In article ,
"Pete KE9OA" wrote:

This link from Mini-Circuits explains it very clearly, with spectrum
analyzer plots, etc:
http://www.minicircuits.com/pages/pdfs/mxr1-18.pdf
I am not sure about this calculation that is mention here but as an example,
let's consider 0dBm for each tone into the RF port of the mixer under test.
If our IMR (delta) is 60dB, our IP3 will be +30dBm,
since IP3 = [(IMR/2) + Pin]

If we had -4dBm for each tone, using the same mixer as the first example,
the IMR would be 68dB.

If we had -10dBm for each tone, using the same mixer as the first example,
the IMR would be 80dB.

Once again, IP3 is calculated in this manner: IP3 = [(IMR/2) + Pin]

This is the method that Mini-Circuits, Synergy Microwave, Watkins-Johnson,
and other vendors in the industry use when making this calculation. The only
difference is that the term "Delta" us used for the IMR spec. I hadn't heard
of the "IMR" term until I did that stint at Motorola last year.
Take a look at the PDF link, and it will become very clear.
One characteristic of the IP3 terms is that as you increase the level of the
two tones at the input port of the mixer, the 3rd order distortion products
will increase by a 3:1 ratio over the desired tones.
As an example, if you increase your two RF tones by 1dB, your 3rd order
products will increase by 3dB.
3dB increase for each tone will cause a 9dB increase in the 3rd order
products.
This example is only valid if you are operating within the linear range of
the mixer. The linear range is defined as the range where conversion loss
(or conversion gain) is constant as you increase the signal level at the RF
port.
Consider that your typical Level 7 mixer has a conversion loss of 6.5dB.
There will be a point where you will increase your input tones and the
conversion loss will be 7.5dB. This is your 1dB compression point.
Now, the IP2 calculations can get confusing, since there are different
methods of measuring it. The [RFin + (I.F./2)] method is commonly used. With
LO power applied to the mixer under test, a desired frequency is applied to
the RF port.
A measurement of outpur power at the I.F. port is then recorded.
Next, a frequency of [RFin + (I.F./2)] is applied to the RF port. The power
at the I.F. port will now be between 50dB to 80dB below the initial recorded
value. This is your IMR (or Delta).
IP2 is calculated as (IMR + Pin). If your IMR is 70dB and your Pin is 0dBm,
your IP2 is 70dB.
If your IMR is 70dB and
your Pin is -10dBm, your IP3 is 60dB, etc,etc,etc.
I hope this provides further clarification!

Pete

Input tone power (both) -4.0
Output tone power (both) -3.8
(2*F1) - F2 power was -59.7
then the IP3 would be +24

The formula I quoted IP(n)= Pin + (delta P/n-1) is a classical
derivation of a 2 tone result in the passband of a broadband circuit
such as an amplifier. Delta P is the difference in the output tone level
and the intermodulation product level. You can use it for the input IP2,
IP3, et etc. The formula can be used to calculate any intermodulation
product as long as the following conditions are met:

1. The tones and intermodulation products you want to make a measurement
on all have to be in the circuits passband.

2. You have to be in the circuits linear range.

3. You have to be within the dynamic range of the measurement equipment.

The Mini-circuits pdf is about a making these measurements on a mixer
and so it requires a third generator as the Lo.

IMR is intermodulation ratio. The definition appears to be the delta of
the input tone level power and the measured spurious response, which is
the intermodulation product I speak of or in other words is a difference
dBc (dB below carrier, this being the input tone). This being the case
then IP3 = Pin + (IMR/2) has the same meaning if there is no gain. If
there is gain then you would get a different answer. I think you are
better off using the formula I referenced as both input power and gain
or loss are accounted for.

The test setup has several amplifiers so I don't know how they actually
expect to make a measurement on the DUT. Also troubling to me that they
state the IP3 measurement can only be made at some input power level and
that it you will get a different result at a different input power
level. Well, you will get the same result at different power levels as
long as you account for it and conditions #2, and #3 above so I don't
understand their problem with that.

They are also using filters. Using filters is OK as long as you don't
violate condition #1 above.

--
Telamon
Ventura, California