Re: [CAD_CAM_EDM_DRO] Interesting Gecko project

Nicolas Benezan wrote:

> > It also seems to have some filtering and optoisolation for the encoder
> > signals. It took me a while to realize the board on top with the SMT chips
> > and discrete components is the Gecko Servo board without the case.
>
> No, it is not the gecko board, it's my own. I think I should come up
> with a real translation of the whole page, but for now, let me explain
> in a few words what the "GeckoFilter" does:
>
> 1. It does some filtering of the supply and motor lines (additional
> decoupling capacitor on the supply line, differential choke on the motor
> line to filter HF common mode noise plus some snubber RC networks). This
> is the board on the bottom.
>
> 2. The sensitivity of the encoder signals to noise problems is greatly
> reduced by transmitting them differentialy (twisted pair instead of a
> single wire for each phase A and B). This feature is implemeted with a
> small driver board which is directly mounted on the encoder (last
> picture) and a differential amplifier/receiver (left half of the top SMD
> PCB).

I made my own PWM servo amps, and did put differential 2-pole
LC filters on the output. I also have several levels of capacitors on the
DC input, enough that I have to ramp the DC bus up with a 2 relay and
resistor scheme to prevent tripping the high-side current sensor.

I had single-ended encoders, and bought the chips to convert to differential
right at the encoders, but never used them, as I have never had any problem
with encoder noise. I use shielded cable for the encoders, however.

The PWM servo amps are switching 69 V at 100 KHz with fast FETs, so
they should be a potent noise source. But, I can even run an AM radio
near the CNC system.

> 3. Easier driving and polling of the ERR/RES pin. Instead of one
> tristate bidirectional signal, the ERR/RES signal is split into two
> seperate ENABLE (to gecko) and STATUS (from gecko) signals.
>
> 4. The resolution and accuracy of a limit/reference switch is enhanced
> by logically combining the reference switch signal with the index output
> (one pulse per rev) of the encoder.

Many systems handle this in the encoder counter. both the Servo to Go card
and my PPMC hardware do this. The home switch is read by a digital I/O
section, and the machine slows down and starts looking for the encoder
index on each servo cycle.

> > Wow, a pond sounds like a big project!
>
> Huh? What do you mean with that?

Oops, sorry! I think the guy who wrote ABOUT your project must be the
one who is making the pond.

Thanks for your comments on this project. The G-code interpreter in an
embedded processor sounds like a BIG piece of software, if you have
smoothing and cutter offset compensation usable at the same time.

Jon