Motoman K6SB Retrofit

 

so i guess SGDRs are "dumb" amplifiers with a current feedback only. encoders connect to a separate board, not to servopack.

Hm, how does the amplifier unit know how to distribute the commanded current between the motor phases if the encoder isn’t connected to it - these servos don’t have halls or resolvers, right?

i don't have any experience with those older pwm interfaces, so it's just my speculation
i think amps usually get 3 pwm diff signals for u,v,w phases supplied by the control board. which is fed by the position data by the encoder. i believe encoders used on USA*EM motors provide A,B,Z,U,V,W diff lines - the later three should be commutation information.
i think i've seen park/clarke transform hal components in the linuxcnc docs. so it should be possible to generate u,v,w phase currents based on encoder data.
also, have a look at this thread: forum.linuxcnc.org/30-cnc-machines/20837...vo-drive-integration
it seems to me that your amps should be somewhat alike. and PCW - looks like he knows this stuff and probably Mesa has required interface hardware

The park and clarke HAL components appear, to me, to be entirely un-useful.

However, the "bldc" component knows how to calculate phase duty cycles from commutation and encoder feedback.
It can be used to drive a Mesa 3-phase PWM module, or even a software PWM.

andypugh wrote:

The park and clarke HAL components appear, to me, to be entirely un-useful.

However, the "bldc" component knows how to calculate phase duty cycles from commutation and encoder feedback.
It can be used to drive a Mesa 3-phase PWM module, or even a software PWM.
 

That’s what I had been thinking I’d use if I rolled my own from half bridges- the bldc Hal unit with encoder feedback. 

By “commutation”, do you guys mean halls/resolver on the servos, or is there another method I’m unaware of? Cursory glance suggests these servos have only the encoders for feedback, but I’m prepared to be wrong. 

"bldc" is meant to be able to work with whatever there is. For non-absolute encoders it needs to be able to move the motor (so it might struggle on a loaded Z axis) though it can then make a better guess when it sees an index.

phasefreq wrote:

Anybody by chance know the harmonic drive ratios on these arms? Or know which document would contain them? I’m told they are params in the ERC controller, but since I lack said controller that doesn’t help me much

I missed these posts before until now, what information is currently outstanding on your project? The best place aside from motoman, who dumped all their manuals for free, is the one site posted earlier in the thread,
spaz.org/~jake/robot/
this page, and they have a pile of info on their hackerspace webpage,
if you still need the gearbox/servo info let me know, its just been a while since I went super deep into a
motoman so ill have to dig around for my documents/hard copies. your model is one model older than mine but a lot of things are still the same.