Is it possible to calibrate the 2 counters on a PCI 6221 ?

[raphael-univ-metz]

Hello,

I make a measurement of a 5MHz frequency with a PCI 6221. I use also the large range measurement and the high frequency measurement. The large range measurement have better results.

But, there is a shift between the measured frequency and the frequency generated on the entry of the 2 counters. It is not an offset but rather of a line according to the generated frequency. Is it possible to carry out a calibration of the counters or of the methods of measurement ? Perhaps on the PCI 6221 it-self ?

Thank you by advance.

Raphael

[jasonv]

How much of a drift are you noticing? It could be possible that the source the 5 Mhz frequency is being generated from is less precise, or maybe the frequency you are reading using the 6221 is within the specified tolerance for the device. Were you able to verify using another device that the 5 Mhz source was in fact precise?

Checking the specifications for the 6221, I noticed that the counter base clock is rated to 50 ppm.

A few other things to try would be refining the minimum and maximum values for frequency specified for the channel to be in a more narrow range, or possibly providing an external clock that is more precise than the one on the 6221 board.

[raphael-univ-metz]

Hello Jason,

The drift I notice is approximately +30Hz. I have already check the 5MHz source with an other device and it is ok, there is no drift. When I refine the minimum and the maximum value of the frequency specified, it does not change the drift. I think that as you told me, the drift is caused by the 50 ppm rate of the timebase frequency. I also post an other message in order to know how the measured frequency is calculate with the large range measurement method. Thus, I could perhaps find the exact value of the time base clock. If you know how this measurement is taken, that would be great to you to send me a message.

Thanks for your help.

Raphael

[Kevin_Price]

Raphael,

1. Your measurement error of +30 Hz in a 5 MHz measurement is within the device's specs. It seems to indicate that your own specific card has a 6 ppm error in its timebase at your operating conditions.

2. As far as I know, the calculation that is done uses the assumed nominal value of 80.00000000 MHz when it converts from tick counts into time or frequency. I base this on observations I made while trying to generate various frequencies and querying for the actual high and low time properties. The values returned were always an integer multiple of 1/80 MHz.
Unfortunately if I'm right, there's no information built into the board that will give you the exact actual timebase. You'd need to perform a calibration against an external standard.

-Kevin P.