How to check correct phasing?

Hello,

I have a customer that is experiencing some problems (occasional runaway at first movement) with commutated motors (using a GeoBrick), hall sensors.

 

Since I'm not there the best I could do is to make him plot the hall sensors U V W values vs the phase position.

 

All the motors should be mechanically phased.

 

the ixx91 value is $CB0000 for all the axes, that inverting the formula means that HEZ (Hall Effect Zero, transition from 3 to 1) point is expected at 61.875 degrees.

 

The motors are named Z1,Z2 and Z3; Z3 was recently changed because of a defective Hall sensor. Z2 is the one giving troubles recently.

 

Looking at the plots I can only see that the Z3 seems to have roughly a 20° error (but it is not the one having problems)...

 

Any suggestions on what else should I check?

Z1.pdf

Z2.pdf

Z3.pdf

Could you ask the customer to try using the HallsUVW.exe application available from forums's FileDepot under Turbo PMAC, Tools, HallsUVW. This application will calculate the proper value for Ixx91 for each motor independently. This should give us a good comparison point.

I've already sent the tool, as suggested also by DT Europe.

I'm waiting for a reply, but my aim is also to understand what's going on.

 

Another identical application is not running into these issues so I was wondering what could cause the runaway.

So far the possibilities are:

- bad/intermittent contact in the hall sensors feedback / hall sensor close to failure

- encoder that slips on motor shaft (but doesn't seem the case)

- other ideas?

You might want to check if the encoder is losing counts. That is the encoder count error bit, suggested M-Variable Mxx18.

 

I know that you are having issues with halls phasing for now, but for future applications and if you are not doing it, you might want to look into implementing fine phasing (described in the Geo Brick Drive manual). This gives you the best torque and efficiency out of the motor.

The problem only occours at the very first phasing, so it looks unlikely that the encoder is missing counts.

However, I will ask to make the test.

 

I appreciate the suggestion for fine phasing, I will look into it for the future.

These are the results of the HallUVW utility, repeated a few times with the same result.

 

It looks the only motor with the correct phase setting is #2, (ix91 is $CB0000 for all motors), while motor #3 (Z2) has about 5° and #5 (Z3) about 30° shift.

 

I can't believe that a 5° shift can cause a runaway while a 30° shift doesn't.... any ideas?