Help picking mesa board redcaps/bldc component

I am going to be helping a friend retrofit a Kryle VMC to linuxcnc. The mill has fanuc redcap motors. The plan is to use AMC servo drives (BE25a20AC). The commutation signals from the motors are 4 bit fanuc encoded and will need to be translated to industry stand 3 wire signals for the AMC servo drives. Andys BLDC component does the translating function nicely but with 4 or 5 servo motors it entails quite a few I/O lines needed.

So, each motor has 4 commutation outputs, times 5 motors means I will need 20 high speed input lines into linuxcnc. After translation by the BLDC component, I will need 3 commutation outputs x 5 motors totaling 15 high speed outputs. I say high speed because with a servo max rpm of 3000 the commutation signals will be fairly high frequency.

Question is, which mesa daughterboard would be best suited to handling the 20 inputs and 15 outputs needed for commutation signal translation? We already have the 5i25/7i77 combo for encoder inputs and general I/O, so a compatible daughterboard would be needed.

Thanks much ! Its going to be fun getting these redcap servos running under linuxcnc.

George

Another option is to use the Pico Systems Fanuc converter board to turn all the Fanuc proprietary signals into industry-compatible "Hall signals". Then, you have some choices. You could use other servo amps that generate their own 3-phase drive, and just feed one +/- 10 V analog signal to each axis. Or, you could use the Pico Systems universal PWM system (controller and drive) to run the motors. I don't know what size motor you are talking about, our servo amps currently handle up to 20 A.

The Fanuc converter uses a best guess scheme to convert the Fanuc commutation signals to the 3 standard commutation signals until it has passed the index mark, then it counts encoder pulses for more accurate commutation timing.

Jon

jmelson wrote:

Another option is to use the Pico Systems Fanuc converter board to turn all the Fanuc proprietary signals into industry-compatible "Hall signals". Then, you have some choices. You could use other servo amps that generate their own 3-phase drive, and just feed one +/- 10 V analog signal to each axis. Or, you could use the Pico Systems universal PWM system (controller and drive) to run the motors. I don't know what size motor you are talking about, our servo amps currently handle up to 20 A.

The Fanuc converter uses a best guess scheme to convert the Fanuc commutation signals to the 3 standard commutation signals until it has passed the index mark, then it counts encoder pulses for more accurate commutation timing.

Jon

I was aware of the pico converter but have decided to go with Andy's bldc component for the commutation conversion. We didn't like the idea of the servo rotating as much as a full turn before determining commutation. (I believe Andy's BLDC component can operate in that mode also if one desired) The same with the drives, we already have AMC BE25a20AC drives which are well matched to the servos and have a internal power supply as a plus.

So, we are pretty locked in on using mesa boards, amc drives and Andy's bldc component for the commutation conversion.

It is also woth noting that using the existing fanuc servo drives is not a option, the owner of the VMC hates anything Fanuc plus the Fanuc drives use the newer serial communications interface. Getting rid of the fanuc electronics is one of the main reasons for the linuxcnc retrofit.

What is the encoder part number?

If the drives are new enough to use the Fanuc serial interface (FSSB)
I would expect that the encoders are serial pulsecoders not the older
quadrature+commutation devices

gmouer wrote:

I was aware of the pico converter but have decided to go with Andy's bldc component for the commutation conversion. We didn't like the idea of the servo rotating as much as a full turn before determining commutation.

No, you don't understand how it works. It uses the existing Fanuc commutation info right after start-up, so it DOES provide commutation signals at all times. The servo motor is NOT rotated by the converter. When you make the first move after powering up the control, the Fanuc commutation info is decoded and used to produce the 3 commutation signals until the encoder gives the index pulse, then quadrature counts are used to generate slightly more precise commutation timing.

It is also woth noting that using the existing fanuc servo drives is not a option, the owner of the VMC hates anything Fanuc plus the Fanuc drives use the newer serial communications interface. Getting rid of the fanuc electronics is one of the main reasons for the linuxcnc retrofit.

Yes, I can mostly agree with this. The really old stuff from the brush motor days is well past its prime, and with anything newer, you have NO documentation at all. Fanuc will tell you how to wire a cable, and that is ALL you will ever get.

If the drives are serial, you'd better make sure the ENCODERS are what you think. The first generation of red cap motors had classic ABZ quadrature encoders with their proprietary 4-wire commutation scheme. The later ones all use serial encoders. I do have converter boards for these encoders as well. Everything they send is in a 77-bit serial string. Serial encoders with an "A" in the type, like (alpha)A64 also generate low-res absolute position data, which is used by my converter before the encoder has passed the index mark to generate commutation. Serial encoders with an "I" in the type, such as (alpha)I64 do not have this info, and there's no way to provide commutation data immediately at power-on unless the position is battery-backed up.

Jon

No question that they are standard quadrature encoders. Its been a while now but we did research the part number plus we have the wiring diagrams for the machine which clearly show standard quad encoder wiring.

I could be wrong about the drives themselves, makes no difference, my friend wants to use the AMC drives and they match the motors well except for the commutation lines of course.

What about the original question? What mesa board would be suitable for handling the multiple I/O lines with enough bandwidth for the commutation pulse stream speed? (minimum of 20 inputs and 15 outputs)

For that many additional I/O I would consider a 7I69 or 7I90hd running in
sserial-remote mode (48 or 72 TTL level I/O respectively)

Also if linuxcnc does the commutation you want the servo thread to
run at least 20 times as fast as the commutation waveform. This means
at least 5 KHz for a 8 pole Fanuc motor at 3000 RPM.

This means you need a decent CPU (Atoms need not apply, they trip over their shoelaces at about 2 KHz)
I would suggest a ASrock H97M Pro4 with a G3258 CPU
As I know this will run a 5i25/7I77 combo up to about 7 KHz

Thanks much Pete, just the info I was looking for. You beat me to it on the CPU issue. One of my next questions was going to be on the thread speed necessary. I do have one question though, the thread speed you mentioned for 3000rpms on a fanuc motor, would that thread speed be ok for multiple servos? There will be at least 4 servos, possibly 5 allowing for a future 4th axis rotary.

How is available stock on the boards you suggested?

Thanks again !

A few more questions to assist in mesa board selection. (assuming Pete will respond)


I looked at the 2 boards suggested, and have so far only used 5i25/7i77 boards for prior retrofits. I am unclear on interfacing the suggested boards. I see the boards can take a serial communications, I presume the board would be connected to the sserial port on the 7i77 ? The I/O boards have 50 pin headers for I/O, what additional boards would be suitable to get to some screw terminals for connection to cabinet wiring? We are leaning towards the 72 I/O board because the many unused I/O's will come in handy for other uses in the VMC retrofit BUT it appears the boards I/O is strictly TTL level compatible, what would be the recommended way to interface to 24vdc machine I/O lines? (the commutation lines would be 5V of course) Are 50 pin interconnect cables available from mesa? Lastly, all the software I seen for the boards were for Pwm/encoder/stepgen combinations, I did not see any for plain digital I/O use, would that need to be something custom?

Sorry for being a pain with this but we are getting a shopping list together to place a order.

George