[gr0kr]

I am working on converting a CNC mill to LinuxCNC and am trying to decipher all the necessary interfaces. In particular, I am trying to get more info on the servos.

Known:

1-Axis 200V S-Series Servo Amplifier Module: A06B-6058-H006
* Lower Board: A20B-1003-0080/04A
* Front Board: A20B-1003-0090/02/7B

20S Motor w/ Incremental 2500 PPR encoder: A06B-0502-B005#7008

Cable K1 (NC Controller to Amplifier) Lists the Pins (GND and COMx Ignored):

• PWMA (ALM1)
• PWMB (ALM2)
• PWMC (ALM4)
• PWMD (ALM8)
• PWME
• PWMF
• IR
• GDR
• IS
• GDS
• MCDN
• DRDY

I'm guessing DRDY is Drive Ready. Other than that, I am having a hard time tracking down the control protocol and more details on the pin descriptions.

It appears that IR/GDR and IS/GDS are RS-422 pairs. But that still doesn't tell me what IR/IS are for?

The Encoder also lists the expected A/B/Z pins, but it also lists:

• CN1
• CN2
• CN4
• CN8
• OHA
• OHB

From a previous post, I gleamed that CNx provides Commutator Orientation. Though I don't quite know how it is arranged or what the pulses look like.

OHA and OHB look to be a RS-422 style signal also. Is this for Overheat?


Request: Can anyone enlighten me with details on these pins and how they are typically driven?

Any insight/pointers/info would be greatly appreciated!

[Al_The_Man]

The OHA and OHB are an internal thermal switch (contact closure).
The Encoder commutation attached.
The problem is that the commutation pulses do not form normal pulses for converting to BLDC directly, without some extra circuitry for these 8 pole motors.
Al.

[gr0kr]

Thanks Al!

The OHA and OHB are an internal thermal switch (contact closure).
The Encoder commutation attached.

That helps a lot! I was guessing some kind of binary counter based on the naming, and that diagram explains it perfectly.

The problem is that the commutation pulses do not form normal pulses for converting to BLDC directly, without some extra circuitry for these 8 pole motors.

This by itself doesn't scare me as I can process it via LinuxCNC / Anything-I/O FPGA. I just still don't know what kind of inputs the amplifier expects. I suspect reading up on 8-Pole AC Motor Control Theory might shed light onto the 6x PWMs (any hints appreciated), but the other pins and suggestion of Bi-Directional data (binary alarm code response?) still boggles to me.

[gr0kr]

One theory that I made up:

• IR and IS are Read and Send signals (It doesn't make sense that there would be two lines for this).
  
• The PWMs are for a Delta (or possibly Wye) half-bridge configuration and represent each gate-set's individual control.
  
• When IR is high, PWM's A-D become outputs which can be used for state feedback.

Any yeh/neh's on whether this is roughly close or perhaps insight into where I can dig to confirm/deny this?

[cnc2149]

The IR and IS are not Read/Send. They are actually an analog representation (+4.5V to - 4.5V) of the current feedback. The motor has 3 phases (R,S,T) in FANUC nomenclature. The IR represents the current in the R phase. IS represents the current in the S phase. They algebraically calculate the IT phase current. The current is sensed in the amplifier with shunt resistors or CTs in 2 of the 3 motor phase leads. It is conditioned and sent back to the controller as the IR/IS signals. Since all 3 loops (Position, Velocity, Current) are closed in the controller, the feedback of all of them must read by the CNC.

Hope this helps....

[gr0kr]

The IR and IS are not Read/Send. They are actually an analog representation (+4.5V to - 4.5V) of the current feedback. The motor has 3 phases (R,S,T) in FANUC nomenclature. The IR represents the current in the R phase. IS represents the current in the S phase. They algebraically calculate the IT phase current. The current is sensed in the amplifier with shunt resistors or CTs in 2 of the 3 motor phase leads. It is conditioned and sent back to the controller as the IR/IS signals. Since all 3 loops (Position, Velocity, Current) are closed in the controller, the feedback of all of them must read by the CNC.

That helps a lot! And it's nice to know there is current feedback available.

Can you confirm/deny my theory on the PWM control?

With my Read/Send theory gone, and assuming that alarm codes are sent over PWM A-D at some point. When should I read those? When MCDN is low?I

Is this a reasonable summary of the interface?:

• PWMA/comA - R High Side Input [RS422]
• PWMB/comB - R Low Side Input [RS422]
• PWMC/comC - S High Side Input [RS422]
• PWMD/comD - S Low Side Input [RS422]
• PWME/comE - T High Side Input [RS422]
• PWMF/comF - T Low Side Input [RS422]
• DRDY/spare - Drive Ready Output [RS422]
• IR/GDR - R-Side Current Output Feedback [Analog +/- 4.5V]
• IS/GDS - S-Side Current Output Feedback [Analog +/- 4.5V]
• MCDN/GND - Amplifier Enable (active High or Low?) Input [RS422]

I'm just trying to get as clear of an idea before wiring it up. Depending on the fault-detection circuitry, I can probably probe out the PWM to R/S/T associations.

Cheers!

[andypugh]

The problem is that the commutation pulses do not form normal pulses for converting to BLDC directly, without some extra circuitry for these 8 pole motors.

The LinuxCNC "bldc" component can use the Fanuc 1/2/4/8 encoder signals directly, and can cope with any pole count.

Also, the same component can output the required 6 PWM amplitudes for a 3-phase PWMgen, and there are 3-phase PWM generators in some of the Mesa firmwares.

[Al_The_Man]

Are they utilizing the C1-C8 pulses for BLDC commutation rather than the original Sinusoidal method?
Any link to the explanation?
Al.

[andypugh]

Are they utilizing the C1-C8 pulses for BLDC commutation rather than the original Sinusoidal method?.

It would typically use the C1-C8 pulses for initial commutation, then use the first transition as a reference to switch to sinusoidal commutation based on the incremental encoder value.