Re: [CAD_CAM_EDM_DRO] Re: Gecko 320 VS. 340

mariss92705 schrieb:

> Right now it functions just like the G320 except there is a noticable
> discontinuity or "tick" on every Hall sensor edge (every 60 degrees).
> This is something I would like to attenuate or eliminate altogether.

This is caused by the unmatched delta voltages when switching the
windings. Say you have 70V supply and 20V average (PWM) motor voltage.
When switching, the switched off winding (#1) is free wheeling what
results in an effective (clamping) voltage of -70V. The other winding
(#2, switched on) gets +20V average from the PWM output.

The motor windings behave like inductances, the rate of current change
is proportional to the applied voltage. So in winding one the current
drops more quickly than it rises in winding #2 what causes torque ripple
"ticks".

This could be avoided by clamping the switched off winding to zero volts
(short circuit) instead of letting it free-wheel (open circuit clamped
to supply rails). But this works only if the motor has a perfectly
symetrical rectangle waveforma, most motors haven't.

> One promising idea that has ocurred to me is to phase lock a ramp
> function off of the hall sensor edges. This ramp then would be used
> as an attenuation function for the "off" winding. Rather than
> abruptly switching a winding "off" or "on", the current would be
> gradually applied or removed, eliminating the discontinuity. The
> behavior should then approach what one would expect from a
> sinusoidally commutated motor.

I once had the same problem. I solved it by sampling the waveforms at
the motor windings when the motor is driven externaly (as generator). I
stored the waveform data in an EPROM or flash device with the encoder
counter feeding the address input of the EPROM. Of course, the counter
has to be synchronised either with the encoder index output or with the
hall sensors. The EPROM output (one for each phase) is then multiplied
with the PID control output to get the PWM input values. This gives a
perfect match of the PWM output voltage and the generator voltage of the
motor. The motor runs very smooth with no ticks or torque ripple.

If your EPROM has more address space than you actually need, you can
even store different waveforms (sinusodial or trapezoidal) and select
the address space with jumpers. The disadvantage is that your waveform
data has to match the encoder pulse count and the pole count of the
motor. If you had a microcontroller with ADC in the gecko, you could
implement something like a self-learning-mode, but I think that would
make it too expensive and to complicated.

I know... The idea of the gecko drives is that they are simple and have
no microcontroller. Multiplication of the waveform data and PID values
is of course easily done with a processor and difficult with analog
circuitry, but not impossible. One trick is to vary two parameters at
the PWM simultanously: comparator voltage AND gradient/amplitude of the
sawtooth generator. Think about it.

Ciao
Bene

--
!!! Achtung, neue Adresse !!!
Benezan Electronics, Maria-Merian-Str. 20a, 70736 Fellbach, Germany
Kundenspezifische Elektronikentwicklung, Sondermaschinenbau