Current or Voltage Mode

[knji]

I have a servo amplifier from Advanced Motion Controls capable of operating in Current or Voltage Mode. I will be using an external controller to regulate the position of the motor shaft so no feedback is sent to the servo amp.

I need to be able to model the servo amplifier mathematically for control purposes. My servo amplifier is connected directly to the motor and in an open loop fashion, i.e I am not supplying to it any current feedback for I assume this is done internally.

With this in mind, I am modelling the servo amplifier simply as a linear gain V_out = Ka*V_in where V_in is the voltage fed to the amplifier and V_out is the amplified voltage fed to the motor. This is assuming that the servo is operating within its linear range.

Is this a good approximation of the servo amp or do I have to include the internal current loop gain circuitry?

TIA,

Klaus

 

[Skogsgurra]

You can get a very rough simulation doing that. And if that's all you need, it's fine.

But if you need any precision in the results you will need to include the dynamics of the electronics and the motor. And also include current limit influence - you can have a very fast simulated servo if you do not include the current limit. But not so fast IRL.

 

[knji]

Thanks for the response. I'll need better precision than what I am currently getting in order to design my controller.

I am including the motor dynamics into my simulation. Actually, I am coupling the dynamics of the motor to that of the servo amp but assuming a linear relationship from V_in to V_out at the amp. The resulting model takes in a supplied voltage V_in and outputs motor shaft displacement q_m. Motor armature receives V_out as input.

When I simulate the mathematical model and compare the results with real data for the same input voltage, the pattern is same. However, the amplitude in from my math model is always larger than that from the real model.

I'm taking viscous and Coulomb friction into consideration since the motor's shaft is connected to a bearing. I am also including the motor's internal viscous and Coulomb friction provided by the manufacturer. So, I am thinking that the effective real voltage available at the motor terminal is not exactly equal to Ka*V_in, the linear model I used for the servo. I am also assuming that saturation is not reached.

When you say that I should take the electronics into consideration, does this include the internal current feedback loop? If so, can I simply apply the relationship

V_out = Ka V_in - Ka Kc Ia ?

Ka represents the amplifier gain, Kc the internal current feedback loop gain and Ia the armature current.

TIA

Klaus

 

[Skogsgurra]

No, you have to include the limiting action as a condition. Like this (in some kind of pseudo-code).

constant Current_limit
input = Armature_current

if ABS(Armature_current) > Current_limit
then Armature_current := Current_limit*SIGN(Armature_current)
endif

The electronics can usually be represented by a first order function. I.e. a simple time constant - usually in the 1 to 3 ms range for the AMC DC servo amplifiers.

 

[sreid]

Voltage mode and current mode are modeled entirely different.

Start with a unity gain voltage power amplifier and do a Bode plot. At low frequency the motor velocity will essentually follow the input voltage. At the motor mechanical break frequency, the velocity will start to roll off. At the motor electrical break frequency, the velocity will start an additional roll off.

With a current amplifier (say unity gain, 1 A/V), the motor will accelerate in response to the command. The amplifier/motor combination act like an integrator with the motor velocity amplitude linearly decreasing with frequency. And this will continue up to the current amplifier's bandwidth which can be made very high .

 

[knji]

Thanks for the response guys. It seems as though I should instead be working in current mode given the much simplier relationship from armature current to torque (shaft position).


sreid, YGM.

TIA