Re: [CAD_CAM_EDM_DRO] LARGE Gantry table: Geckos on large servos.

Kim Lux wrote:

>Comments below.
>
>On Wed, 2004-03-10 at 11:53, Jon Elson wrote:
>
>
>>The Gecko doesn't have output filters, which causes 80 V (or whatever the
>>DC supply voltage is) square waves to be applied to the motor's terminals
>>at all times.
>>
>>
>
>I think you mean whenever the motor is commanded to move by the drive.
>You are saying that the drive controls the motor velocity/power with
>pulse width modulation and that the voltage applied is the supply
>voltage, ie 80V in this case.
>
>
>

Yes, at idle there is a square wave where 50% of the time there is 80 V
across
the motor one way, the other 50% of the time the polarity is reversed. The
motor experiences a triangle current wave of some amplitude determined
but the applied voltage and the motor's inductance (and the switching freq
of 25 KHz.)

>If the motor had 80V across it all the time, it would be moving all the
>time.
>
>

But, it reverses polarity 25,000 times a second. The net current
averages out
to zero.

>
>
>>This may cause high circulating currents at 25 KHz to flow
>>through the transistors.
>>
>>
>
>I think the only time current should be flowing through the <output>
>transistors is when power is being supplied to the motors.
>

Yes, seems reasonable, but MOST PWM full-bridge driver schemes use
the 50% duty cycle scheme.

>
>It may also cause high currents to flow just to
>the windings if the capacitance to ground is low, or if there are EMI
>filters in the motor. It might be necessary to put an inductor in series
>with the motor.
>
>
>
>I'm not following this argument.
>
>

If the motor's inductance is too small, the triangle wave current
becomes large.
The solution is to increase the inductance with a series inductor.

>
>
>>Finally, if the Gecko failed when stopping, the problem is due to the
>>back EMF of the spinning motor PLUS the extra voltage created by
>>currents in the motor inductance produces a DC supply voltage GREATER
>>than the DC supply normally produces. In other words, during deceleration
>>of large motors and heavy inertial loads, the bus voltage can rise to
>>dangerous
>>
>>
>
>I've had this argument with other people. Here is my understanding:
>
>a) the terminal voltage of a DC <servo> motor is equal to the IR "drop"
>of the armature, plus the KV product of the motor's speed.
><big snip>
>
>Therefore the terminal voltage during deceleration can never really be
>greater than the supply voltage.
>
>

This is true in a completely linear system. But, this is NOT a linear
system,
it is a switched-mode system, and inductance becomes VERY important.

So, the motor can be thought of as the ideal DC motor, as you described
above,
PLUS an inductor in series with it! Now, if you are extracting energy from
the spinning motor, you are applying the reverse polarity from the power
supply across the motor for short periods, and allowing the current to
build to a large value. Then, you reverse the polarity for a while as part
of the PWM switching cycle, so that the motor back EMF and the power
supply are of the same polarity. But, there's also that inductance with the
large current in it. As the polarity reverses, the like polarity of the
two sources (DC supply through transistors and the back EMF) would
act to reduce the large current in the inductor. AND, when you reduce
the current flowing in an inductor, you get a large voltage which acts
to raise the motor terminal voltage appreciably, as it forces a large
current back into the power supply.

>Think of the decelerating motor as a generator being turned at a
>constant speed and the deceleration being a bigger and bigger load on
>the motor. As the load increases, the terminal voltage drops.
>
>

But, if you were to short the motor for an instant, and then when the
current
became very large, break the circuit, you'd get a big spark! This is
exactly
what happens in the simple PWM servo system every PWM cycle.

These systems are always on, one polarity or the other. The PWM refers
to the duty cycle of the waveform. There are other designs that provide
for the source transistors to be off for part of the cycle, but the
Gecko 320/340
is not of that type.

Jon