very sophisticated servo, because the audio level is driven by data, not by the fader.
128 steps has long been deemed insufficient; 256 steps is considered a minimum, for about 0.25dB resolution in the critical -30/+10 range.

But if I take a stereofader, then one Channel controls the audio signal, the other Channel gives the information about the position of the fader.

  Again, I don't understand.  There are two basic possibilities. Either you run audio into one track and another track provides servo/position information, OR audio is ran through a VCA or DCA controlled by the automation data.
If you use one track for audio, then the other track is not another audio track, it's a dedicated servo track, typically 10k linear taper, as opposed to 10k with an audio taper.
If you use VCA/DCA, you just need one servo track.
Check the P&G catalog, they have both types available.

Thus, no unwanted noise could occur because I would drive completely analog with the audio signal.

Noise can be transmitted by cross talk, due to capacitive coupling or spike currents polluting grounds.