Setting up 5i25+7i85+8i20

stes wrote:

I'm trying to set up these but I can't find xml file for 7i85

You might be able to get a start by pretending that you have a 7i77 instead. Clearly many of the 7i77 pins won't actually exist, but the encoders "belong" to the 5i25, and the 8i20 pin names will be the same too.

There is information that you might find useful in the future that can be gleaned at the command line.

Install the 5i25 and connect the 7i85. Optionally connect the 8i20.
Open a terminal window and typeYou should be able to copy and paste the output from "show pin" and "show param" into a text file for future reference.
These are the pins that can be used in the HAL file, that you can either create by hand, or have PNCCONF create for you, or (most typically) a bit of both.
The second dmesg will show you which FPGA functions are on which 5i25 pin (this isn't all that much help in telling you which terminals those are on the 7i85)

stes wrote:

Andy,

I have noticed that you use a 8i20 to drive ac servo motors also. Is it complicated to set it up?

I think the answer is "Yes", mainly because AC servos are themselves fairly complicated.

What motors do you have?

The complexity is not with the 8i20 (which simply needs to know what current to supply, and at what phase angle) but in working out where the rotor is. How you do this depends on the motor and what feedback it offers. it is simpler for spindles than for axis motors.

stes wrote:

I'm using Sanyo Denki P5 1kw. Would it be easiest to detect the phase by powering one of the phases at start up?
This motor has hall effect sensors also.

Using Hall sensors is the best way to get the motor initialised. This is the test script I use to set up a motorSave that in a file somewhere called 8i20.hal. Then edit it to suit your hardware (you won't have any "7i76" pins with a 7i85, for example)
Then start a HAL session and load the script:Now open a some halmeters to check the hall inputsand check that the halls twiddle as you manually rotate the motor.
Now you can apply a small current to the motor and see if it spins. It probably won't as there is only a 1 in 42 chance that "34" was the correct hall pattern.The motor might kick, it might buzz. It might even spin.
Now try all the hall patterns and see which one works best.Make a note of which ones spin the motor best. You will probably find two that work well, but which spin the motor in different directions. Choose one, that is your hall pattern. A further test that it is a good pattern is to try it at higher and lower currents, and make sure that it will reverse the motor properly.and so on.

Running your motors like this will work, but isn't ideal. This is trapezoidal commutation, and it is noisy and less efficient. The next stage is to switch the "bldc" component to "qh" mode, where it uses trapezoidal commutation to start, then uses the quadrature encoder for sinusoidal commutation after the first Hall transition.
So, exit halcmd, and edit the 8i20.hal file:
The lines in red need to be set to suit your hardware. Note that there is a 50:50 chance that the bldc scale will have to be negative.

Now you should find that the motor runs smoothly under speed and direction current commands, and you can integrate the bldc and 8i20 HAL into your machine HAL configuration.

One troubleshooting tip. At the halcmd: promptwill show you the 8i20 status, if it isn't working then there should be a clue as to why not in there, such as undervoltage, or enable not set, or whatever.

I forgot the link to the bldc docs:

www.linuxcnc.org/docs/html/man/man9/bldc.9.html

I guess that is a file path problem. You could try an absolute path from root. eg
/Users/yourname/linuxcnc/configs/the-file

(That's a guess, I am on a Mac right now and can't actually recall the Linux user paths. could be "Homes" for example)