Possible to have the IIP3 point < P-1db point?

[joakim]

When I simulate a resistive FET mixer. The IIP3 points seems to be lower than the 1 db Compression point, P-1dB is this possible?
I do the simulations in @DS. The results also seems to be hardly dependent on the maximumorder. Therefore I have increased the maximumorder to about 20.
I know the rule of tumb
IIP3 ≈ P-1dB + 10
Therefore, I don't belive it is possible to have IIP3 < P-1dB?
any advices how I can simulate this as correctly as possible?

/Joakim

 

[g579]

P -1dB used to be an output data for a device.
IIP3 is an input data for a device. If your FET mixer has gain, than the situation you described may be normal.
Try comparing OIP3 and P -1dB,
g579

 

[joakim]

The mixer has no gain, the conversion loss is about 8 dB.
Is it still possible to have IIP3 < P-1dB ?

tnx for your support

 

[alanphydenics]

Hi Joakim,

You can refer to razavi's book, RF Microelectronics. From the book, we know that A1-dB=sqrt(0.145a1/a3), AIP3=sqrt(4a1/3a3), so the AIP3=A1-dB +9.6. Here razavi assumes a3 <0, that means when increase the input power the gain of the network decrease. For your circuit, if A1-dB > AIP3, that means, a3 >0, so the 3rd order harmonic increases fast and reaches 1st order signal before 1st order gain degraded 1 dB, in fact, if a3 > 0, the 1st order gain increases with the input power increase. Normally it is not the case for mixer.

So I suggest you: 1) check the setup for simulating the 1st order gain, you can sweep your input power from a very small level, e.g. -80 dBm. You should get the right 1st order gain curve. Here remember the power gain for mixer is the conversion gain, since your output frequency is not same as input frequency, and you should also care about the input and output impedance, different impedance, different power gain.

2) Same method, when you sweep your input power (single tone), at the output should have the 3rd order signal, use mix function in ADS to figure out the power of this 3rd order signal, remember the 3rd order signal is like 2ω1-ω2. Plot out the 3rd order output power with the 1st order signal power at the output vesus input single tone power, extend these two curves, to get the right IIP3. You also can use the two-tone test method to get the IIP3, but if you do not select the power and delta frequency rightly, the two-tone method is not accurate.

In ADS there are some examples and design guide for the setup of simulation mixer. You can refer to them. I think your problem may due to the wrong simulation setup. Hope this can help you to get the right result. If possible, you can email me your design file for further discussion.

 

[joakim]

Thanks for your answer alanphydenics.
I tried to plot the conversion gain curves as well as the IMD curves 2w1-w2 in the same graph when the input power was sweeped. I used the SOI and TOI simulation from the mixer design guide, with the only change that the input power was sweeped.

Then I used RF=mix(Vload,{1,1,0}) and IMD=mix(Vload,{1,2,-1}) (I have a upconverter mixer, RF is therefore my output)
and ploted RF and IMD in the same graph.
The RF output seems okey, the slope is about 1 and the compression points seems to be the same as from the compression point simulation with the XDB controller.
Now, the third-order IMD should have a slope of 3 right? However the slope of the 3rd IMD is more like 1.5.
Maybe I should not care to much about the IP3 but it's annoying when basic theroy and simulations don't agree.
I sweeped the inpower from -20 dBm to 20 dBm, the compression point where something about 15 dBm.
Maybe the input power level is to high, to get an accurate value for IP3?

Best regards
/Joakim