Reducing servo signal 'granularity', is it possible?

E

elanman99

Senior Member
I know the actual mechanical/electronic resolution of a standard model servo is the ultimate limit to controlling its resolution but in order for it to use that resolution the Servopos or Pulsout signal has to have equal or better resolution.

With a 10bit ADC in the Picaxe there are 1024 discrete steps, however using values over the range of 100 to 200 for the Pulsout command means there are only about 100 increments of pulse width being sent to the servo.

Have I missed something really obvious? For the project I am doing I would like to get the best possible performance. I was considering using a really expensive brushless digital servo with Hall device encoder, but now I not sure if it would be any benefit.

Thanks

Ian
 
G

goom

Senior Member
I believe that one way is to increase the speed of the Picaxe. At 4Mhz, to get the 1 to 2 ms pulses, one would use pulsout values of 100 to 200. However, at 16MhZ, pulsout would need to be 400 to 800 for the same 1 to 2 ms output, i.e 4x better on granularity.
 
E

elanman99

Senior Member
bluejets said:
Digital servo....no deadband...would have to be an improvement i would think.
Click to expand...
Servo with no deadband? I would like one of those!

I wanted to get really smooth movement, but with only 100 steps from end to end it will be really jerky, especially as my servo is one I made myself with 8 turns end to end (bit like a sail servo)

Ian
 
M

MikeM100

Member
A idea ! Rather than use model servo electronics with PW output from PicAxe why don't you make your own digital feedback servo system ?

Using 10 bit A-D inputs for both the manual control pot and the feedback pot - compare the two to derive a > , < , = to, 'error' signal to drive a bridge output. You can then set your own 'deadband'. Probably a bit more complicated as may possibly need a PID control loop ?
 
E

elanman99

Senior Member
MikeM100 said:
A idea ! Rather than use model servo electronics with PW output from PicAxe why don't you make your own digital feedback servo system ?

Using 10 bit A-D inputs for both the manual control pot and the feedback pot - compare the two to derive a > , < , = to, 'error' signal to drive a bridge output. You can then set your own 'deadband'. Probably a bit more complicated as may possibly need a PID control loop ?
Click to expand...
I am running out of time for this project but its an idea that is definitely worth following up. I suppose it would need the fastest Picaxe chip because servo and feedback systems usually need as much horsepower as you can afford!

Might be better to use some sort of digital encoder for feedback rather than a pot?

Ian
 
A

AllyCat

Senior Member
Hi Ian,

elanman99 said:
Have I missed something really obvious? For the project I am doing .....
Click to expand...
You haven't said what you're actually trying to achieve or why you need high resolution.

But perhaps you should be considering a (geared) stepper motor?

Cheers, Alan.
 
E

elanman99

Senior Member
AllyCat said:
Hi Ian,

You haven't said what you're actually trying to achieve or why you need high resolution.

But perhaps you should be considering a (geared) stepper motor?

Cheers, Alan.
Click to expand...
'High resolution' is a vague definition and means little without a proper specification. What I am building is a lens focus servo for a camera in an underwater housing, the cameraman will control the focus using a pot and knob. There are ready made (really expensive!) systems available for pro camera gear but this is a one-off camera in custom housing and has limited physical space. The other complication is that four different lenses will be used and each one has a different diameter focus ring in a different position, and each lens focus ring has a different amount of rotation to cover the range.

I have made a custom servo with a geared motor, 10 turn potentiometer, and found a way to couple it to the different lenses. I want the cameraman to have the finest, smoothest control of focus as with luck the results might be used for a HD TV film and especially for HD, good focus is paramount.

If I had the money (and skills) I might have used a Swiss precision brushless motor and optical encoder with the manufacturers own servo driver PCBs, however I am hopeful that if I can get the best out of the available hardware (servo, Picaxe etc) that it will be good enough for the purpose and I can say I have done it myself (not true, I have had lots of help from this forum!)

Ian
 
E

elanman99

Senior Member
Not trying to be pedantic but some deadband must exist. For aeromodelling purposes it would be too small to be important and Hitec are quite reasonably calling it zero. Their documentation is pretty sparse as regards containing any real numeric values though.

Ian
 
A

AllyCat

Senior Member
Hi Ian,

You may be confusing a (model maker's) "servo" with the more generic "servomechanism", "servo loop" or just "feedback control loop". "Continuous Rotation Servos" (the same as your winch servo?) don't even use feedback control (or servo action), so fundamentally must have a deadband. An analogue servo loop can certainly have a "zero" deadband, but it may "jitter" or "dither" due to electrical noise.

If your cameraman is using a "pot" to set the focus then it isn't really even a closed loop system (only his eye-brain-hand coordination closes the loop). Presumably you actually need the servo/control loop to achieve "repeatability" after changing lenses?

If you basically have a reversible motor and a position sensor with A/D conversion to detect if the lens position is "correct", then the positional accuracy will be (at best) the resolution of the A/D converter. That is 1:1024 (10 bits) with a PICaxe, although there may be a few "tricks" to try, or use a better external converter.

However, there are already various "macro rail" (lens control) projects on this forum and I believe that they (all) use a stepper motor (connected to a "lead screw"), because that provides the potential for much higher resolution and repeatability with the simplest and cheapest hardware.

Cheers, Alan.
 
E

elanman99

Senior Member
Alan

As far as I understand it, the term servo is mostly used on this forum to refer to a model servo. One that has been modified for continuous rotation is still physically a servo but is somewhat open loop. It has become a reversible variable speed motor and as you say, it has to have some deadband to make it usable.

The servo I have put together uses the PCB out of an analogue model servo and I suppose is like a 'sail' servo, I'm not sure whether a winch servo is the same thing as presumably a winch servo is just a reversible continuous rotation motor.

In my project, the cameraman is the ultimate feedback device. He adjusts the focus using his eyes as the input sensor and his fingers rotate the pot. What he needs is for the pot knob to feel as if it is actually turning the lens focus ring. I know 10 bit resolution is the limit of the Picaxe (input side) but my original enquiry was about increasing the output resolution which on my first understanding as only about 7 bits.

I think stepper motors are used on macro rails for purely practical reasons (cost, availability, ease of use, etc) but they are no more inherently capable of better resolution than a geared motor servo system. My only experience was with a macro rail that used a brushed motor without any feedback to drive the carriage on the track with a toothed belt rather than a leadscrew.

Just re-read your post but not sure whether you were referring to track movement or lens control by steppers? I doubt any commercial 'follow focus' servo systems that operate 'FIZ' on the lens use stepper motors, they would not be fast enough, unless large and heavy, but worst of all they would be be too noisy.

Ian


(FIZ = Focus-Isis-Zoom)
 
R

rossko57

Senior Member
I would think an advantage of stepper motor drive is having position control without any form of position sensor.

Here's an old idea, sometimes used for iris controls and fluid valves; a mechanism is sprung in one direction, and pulled the other way by a solenoid coil. Driving the coil with PWM allows position control, again with no sensor. There is a penalty in continuous power consumption, and it collapses to one end at power off, which may not be acceptable in every application.
 
You must log in or register to reply here.
Top