Hi, I have read that it is possible to accurately measure the noise figure of a device without having a calibrated noise source. This is done by including an extra amplifier and an attenuator in the process. If the amplifier gain and the attenuator loss are very accurately known then knowing the exact ENR of the noise source is not necessary. The device noise figure can be then determined as accurately as doing the normal Y-factor method with a calibrated noise source.
If this is correct then surely the next step would be to do the normal Y-factor process and change the ENR of the un-calibrated noise source in the calculation until the result given is the same as in the above procedure. So now the ENR of the un-calibrated noise source is known.
So how accurate could this result be? Could it be claimed that the un-calibrated noise source is now calibrated or only measured?
Would it be possible to create a spreadsheet to calculate all this using only the noise marker readings from something like an old 859xE or 856xE spectrum analyzer? (As an aside I don't think this would work using an E440xB spectrum analyzer would it?)
My thought is that a hobbyist could buy 3 of the small amplifier modules seen on auction sites, the best being the ones that use the SBB5089 MMIC, as it have the right gain and a fairly flat frequency response. Also needed would be a termination and 2 attenuators. On one of the amplifiers the termination would be placed on the input and one of the attenuators placed on the output. This would be the un-calibrated noise source. The other preamplifiers would be the one normally used on the front end of the spectrum analyzer for a Y-factor method measurement and the last is the DUT to be measured.