1-tone vs. 2-tone IP3 measurement for mixer

Normally IP3 is measured using two tones, by measuring spur level 2f1-f2, 2f2-f1 at mixer's input side.
For example, in a down-conversion case, the 3rd order spur will be: LO-(2f1-f2), or LO-(2f2-f1) at output side.

Can IP3 for mixer be also measured using 1-tone, by measuring 3*f_in at mixer's input side??
For example, in a down-conversion case, the 3rd order spur will be: LO-3*f_in at mixer's output side.
This is actually the mixing of 3rd order harmonic of input tone. Maybe it really should be named IP3H? The letter H denotes "harmonic".

Is 1-tone measurement also valid approach for IP3? What is the difference in value between these two different measurements?

Thank you for the help!

The basic reason of the two tone measurements, both for linear (amplifiers) and nonlinear (mixer) devices is that all the signals f1, f2, 2f1-f2, 2f2-f1 are IN BAND of your circuit while using the H3 it will be probably out of band, so it will be attenuated by frequency response of the circuit.

If in your case 1-tone signals at input and output of the mixer are inside respective bands (both fundamental - of course, and H3), the measurement won't be affected, but I do not expect this will be true. It depends on LO & RF bands.

I hope it can help.

Mazz

Thank you Mazz.

Unfortunately, for my mixer case, both fin and 3*fin are in-band at both input side and output side.
input sidene tone signal fin, and its 3rd harmonic, 3*fin
output side: mixing of these two inputs with LO are also fall into output band. (LO-fin & LO-3*fin are both in-band at output)

My question are whether using single tone and two-tone measurements make any difference? Is the one tone measurement valid in my case?

Can you tell me your RF, LO and IF bands (lower to upper limits)?

We use a nonlinear device to perform the mixing operation, but a mixer actually it is a linear device, which is shifting a signal from one frequency to another, keeping the properties of the initial signal, like phase and amplitude, and doing in this way a linear operation.

In a mixer cannot use a single-tone (RF to be one tone, and LO the second tone) to measure the equivalent of a two-tone intermod test.

Meantime, some of the active mixers follow the same 10dB rule of some LNA's. (IP3 is about 10dB higher than P1dB), so you can use a single-tone to find the P1dB, and from there the IP3.

You may have a more fundamental problem than IP3 measurement. If the 3rd harmonic of the input frequency and the 3rd harmonic of the output frequency are both in-band, you'll have a significant problem with spurious signals getting into the signal chain. You should seriously consider adding filtering at both the input and output to restrict the band of frequencies there. Generally, a bandpass filter at both the input and output are used.

Dave

Mazz said:

Can you tell me your RF, LO and IF bands (lower to upper limits)?

Click to expand...

Mazz, how about make up something: RF input 0.2 to 2GHz, LO=10GHz, output IF: 8 to 9.8GHz.
So for example, 3rd harmonic of input 0.2GHz, will be 0.6GHz. So it is in-band.

---------- Post added at 09:38 ---------- Previous post was at 09:35 ----------

vfone said:

We use a nonlinear device to perform the mixing operation, but a mixer actually it is a linear device, which is shifting a signal from one frequency to another, keeping the properties of the initial signal, like phase and amplitude, and doing in this way a linear operation.

In a mixer cannot use a single-tone (RF to be one tone, and LO the second tone) to measure the equivalent of a two-tone intermod test.

Meantime, some of the active mixers follow the same 10dB rule of some LNA's. (IP3 is about 10dB higher than P1dB), so you can use a single-tone to find the P1dB, and from there the IP3.

Click to expand...

vfone, have you heard of IP2H, IP3H measurement (H means harmonic)? They use single tone RF for measurement. You can find such info from Agilent on the web (I will see if I dig out these).

My question was: what is the relationship between IP3H and regular two-tone IP3?

---------- Post added at 09:41 ---------- Previous post was at 09:38 ----------

RFDave said:

You may have a more fundamental problem than IP3 measurement. If the 3rd harmonic of the input frequency and the 3rd harmonic of the output frequency are both in-band, you'll have a significant problem with spurious signals getting into the signal chain. You should seriously consider adding filtering at both the input and output to restrict the band of frequencies there. Generally, a bandpass filter at both the input and output are used.

Dave

Click to expand...

Hi Dave, assuming we have harmonics in band and we are not able to filter it for now. I am interested in finding out how bad it is. If the distortion due to harmonic low enough, I can still live with it. That is the motivation for measuring it ,so we will know how low or how high it is.

Some mixer like Hittite give a spurs levels in mixer's datasheet, you can find the level in the datasheet.

So effectively you're building a block up-converter, and the harmonic content of the input signal isn't really relevant to the analysis. Whether or not you have a single tone with a 3rd/5th/7th harmonic, or multiple signals you're up-converting, you don't want the output of the mixer to be full of mutual interfering tones. Are you looking to specify a mixer, or do you have a mixer that you're trying to evaluate?

Either way, the original question you asked is somewhat malformed. You have multiple signals input to the mixer, and need to determine what the output looks like. You'll need IP3 in band, with 2 input tones as well as the local oscillator. In addition, you would be well advised to look at IP3 across the input frequency band, as well as with varying tone separation, as the relative phase of the tones will impact the IP3 output levels.

As far as IP3H, there was a somewhat decent treatment in Egans RF System Design book. However, when doing stuff like this, I have to question what advantage you're trying to gain by using a measurement that isn't obviously related to what you're trying to do. If for no other reason than when you're in a design review, if you have to spend 20 minutes justifying why you measured what you measured, the whole thing isn't going to end well.

Dave

Dave, Thanks for your reply. This is a common problem encountered in most wideband systems. need to check different views on this.

The main idea is that in any nonlinear device (amplifier or mixer) you cannot get a true IP3 number, just looking to the harmonic levels when a single tone is injected at the input (and not to IM products as two-tone).
I've seen papers try to prove that it's possible, but experiments I did during years show me that the results are not consistent.

I've measured different IP3 values depending on the separation between the two tones several times. The IP3 is pretty sensitive to the return loss at the 3rd harmonic, so if you've got a somewhat uncontrolled impedance at the output of the nonlinear device, it's not surprising the results are not consistent.