Re: [CAD_CAM_EDM_DRO] Re: Update on closed-loop to the computer with Gecko 320

Mariss Freimanis wrote:

>Jon,
>
>Looks interesting. You can opto-isolate the encoder to the G320. This
>allows you to:
>
>1) Make the encoder outputs available to be read by the PC.
>2) Allow the PC +5V to power high-current (>50mA) encoders.
>
>Use a dual high-speed opto-isolator (HCPL-2531). Place the opto LEDs
>as pull-up loads on the encoder channel outputs. Bias them at 20 to
>25mA to permit a 5mA collector current needed for the G320 (the HCPL-
>2531 has a transfer gain of 20%).
>
>

Thanks for the info. I ordered some HCPL-2530 isolators, will have to see
how they do when I get them. The magnetic isolator I tried was a CMOS-in,
CMOS-out 2-channel device, but with WAY more bandwidth than I needed.
To use an optocoupler, I need a current source. Some encoders have enough
drive to power the LED in an optocoupler and still meet the CMOS levels
needed by my FPGA, but I couldn't count on all encoders working properly
in that arrangement. Well, I think I'll have to put a 74HC14 between the
encoder and the LEDs in the opto to be sure all encoders will be able to
drive it.

>Place both emitters on ENC-, take the collectors to the CHANNEL A, B
>inputs. Have a 470 ohm resistor from ENC+ to ENC-.
>
>

Huh? Did you mean a 470 Ohm R from ENC+ to the A and B inputs?
That must be what you meant there.

>I assume you have done something like this in your scheme. If you
>feel you are up to modifiying SMT components (a low-power microscope,
>tweezers and a hot-air rework station helps), I can send you a
>circuit with landmarks to modify the drive to have a proportional
>loop response only. Maybe that will help.
>

I might try that sometime. But, I seem to have gotten a very smooth and
stable response with what I'm doing now. Something I'm still studying
is a very asymmetrical response of the Gecko 320. It seems to be much
stronger and more stable in one motor direction than the other. The setup
I am using at the test bench has a 1.8 A lab power supply with a 3900 uF
capacitor to handle the current peaks. This may not appear stiff enough,
and a fluctuation of the supply voltage may upset the I and D terms of
the servo loop. Any thoughts on this?

(Something that I learned when developing my PWM servo amp about 5
years ago was that gain is proportional to power supply voltage. When
you look for the cause of this, it becomes completely obvious, but was
not something that I anticipated from working with linear power amplifiers,
which don't behave that way.)

As for SMT, I'm doing stuff much harder than anything in the Gecko drives.
My USC board has a 144-pin FPGA with a 0.5 mm lead pitch! Now, that
is fine spacing. I also use some other chips with that lead pitch, and some
SSOP chips, too. But, for a commercial product, I'd rather not have to
hack the innards of the Gecko products, just use them as is.

Something else I had considered, if the loop stability proved to be
difficult,
was to turn the gecko 320 into an open-loop servo amp, by not connecting
any encoder to it. I could then just send step & direction signals to it,
moving the commanded position within a range of +/- 127 counts, to
drive it between the range of the P term. (Here, I think the removal
of the I and D terms might be needed.)

Jon