Measure differential OIP3

I have a differential output microwave mixer. I am trying to measure the OIP3. But, I am getting different OIP3 calculations for each way I measure it. Does anyone have an example or paper specifically dealing with differential oip3? For instance, if you have a passive balun at the output, and measure the balun's S21 and S31, how do you de-embed the balun loss. After the balun, I can measure one of the tone output powers P0, and measure a 3rd order product X db down. OIP3 = P0 + (X/2) + some correction for the balun loss.

Lets say that at this one frequency the balun loss is measured as 7 dB. In the above calculation, is the correction for the balun loss 7 dB? Is it 7 -3 dB = 4 dB (since when I measure the balun as a 3 port single ended device, it does not take into account the vector addition of the tones that will go on when I connect the two input ports to the two differential output ports of the mixer)? Should it be 7 dB - 6 dB, since the two differential outputs are correlated?

My brain is hurting on this one.

Hi dear Biff44
I just thought perhaps these links can be interesting and helpful for you . see below , please :
http://www.springerimages.com/Images...-008-9272-8-19
http://144.206.159.178/ft/745/9302/180056.pdf
Best Wishes and Good luck
Goldsmith

Hmmm, does the article show the actual calculation?

Apparently !

Springerlink free trial is not working, and S395 for a subscription...I'll keep looking

But you can freely read those pages . just click on the pages that you can see at below of that page . you will see containing literatures .
Best Regards
Goldsmith

Spent some time with a N5242A yesterday, and with the balun saved as a mixed mode S2P file, the network analyzer de-embeds the OIP3 calculation to include the full balun S21 loss at the output. In other words, if the balun has S21ds= -7 dB, it adds 7 dB to the measured P0 of one of the two main tones in computing OIP3. Just FYI.

so it ads 7dB to main tone , and 7dB to IMD3 power, so did OIP3 7dB increase?

There is no mention of adding 7dB to IMD3. Just adding 7dB to the output power Po, to compensate the balun losses.

On the other hand I think 7dB is too much loss for a balun, even it is a high-frequency one. Sometimes is good to use a differential matching network to minimize this loss.

Marki and Picosecond pulse labs baluns run 4 to 9 db of loss, depending on frequenty.

OIP3 = P0 + (X/2) where x is the distance in dbc from one of the peaks to the adjacent 3rd order product top. So adding 7 dB to P0 will also add 7 dB to OIP3 cacluation, and the N5242A OIP3 display jumps up the same amount when the balun is de-embedded.

I am just surprised it is not 7 - 3 = 4 dB.

i agree, surprising part is that oip3 increased 7 dB, rather than 4 dB due to diff to SE conversion.