intermodulation test

When I measurement a mixer's IIP3,the IM3 should be increase 3dB as fundmental increases 1dB.
If the fundmental singal is large, the IM3 increases less than 3dB because of the gain compression.
but if the fundmental is very small,and i use the spectrum analyzer to measure the IM3,the IM3 increases less than 3dB too.
can someone tell me why?

If the signal is small then the IM3 will be very small. Are you sure you are getting an accurate measurement of the IM3 for a small input.

When using a SA to measure distortion, you must rmove the effects of the SA distortion and noise floor from your measurment. The SA will typically have its own IP3 of 40dBm and a NF of 25dB. Also, the two input tones will have some distortion. You will need to characterize this as well. Typically one must use pads with the power combiner to monimize in put intermods. Also you may have to put pads between the DUT and the SA to insure you are looking at only the IM3 of youe DUT.

C

Very important in IP3 measurement is the channel spacing between input tones and also important is the Bandwidth of the system.
In general the IM3 components (2f1-f2 and 2f2-f1) should be 10dB higher than the Noise Floor of the spectrum analyzer.

"{Typically one must use pads with the power combiner to monimize in put intermods. Also you may have to put pads between the DUT and the SA to insure you are looking at only the IM3 of youe DUT"

Could you please explain more ?
Thanks!

wilson

Attached is a Mini-Circuits application note on intermod measurements. Take a look at Figure 1. It describes the importance of using pads to reduce the intermod of the signal generators from getting through. Note also, that you cannot over[pad either:)

Also, when doing the testing, it is good practice to use a variable attenuator on the output of the sources or output of the combiner and vary that (a passive component) rather than varying the signal generator outputs. If you choose to vary the signal generator output, then the source non-linearities will vary with your measurement. If you just change the passive attenuator values, it does not alter the input signal non-linearities.

Hope this helps,

Joe

To Wilson_yu_chen

To measure IM3 you must put two tones into youe device (DUT). One way to do this is to combine two signal generators with a wilkerson power divider. This divider has isolation between the ports of 15 - 30 dB. This means each sig generator is injecting a signal into the front of the other Sig Gen and this will create IM. Also, by adding pads to the output of the DUT, you reduce the IM of the SA. Most SA have built in front end attenuation. If DUT IP3 does not change with varying front end attenuation then the SA IM is not disturbing the measurement.

I used agilent E4438C,which can generate two tone with little distortion .When I test the IM3,the amplitude of it was 15dB larger then the floor noise. So I can ignore the noise. I think it is maybe the SA uncertainty or the dynamic range of the SA.