five different measurement results using five pci-6014 cards

[daduriver]

Dear all,
I have measured the period of a pulse using five pci-6014 cards.The internal clock of 20MHz(so the resolution is 50ns) is used,and the measured results for these five cards separately are:
0.4999852 (s)
0.4999992 (s)
0.4999802 (s)
0.4999853 (s)
0.4999953 (s)
For each card the value is stable,for example, it is 0.4999852(1)(s) for the first card,which means the period of signal is stable at least at the level of
100ns,and the frequency of internal clock is also stable enough. What makes these big difference between these five measurements? Is it possible to measure the period(about 0.5s) of a pulse with uncertainty of 100ns?

[rgharrison]

Hi,

That's interesting. It's possible that the oscillator on each card has a different but small amount of jitter, however not 10us worth! I would try a couple things:

1) I would try generating a pulse from the one counter on a card and input that pulse into the other counter of the same card. You should read the exact pulse width you specified +/- 100ns. The reason it is +/- 100ns is because the pulse train generated is +/- 1 clock cycle and the pulse measurement is +/- 1 clock cycle (20MHz clock becomes 50ns + 50ns).

2) Generate the pulse and input the pulse into more than 1 card at most 3 (because of losses). Any difference in wire material or length will cause the pulse width to change slightly. Also, generating the same pulse width at differen
t periods of time might also change the width because of clock skew/jitter.

Anyway, I'm not sure if those reasons are causing the 10us difference or what but testing each one with their own counter will check their own calibration. Hope that helps.

Ron