Measure time of 5 pulses

[LabLung]

I need to measure the speed of a motor which has an eccentric load so the instantaneous speed of the motor looks like a sine wave. I have 5 hall pulses per one revolution so I want to measure the time of 5 hall pulses and convert this to a frequency to basically filter out the speed fluctuations. The frequency range of the hall pulses to be measured is from 4Hz to 50Hz. I have two motors that must be measured this way and two counters available on a 9172 with a 9411 module. I have tried most of the shipped examples but I have had no sucess trying to filter out the speed fluctuations. I am using LabView 2009.

[srikrishnaNF]

Hi LabLung,

   As you are saying that there will be 5 hall pulses per revolution.So instead of calculating the pulse time,it is better to calculate the no. of pulses in one minute and divide the value with five and you will be getting the value in rpm.

  If you got the some other way,please post to the forum.

Thanks and regards,

srikrishnaNF

[LabLung]

I have to report the speed in Hz and at a much faster rate than 1 minute.

Ideally reporting the speed for one complete revolution at a time would solve my issues.

[johnsold]

Use a running or moving average of the pulse periods.  (I am assuming that you are able to measure the time from one Hall pulse to the next for all the pulses). Since you want the speed based on five consecutive pulses, you really do not need the average, just the sum of the times.  See the attached example in which the pulse frequencies are random but in the 4 to 50 Hz range you expect.  Notice that the speeds are smoother than the pulse times.

Lynn 

[Doiron]

Hello,

Are you simply wanting the average every 5 pulses? You can do this by using a buffered digital period measurement and dividing the total time by 5. See the attached VI (NI-DAQmx Required).

[LabLung]

I can't just average 5 pulses because the pulses the vi samples may just happen to be 2 fast ones from one revolution and 2 fast ones from the next revolution then 1 fast one from the third revolution, then it reports a false overspeed condition.

I have several inerations of averaging methods and none report correctly.

The time to complete one revolution is very consistant but the instantaneious speed is not so I need to measure the time of any 5 consecutive pulses to eliminate the speed variations.

[Doiron]

Did you take a look at the VI I attached? It will return the period of the last pulse. Therefore, you can measure 5 consectutive pulses and average those which sounds like what you are looking for.

[LabLung · Accepted solution]

This will work if it doesn't skip any pulses in the train. I ran the vi and the periods look correct for 5 consecutive pulses. 

Thanks

Brian.