2.7 GHz vs 6 GHz ... EVM Differences

Question

Dear NI RF HW Guys,
I have a 2.7 GHz RFSA as well as a 6 GHz RFSA ....
I see a huge difference in my EVM readings with the
same Labview VI and same DUT transmitting a predefined packet.
I am decoding my frames in 2.4 GHz band.
I get favourable results in the  2.7 GHz RFSA.
Can anyone explain why is this huge difference in EVM,
when I have not changed my Labview VI and the DUT transmit conditions
Regards,
Dharmendra
Solved!
Go to Solution.

Answer 1

"Shouldnt EVM be same irrespective of HW differences ?"
 Not
necessarily. Phase noise and amplitude accuracy have bearing on EVM and
although I am not saying at this point that these are the explanations,
HW difference makes a great impact on EVM performance in general. The
digitizer difference between 5660 and 5661 most likely will not make a
large difference in performance, and indeed you will notice the EVM
specs for 5660 and 5661 are the same.
 "Andy, Just in case you were worried about signal saturation / clipping ...."
I
am indeed worried about saturation. You're generating a 0 dBm signal
into an analyzer whose max input level has been set at -10 dBm. How do
things change when you use the proper reference level of 0 dBm? You
say you are capturing a clean RF signal and I am curious as to how you
determine this? Visually by looking at the spectrum? Signal distortion
leads to amplitude errors and amplitude errors are one component of
what EVM measures. Do you see a difference in EVM between reference
level of 0 and -10 dBm?
Our EVM specs for QAM are
listed in the 5663 specs and can be compared to 5660/5661 EVM specs.
This data clearly shows the EVM performance of the 6.6 GHZ platform is
better than the 2.7 GHz platform.
One difference 
between the 2.7 GHz platfrom and the 6.6 GHz platform is the amount of
'headroom' between the stated reference level of the PXIe-5663 and the
clipping level of the instrument. A -10 dBm reference level places the
clipping/overload level @ 0 dBm on the 2.7 GHz platform. The same
reference level would place the clipping level @ -4 dBm I believe (I
could be off by 1 or 2 dB but not much more) on the 6 GHz platform. In
any case, there is less headroom on the 6 GHz platform since the 6 GHz
platform has 1 dB attenuation resolution (as opposed to 10 dB
resolution on 2.7 GHz platform) and doesn't require as large an amount
of headroom. This could very well be  your issue.
As a
test you should run generator and analyzer examples for the Modulation
Toolkit, QPSK, and see what happens. Also, perhaps we can provide more
assistance if you provide your code and the ability toreproduce the
behavior you are seeing. If we can reproduce it we can figure out why
you are seeing differences. At the very least I recommend using a
correct value for Reference Level.
Regards,
Andy HindeNational Instruments
Message Edited by Andy Hinde on 06-11-2009 10:13 AM
Message Edited by Andy Hinde on 06-11-2009 10:14 AM