Performance measurements

[Stefan Johansson]

Hello. I'm trying to verify the CMRR specifications for the NI PXI5922 Digitizer using the two channels in differential mode, but I don't reach the specification. The signal source is a PXI 5441 AVG.  The CMRR is measured by applying the same signals to both channels on the digitizer via a T-coupling and quite short coaxial cables of the same length from the AVG. And then the resulting voltage is measured to estimate the CMRR = 20*LOG(Vcm/Vout), where Vcm is the applied common mode voltage and Vout is the resulting output. The results I get for 1 kHz is about 94 dB and for 100kHz about 75dB. According to the spec the 5992 should have over 90 dB CMRR up to 300kHz.  I was using the given examples in Labview to generate the signal and making the measurement. So why doesn't our 5922 meet the spec? Is there a better way to validate the CMRR? Or might it be something wrong with the digitizer? Best RegardsStefan Johansson

SP

[sjarago]

Stefan,

According to the most recent specs I found online (http://www.ni.com/pdf/manuals/374049d.pdf), page 4 lists the spec as 50 dB up to 1 kHz.  Below that, in Figure 1, is graph of cmrr vs. frequency.  However, the accompanying text states "TYPICAL."  When a manufacturer states a specification or performance characteristic as "typical" what they really mean is that "after testing numerous 5922's that came off our assembly line, we were able to find ONE that performed exceptionally well and these are the results."

In other words, "typical" specs are not to be trusted and in reality should be ignored.  The actual spec, according to the manual, is 50 dB up to 1 kHz.  It appears your 5922 meets this spec and there is nothing wrong with it.

Hope this helps,

Steven

[Stefan Johansson]

Thank you for the answer! If the spec that you are linking to is the latest then my measurements are not that far from the "typical" curve showed in figure 1. The spec that I have is more optimistic, showing a flat curve all the way up to 200kHz. The 50 dB that you are talking about is specified when the channels are used in unbalanced differential mode. This means that one signal is applied to the "shield" of the coaxial cable, giving the 5922 two differential channels. I want to verify the CMRR when two channels are used as one "real" differential channel and the CMRR for that is in figure 1. Best RegardsStefan Johansson