Servo config, Mesa, H bridges, current sensing?

I just ordered a Sieg SX3L mill which I want to convert to CNC with servo motors.
I like to experiment and EMC2 with servos will let me have full software control and visibility of the position feedback loop.
The plan for the electronics would be as follow.

- Mesa Anything I/O board
- 4 H bridges
- Home, limit and e-stop switches
- eventually spindle control

This brings several questions, advice and feedback is most appreciated!

- Should I put everything in the computer tower? This may mean a separate power input for the
unregulated servo motors power supply. However, it would allow short 50 pins cables to individual
connectors for encoders, motor control, switches...

Or should I have a separate CNC control enclosure? In that case, how do I bring out the 2 or 3 50 pins
cables between the computer and the CNC control (ribbon cable to connector on the casing,
SCSI 1 external cables between the two enclosures)? A neat setup could be a MESA 5I71 PCI to PCI Express
bridge in the computer connecting to a 3X20 in the CNC control enclosure with an external PCI express cable.

- The servo motors are rated 4A continuous 60V/20A peak (X and Y) and 7.6A continuous 90v/38A peak (Z).
If I understand correctly, the PWM and direction outputs of the I/O board could control the motors through a H bridge.
Mesa has several such H bridges. Given that I should get a 50V unregulated power suppply, which model should I take.
The Mesa 7I29 Dual 2KW H-bridge is pricey but should be up to the job; its current rating at 22.5A is lower than the peak
but much higher than the Z axis continuous current. The Mesa 7I40 dual 400W driver looks too small at 7A continuous,
10A peak and the price difference is small.

- I would like to have motor current feedback. This would let me monitor the current and detect problems
(overloaded or stalled motors...). This should be possible by sampling into a AD converter the voltage across
the H bridge current limit sensing resistor. The AD converter would then be connected to GPIO pins of the I/O board.
Do you see a better way?

mdagenais wrote:

- Should I put everything in the computer tower? This may mean a separate power input for the
unregulated servo motors power supply. However, it would allow short 50 pins cables to individual
connectors for encoders, motor control, switches...
Or should I have a separate CNC control enclosure? In that case, how do I bring out the 2 or 3 50 pins
cables between the computer and the CNC control (ribbon cable to connector on the casing,
SCSI 1 external cables between the two enclosures)? A neat setup could be a MESA 5I71 PCI to PCI Express
bridge in the computer connecting to a 3X20 in the CNC control enclosure with an external PCI express cable.

I have used about 2m (6 feet) 50-pin ribbon cables from the PC to a 19" rack enclosure where the PSU and servo-drives are. I did not pay special attention to noise shielding or anythng and have had no problems. For the high-voltage drives (servos and VFD) I've used isolated inputs/outputs, but not for low-voltage stuff (limit switches, e-stop, jogwheel, etc)

- The servo motors are rated 4A continuous 60V/20A peak (X and Y) and 7.6A continuous 90v/38A peak (Z).
If I understand correctly, the PWM and direction outputs of the I/O board could control the motors through a H bridge.
Mesa has several such H bridges. Given that I should get a 50V unregulated power suppply, which model should I take.
The Mesa 7I29 Dual 2KW H-bridge is pricey but should be up to the job; its current rating at 22.5A is lower than the peak
but much higher than the Z axis continuous current. The Mesa 7I40 dual 400W driver looks too small at 7A continuous,
10A peak and the price difference is small.

I've used the PWM amps from Jon Elson, worth a look also. (www.pico-systems.com)
Your power supply needs to be pretty hefty if it is going to deliver 20A peaks (lots of capacitors!).

- I would like to have motor current feedback. This would let me monitor the current and detect problems
(overloaded or stalled motors...). This should be possible by sampling into a AD converter the voltage across
the H bridge current limit sensing resistor. The AD converter would then be connected to GPIO pins of the I/O board.
Do you see a better way?

A simple overload/stall protection would be to just look at the PWM output and if you ever see 100% or -100% then do an e-stop. The PID loop is supposed to be in control of the motor and if you are maxing out on commanded PWM you are not in control anymore.

IRF has a chip (IR2175) which looks at the voltage drop across a sensing-resistor and produces a variable frequency (or PWM?? can't remember) output. This could be wired into a single logic I/O pin and is much simpler than an AD interface.
Beware that the current signal is going to be very noisy. You will probably want some filtering on the analog side and also some in HAL after you have read the current into EMC2.


Anders

I have used about 2m (6 feet) 50-pin ribbon cables from the PC to a 19" rack enclosure where the PSU and servo-drives are.

Did you ran cables from the internal connectors on the PCI card inside the PC directly to the control enclosure, or you used a connector screwed to the PC case, an internal cable and an external cable between the PC and the control enclosure.

For the high-voltage drives (servos and VFD) I've used isolated inputs/outputs, but not for low-voltage stuff (limit switches, e-stop, jogwheel, etc)

The PWM amplifiers (mesa or pico-systems) already provide opto-isolated inputs for PWM and DIR signals, did you add other isolation? I understand that switches are passive devices and need not isolation as much.

I've used the PWM amps from Jon Elson, worth a look also. (www.pico-systems.com)

Interesting. The mesa PWM amplifier provides encoder input massaging (voltage, protection, filtering, unipolar versus differential), is that useful? The pico-systems PWM amplifier offers a nice framework for e-stop (not fault in/out) and more control over the current limit. How did you decide on the pico-systems?

A simple overload/stall protection would be to just look at the PWM output and if you ever see 100% or -100% then do an e-stop. The PID loop is supposed to be in control of the motor and if you are maxing out on commanded PWM you are not in control anymore.

Good point, with the motor speed and PWM duty cycle, one can estimate the current/torque in the motor.

IRF has a chip (IR2175) which looks at the voltage drop across a sensing-resistor and produces a variable frequency (or PWM?? can't remember) output. This could be wired into a single logic I/O pin and is much simpler than an AD interface.

Good trick, many thanks!