Motor fader servos

[Synapsys]

I have read here on the forum that the motor fader servo tracks cannot be multiplexed like pots are. Can someone explain to me why this is?

Thanks :)

[Synapsys]

I have read here on the forum that the motor fader servo tracks cannot be multiplexed like pots are. Can someone explain to me why this is?

Thanks :)

[titan]

Because servo tracks are controlled by an electronic (pulsewidth) For controlling servos you must permanently send every 20ms a signal PWM.

But i m not expert with that, maybe im wrong...

[titan]

Because servo tracks are controlled by an electronic (pulsewidth) For controlling servos you must permanently send every 20ms a signal PWM.

But i m not expert with that, maybe im wrong...

[smashtv]

I have read here on the forum that the motor fader servo tracks cannot be multiplexed like pots are. Can someone explain to me why this is?

They could be multiplexed, but there are several reasons not to, the best one is that there is a tradeoff in performance to keep the motors response fast.

Multiplexed inputs are read one at a time. If fader 1 has a fast move in progress and MIOS is off reading another pot when it's stop time for fader 1, the fader will not hit its target accurately.

Keep in mind that most moving fader designs use one processor per fader to keep up, TK's design is pure genius. :)

Because servo tracks are controlled by an electronic (pulsewidth) For controlling servos you must permanently send every 20ms a signal PWM.

Yes but that is for driving RC model servos, not reading the "servo" track on the faders.

The term servo is used to describe a lot of different types of designs, but usually just means a motor with some type of positional feedback to its controller.

Have fun!  ;)

SmashTV

[smashtv]

I have read here on the forum that the motor fader servo tracks cannot be multiplexed like pots are. Can someone explain to me why this is?

They could be multiplexed, but there are several reasons not to, the best one is that there is a tradeoff in performance to keep the motors response fast.

Multiplexed inputs are read one at a time. If fader 1 has a fast move in progress and MIOS is off reading another pot when it's stop time for fader 1, the fader will not hit its target accurately.

Keep in mind that most moving fader designs use one processor per fader to keep up, TK's design is pure genius. :)

Because servo tracks are controlled by an electronic (pulsewidth) For controlling servos you must permanently send every 20ms a signal PWM.

Yes but that is for driving RC model servos, not reading the "servo" track on the faders.

The term servo is used to describe a lot of different types of designs, but usually just means a motor with some type of positional feedback to its controller.

Have fun!  ;)

SmashTV

[uclaros]

Hi!

What kind of motor does a motorfader have?

And can I use such a motor connected to a rotary pot and make a motorised pot for use with the MF module??

Thanks, uclaros

[uclaros]

Hi!

What kind of motor does a motorfader have?

And can I use such a motor connected to a rotary pot and make a motorised pot for use with the MF module??

Thanks, uclaros

[Synapsys]

Hi SmashTV,

Thanks for your response but I do not agree with the one at a time theory here. The PIC 18f (like all MCUs) can perform an A/D conversion on only one of the 8 analog input pins at a time anyway (i.e. the ADCON0 register bits CHS2:CHS0 select which of the 8 inputs the conversion is done on). I do realize that using multiplexed analog adds some overhead in terms of switching channels. However I do not know if the switching time vs. changing the ADCON0 channel select is significant here.

There is no doubt that TK's design is quite elegant and certainly works well but since Microchip chose to assign the SPI slave select bit (*SS on pin 7) to one of the analog input pins it is not possible to use the SPI  bus and process 8 fader servos on a slave core.

This question was in no way intended to challenge TK's design. It was intended purely as a means to understand fully how MIOS and MBHP work together.

Regards, Synapsys

[Synapsys]

Hi SmashTV,

Thanks for your response but I do not agree with the one at a time theory here. The PIC 18f (like all MCUs) can perform an A/D conversion on only one of the 8 analog input pins at a time anyway (i.e. the ADCON0 register bits CHS2:CHS0 select which of the 8 inputs the conversion is done on). I do realize that using multiplexed analog adds some overhead in terms of switching channels. However I do not know if the switching time vs. changing the ADCON0 channel select is significant here.

There is no doubt that TK's design is quite elegant and certainly works well but since Microchip chose to assign the SPI slave select bit (*SS on pin 7) to one of the analog input pins it is not possible to use the SPI  bus and process 8 fader servos on a slave core.

This question was in no way intended to challenge TK's design. It was intended purely as a means to understand fully how MIOS and MBHP work together.

Regards, Synapsys

[Screaming_Rabbit]

Hi Uclaros

Found this in the French forum http://cgi.ebay.de/ws/eBayISAPI.dll?ViewItem&category=12078&item=5712857016

... not more expensive than an ALPS N-Type Motorfader  :)

Just place a button for the touch-mode next to it.

Greets, Roger

[Screaming_Rabbit]

Hi Uclaros

Found this in the French forum http://cgi.ebay.de/ws/eBayISAPI.dll?ViewItem&category=12078&item=5712857016

... not more expensive than an ALPS N-Type Motorfader  :)

Just place a button for the touch-mode next to it.

Greets, Roger

[smashtv]

Hi SmashTV,

Thanks for your response but I do not agree with the one at a time theory here. The PIC 18f (like all MCUs) can perform an A/D conversion on only one of the 8 analog input pins at a time anyway (i.e. the ADCON0 register bits CHS2:CHS0 select which of the 8 inputs the conversion is done on). I do realize that using multiplexed analog adds some overhead in terms of switching channels. However I do not know if the switching time vs. changing the ADCON0 channel select is significant here.

You are right about the A/D conversion on only one pin at a time thing.  You have to remember here that it is only reading one -pin- at a time, doing that you don't have to wait for settle time (or other waits) that add enough to the time overhead of multiplexing to make it unusable for this.  

If the reading from the pot is not as real-time as possible, the fader will overshoot enough to cause the motors to overwork, so you would have to bring the move speed way down, stealing more time away from other processes (still working on a move that should be done) etc.

Have fun! ;)

SmashTV

[smashtv]

Hi SmashTV,

Thanks for your response but I do not agree with the one at a time theory here. The PIC 18f (like all MCUs) can perform an A/D conversion on only one of the 8 analog input pins at a time anyway (i.e. the ADCON0 register bits CHS2:CHS0 select which of the 8 inputs the conversion is done on). I do realize that using multiplexed analog adds some overhead in terms of switching channels. However I do not know if the switching time vs. changing the ADCON0 channel select is significant here.

You are right about the A/D conversion on only one pin at a time thing.  You have to remember here that it is only reading one -pin- at a time, doing that you don't have to wait for settle time (or other waits) that add enough to the time overhead of multiplexing to make it unusable for this.  

If the reading from the pot is not as real-time as possible, the fader will overshoot enough to cause the motors to overwork, so you would have to bring the move speed way down, stealing more time away from other processes (still working on a move that should be done) etc.

Have fun! ;)

SmashTV

[Synapsys]

Hi SmashTV,

Thanks for the info. This makes more sense to me. I have looked at the specs for the 4051 (Fairchild) and the propagation delay from channel change to output is on the order of 1.2us with a signal in to signal out delay max at 55 ns. This means that for eight faders a delay of  about 10 usec might be expected. I don't know if a 10 us delay is so long as to cause slider overshoot. However, I have not done any experimenting with this yet so I am not sure. If the 1.2 usec per channel switch is too long than an alternative may be to use two 4051 chips connected to two AIN pins and alternate which one is begin converted and switching the other during the conversion time. This would effectively eliminate all the delays. I am currently tied up with another project but when I get more free time I will investigate this to further quantize these delays in terms of MISO scan resolution for the servos.

Regards, Synapsys :)

[Synapsys]

Hi SmashTV,

Thanks for the info. This makes more sense to me. I have looked at the specs for the 4051 (Fairchild) and the propagation delay from channel change to output is on the order of 1.2us with a signal in to signal out delay max at 55 ns. This means that for eight faders a delay of  about 10 usec might be expected. I don't know if a 10 us delay is so long as to cause slider overshoot. However, I have not done any experimenting with this yet so I am not sure. If the 1.2 usec per channel switch is too long than an alternative may be to use two 4051 chips connected to two AIN pins and alternate which one is begin converted and switching the other during the conversion time. This would effectively eliminate all the delays. I am currently tied up with another project but when I get more free time I will investigate this to further quantize these delays in terms of MISO scan resolution for the servos.

Regards, Synapsys :)