[justCNCit]

Hi, I was wondering how servo drives work. I am familiar with the principles of stepper motor drives, where an IC is used to provide phase signals to a transistor of some sort(usually a mosfet). Of course this is done with the required regulations in place, such as voltage & current regulating components on the board.

However with servos I understand the voltage may change from like 900v to nothing in a millisecond. I see a german guy has commercialized a board kit to drive a servo, but I am curious as to what it is that controls the current.

There must be some device in there that is able to control current and volts to a point where it can bring the voltage up to 900 v in a matter of milliseconds & current to full and then back to zero in the same matter. How is this done? I understand Mosfets are meant to switch current on and off in stepper setups, however with servos I am at a loss.

I realize I have not done much research into this subject and appreciate very much any replies that are made in advance.

[pointcloud]

Look at this... It will give a little insight to your question.

http://www.geckodrive.com/photos/G340_REV-4_MANUAL.pdf

There are many other dirves and they work a little diferently... But for the most part that are the same...

[epineh]

I am not too sure where you got the 900V from, but here is my version of what you are after

A servo uses an encoder to provide feedback to the controller to allow precise monitoring of the servo's position. The controller reads the encoder position and "holds" the servo position to the setpoint value. So lets say the machine is sitting still, you don't want the servo to move but you still want it to hold its position until told otherwise. What the drive will do is to use what is called "locked anti phase" PWM to do this. Stick with me, it isnt as scary as it sounds. Imagine a square wave signal, or Pulse Width Modulation wave signal, you will have full DC positive applied for half the wave then full DC negative for the other half, now picture this happening 20 000 times a second, which is pretty usual to have a 20 KHz frequency.

So the servo (basically a good quality DC motor) is effectively "rocking" back and forth but the end result is that it "holds" its position. Now the encoder is helping here, as it detects a little movement, it tells the controller which in turn corrects, check position, corrects, and so on. If you grab the servo shaft and try to turn it against its will, it will sense this and "fight" you for its position. If you force it past its position by the preset following error, the drive will fault and give an alarm.

So when you want the servo to move to a certain point, you change the setpoint and the drive gives, say 80% to the positive signal and 20% to the negative, the result is more positive than negative and the motor turns.

Now all of this is variable, I am only giving you a concept, the servo is tuned to the drive and machine it is running, if the machine is told to rapid, then the PWM signal will be as close to 100% in whatever direction as the drive allows, usually not quite 100% as the bootstrap caps need time to charge. So the effect is to apply DC to the servo and it shoots off in whatever direction you told it to, you can imagine that it will have a "top speed", this will be known or worked out during machine commisioning and the software will not exceed this and avoid following errors.

As for current limit, a servo is generally a brushed DC motor of sorts so in general use will not need to worry about excess current, as mentioned during rapid's full DC is applied to the motor and the motor rated voltage is usually what the drive can supply or fairly close. What can happen though is the servo can stall for whatever reason, hitting a hard stop etc, this is when current limit is needed, there are a few methods, I don't claim to be an expert here but basically you can use a sense resistor to measure current, by monitoring the voltage across the resistor, which is in series with the motor and calculating the motor current. When this value gets too high the drive can send a fault signal back to the PC.

I have heard mention of other forms of current limit, namely "highside" current limit, which from what I gather doesn't use sense resistors, I am not too sure how this is done, possibly by monitoring the Vce Sat voltage across the output MOSFET's, as I said, I am no expert, but I hope this gives you a bit of an insight.

I am sure someone will add to/correct this but at least it is a start

Cheers.

Russell.

[justCNCit]

So then, the controller uses MOSFETS to switch current on/off, and can do this fast and precisely enough to close/open the circuit for a percentage of the 1/20,000 part of a second?

I then gather that the voltage regulates itself but using IC's you are able to control the MOSFET at a speed where this can be done?

[epineh]

Not so much switch on/off but forwards/backwards, imagine you had a switch that applied either full motor voltage to make the motor spin flat out in one direction, and if you flick the switch it makes tha motor spin flat out in the other direction. If you wanted it to stay "still" you would have to toggle the switch very quickly in both directions, the motor would spin a little one way, then the other with the shaft averaging out to staying "still".

If you wanted the motor to spin in one direction you would just leave the switch to one side.

Now the electronics does this a little better than we can with a switch, and coupled with an encoder does it very accurately, a servo will appear to be perfectly still even though it is being "switched" backwards and forwards very quickly. It also allows for varying waveform (PWM) to vary the speed from standstill to full speed and everything in between.

Russell.

[justCNCit]

thanks for the explanation, thats what I was looking for, as well as the PDF.

To prod further, how I understand it the MOSFETS are able to switch current on and off in the matter of a percentage of 1/20,000 of a second? These in turn are controlled by some kind of microcontroller?

[epineh]

Yup, there are many different types of drivers, generally with some kind of "brains" micro's, FPGA's dedicated hardware, you name it.

You can also have the smarts done on PC, EMC will output PWM directly from the parallel port, add a basic H Bridge or two (or more ) and you have an instant servo drive !

I am currently in the process of doing just that, when time permits.

Russell.

[Al_The_Man]

With a modern servo using torque (current) control, the drive current output is directly proportional to the voltage input, With this method, if the motor is commanded to stationary position, and if the encoder indicates that it is in co-incidence, there is no need to output any command voltage, hence no current will flow, all outputs are essentially off.
Unless something attempts to move the servo.
The 50% on 50% off technique was used primarily in velocity drives in order to hold the servo at zero position as the control was weakest around the zero point.
That is why on older systems you could hear the servo's 'singing' at rest, modern servo's are much quieter both at rest and in motion.
Al.