Dasylab: Cannot get a precise reading of counter inpu

[gab_laj]

I'm trying to read the frequency of a prox- meter installed as a encoder (reading of a squared-tooth wheel)

I'm using Dasylab 11 and MC's USB-1408FS.

The problem I'm getting is that the reading obtained by the counter input module is not continuous: when I set a certain speed up, It ''hesitates'' between two value, the real speed (verefied by a stroboscope) being given by an average of these two speed...

I've tried setting a faster sampling rate, but dasylab still reads a constant speed as a hesitation between two speed...

Any ideas??

I've attached a graphic of three speed variation, notice the ''steps'' taken by dasylab..

[TomRizzo]

Are you using the counter input?

Also, what is the frequency you're are measuring (ppr)?

[gab_laj]

I am using the counter input, I'm reading 20 ppr at a speed of (maximum) 500 RPM, usually more around 200-300 RPM

[CJ_Butler__DASYLab_Pro]

The counter on the USB-1408FS is “software polled”, not read at a hardware pace. That will certainly affect the quality of the incoming signal.

Try reducing the block size to about 1/10th of the sample rate. For example, sample at 100 samples/second, block size of 10.

[TomRizzo]

Also,

the spikes you see are probably the mechanical error of the teeth in your wheel.  This is a common effect.  Basically, if you would connect the output to an ossiloscope, you would not see a perfect squarewave but a moving one changing frequency +/- few hertz.

The signal shoulod be approximately 166 Hz at 500 rpm so that you could use an analog input as well to measure the speed

[gab_laj]

I've change the sample rate and block size to 100 and 10.

When I do so for the PC-clock (by un-clicking automatic), I get readings that change from 70 to 80...

I have the impression that this particular factor is behind the fact that my readings cannot be as precise as I want...

I've tried putting 1000 Hz(sample rate) and 100 as block size, dosen't work also...

Thanks for the tips !

Perhaps you've got an idea...

[gab_laj]

I will try the analog input solution,

Yes, part of my error is definitely mechanical, because the fluke connected to the same input changes a little bit (but not as drastically and clearly separated in to frequencies per speed)....

[TomRizzo]

For a good counting module I would suggest

http://www.mccdaq.com/usb-data-acquisition/USB-4300-Series.aspx

This unit is designed for counting pulses