mixer noise figure measurement

i developed a passive mixer and now i want to measure the mixer noise figure. i don't have noise meter in my lab and the equipment available is a VNA, spectrum analyzer and power meter.
is it possible to measure mixer's noise figure using a power meter?

what should i do if i want to measure noise figure using spectrum analyzer. Do i need a very good spectrum analyzer?

That's possible but the result is not accurate. (I suppose you have another signal generator to give CW as LO)

Firstly, you should do the souce calibration to make the input power more precise. (Maybe you can use VNA to offer the input signal, and the signal could be another CW, but make sure the source calibration is done!) Then set the input power at a proper level, marked Pi_cold; watch the mixer output with the spectrum analyzer, and record the peak power, marked Po_cold; enhance the input power to a relatively higher level, marked Pi_hot; and record the corresponding output power Po_hot. The ratio of Po-hot to Po_cold (both are ture value, not "dB") is Y factor. And ENR is (Pi_hot-Pi_cold)/Pi_cold (ture value again). With ENR and Y, you can calculate NF.

If you have a noise source, things will be easier. But the ENR of the noise source should be high enough and a pre-filter is necessary. The loss of the prefilter could be measured precisely and removed directly from the test result.

You can measure it accuratelyl, but you should have a noise figure option with the spectrum analyzer. For example, you should have 219 option for Agilent spectrum analyzer.

thanks for the help
:D

A Noise Source is required.

If you have no Noise Source, some method based on "room temperature" loads and KTBGav knowing may be used, but they are so inaccurate....

My suggestion is to give up the idea to measure directly noise figure.
You may have an idea of the Noise Figure, indirecly, by spar measurement.
For a passive circuit, like mixer, NF>Gav but not much more. Add arbitrarially 1dB.

where Gav is Available gain.

Since Gav may be not easy to measure , you may assume that Gav≈Conversion Gain.

The overall uncertainty due to the aproximation is smaller than uncertainty obtained by NF measurement without a Noise Generator

Simply, for a passive mixer ,the NF of a mixer is equal to the conversion loss of the mixer if and only if you use a image reject filter before the mixer, otherwise you should add 3db to the conversion loss to take account the effect of the image noise in NF.