Issues with measuring return loss

Hi All,
I am trying to measure the return loss of a number of passive loads using a directional coupler + signal generator(SG) + spectrum analyzer(SA) with the setup shown.
I basically generate a CW at 500MHz with the SG and feed to the coupler and
a) terminate the out-cpl port and measure power from the in-cpl
b) terminate the in-cpl port and measure power from the out-cpl
subtract the two measurements to measure the return loss.

The parts/instruments used are:

1. SG - N5181A
2. SA - N9010A EXA
3. Coupler - Minicircuits ZFBDC20-62HP+S+
4. Minicircuits 50Ω terminator

I have noticed that my results are different from the VNA measurements:

I am using a small RBW and 10dB attenuation on the SA.
What is causing this error in my measurements? Is there a way to improve the accuracy of my measurements?

Thank you

I would like to say your method is quite good and the results are good as well. It is actually the principle of network analyzer. If there is phase detector added, then it becomes a VNA. The only thing you forgot is the loss of the coupler itself. The actually power injected to the DUT is lower since there is a small loss from coupler IN port to OUT port. So the actual returnloss should be slightly lower, just like the comparison with VNA.

VNA is a calibrated equipment but yours is not.. However the results are not so discrete..

Thanks for your comments cariban.
But I'd like to improve this to have a better match with the VNA, is there a way to achieve phase detection by setting up the SA to trigger with the SG?



BigBoss,
I understand, but the question is what is causing the difference and how can I improve it?
Thanks

There should be some phase-detector chips available. For the amplitude, you can just do similar "calibration" as VNA. Fit an "OPEN" or "SHORT" kit for VNA to the OUT port. The reflection should be very close to 1. Then measure the power difference between the two coupled port. That is the offset you should use to compensate the measurement.

1- )Are you adding/substracting any connectors while you are measuring? Like a male/male or female/male connector.

2- What is the "through" loss which you are using?

3- Also, the 10db attenuation you are using on SA may not be equal at all conditions.

Truth is, the difference you are trying to measure and the results are so close that, it is not that important any more :)

Difference between your setup and VNA is at max 1db and at min 0.2db.