Welcome to LinuxCNC.org

I picked up some cheep brushless servos off ebay.
their specs are:
Torque Constant (Kt) : 1.8 in-lbs/ amp
Voltage Constant (Ke) : 19.9 - 23.1 V/KRPM
Rated Voltage : 230 volts
Continuous Speed (no load) : 6000 rpm
Peak Torque : 18.5 in-lbs
Continuous Torque : 6.1 in-lbs
Inductance : 5.8 mH
Encoder count: 2,000 ppr
Encoder Voltage : 5 Vdc
Encoder current: 225 mA

I was thinking about using this drive:
www.a-m-c.com/download/datasheet/sx25a20.pdf
it is a sinusoidal drive that takes 2 analog inputs.
I found some cheep used ones.

What hardware would you recommend to run these?

Todd Zuercher wrote:
So, they are 10A 120V motors and that AMC drive looks like a decent fit.
How do you propose generating the 2 analogue voltages? They need to be synchronised to the rotor movement. I think that can be done in HAL with the bldc component and digital to analogue conversion (possibly PWM)
Do the motors have Hall sensors as well as the encoder?

That AMC drive appears to cost about $450, have you located a cheaper source?

Both the Pico (www.pico-systems.com/acservo.html) and mesa 8i20 (www.mesanet.com/motioncardinfo.html) are a fair bit cheaper, and are easier to integrate into LinuxCNC
The Pico drive expects Hall sensor feedback. If the motors don't have Hall sensors then there are inelegant ways to simulate them but that wouldn't be ideal.

I found some used ones for about 1/5 the new price.

My 2nd choice is probably the Mesa 8i20. But they cost more than twice the price of these used ones I found.

I was hoping that someone had tried one of these AMC drives, and that I could learn from their experiences.

The motors have a hall signal. I think it is another set of differential signals built into the encoder. They also have a temperature sensor (I need to learn more about it to see if I can make use of it.)

What other hardware do you have? I am guessing more than a parport?

I think you would want some high-quality D to A to control that drive.

I don't have anything yet. I'm making my shopping list. This is why I am asking for suggestions. I was thinking about some combination of Mesa hardware.

Todd Zuercher wrote:
The 5i25/7i77 combination might work. It has 6 analogue outputs, which could be used in pairs to control the AMC drives.
There should be enough IO to connect up the Hall sensors too.

That is one thing I was thinking, but I would need two of the 7i77s because I need 4 axis. For a total of 8 analog outs. The halls are differential signals like the encoders. Could I use the extra encoder inputs for them? If not, how do I use the regular inputs for this kind of signal?

It may be easier for me just to use the 8i20s and wait for the new Mesa 7i85 to come out.

Todd Zuercher wrote:That's a pity.
I don't really understand those drives, I don't know what they are meant to be driven by. Do AMC give any clues?
If you haven't already bought the drives then I would hold off until compatibility is clear.
A single 7i65 combined with a 5i20, 5i23 or 7i43 would be another option.
You could use extra encoders for them. What you would do is tell LinuxCNC to only enable 4 encoder modules in the firmare That would leave the pins that would have been allocated to encoders 5 - 8 available in the HAL as GPIO pin, but they would be physically connected to the differential transceivers on the 7i77 boards, allowing you to connect the Hall sensors.
It is something of an unknown whether the 7i77 analogue outputs are suitable for creating commutation signals for those drives. They might be too slow, for example.
Maybe, though you might still need some advice from Pete about how best to connect the diferential signals.

The 7I77 analog outputs have a ~2 KHz bandwidth so should be OK unless you are talking
about really high servo speeds (say >3600 RPM on a 4 pole motor = > 120 Hz sine waves)

Note that both the 7I65 and 7I77 encoder inputs are multiplexed inputs so reading the GPIO
bits will just return bit-hash since the GPIO pins are a combintation of both the even and odd
encoder input pins.

The demuxed A,B,I pins are available from the firmware but the hostmot2 driver does not present them to HAL

The differential Hall signals can likely be read using single ended inputs (like the 7I77 inputs) or OPTO inputs (like a 7I37) as they do not need much speed

another option is to ignore the Halls and use index rotor alignment

How does one wire the differential hall signal to a single ended input?

Since two of the axis are fighting gravity, I thought it would be better to use the halls.

1500 rpm will be fast enough unless I decide that I need to add more gear reduction.

So lets try to compair some of my options:

4 AMC s25a20 drives with 2 7i77s and 5i25
Should have plenty of IO to handle hall sensors
May have dificulty setting up generating 2 sine wave analog signals/drive.
Cost: $90(4) + $398 = $758
4 8i20 drives, with one 7i85 and 5i25
the 8i20s should integrate into the Mesa hardware and LinuxCNC easier?
The 7i85 is not avalible yet (maybe a month or more away)
May have to add more IO to handle hall sensors
Cost: $239(4) + $79? + $89 = $1124 (more if more IO is needed)

I didn't include costs like power supplies, ect. since those costs should be the same regardles of which these 2 paths I take.

Have I over looked any big stumbling blocks?

Any one see better options?