A resistive has a POSITIVE conversion gain--- Help me

时间：04-08 整理：3721RD 点击：

I am designing a passtive. However, it shows a POSITIVE CG.
Is there anything wrong in my circuit or in my stimuliation?
Help me, everyone~~ Thank you very much!

The circuit is active mixer so it can have possitive CG.

Passive = NO active devices and as kspalla said yours does have active
devices

what made you think it was a passive mixer ?

cheers
Dave

Thanks for your replay~

I have different idea with you.

I think passive mixer does not mean NO active device. The reason why this circuit is called passive mixer is that the maximum CG is 2/pi, equals -3.92dB. In other words, the circuit can never has positive CG.

In the circuit, no DC current flows into the MOSFET. And the MOSFET operates in triode region when LO is high. The MOSFET acts as a resistor.

Detail analysis can be seen in many books and papers. And they all consider this kind of circuit as passive circuit.

Here are two papers. You can download from IEEE.
1 Ghulam Mehdi: Dual Band Offset Mixer for On-chip RF Test in 0.13 um CMOS
2 Krenar Komoni, Sameer Sonkusale, Geoff Dawe: Fundamental Performance Limits and Scaling of a CMOS Passive Double-Balanced Mixer

Added after 1 minutes:

Passive = NO active devices and as kspalla said yours does have active
devices

what made you think it was a passive mixer ?

cheers
Dave

How did you say it has a positive gain? How did you measure the conversion gain?

How did you say it has a positive gain? How did you measure the conversion gain?
Stimulation steps follow the file” LAB2_Mixer_Tutorial” which is attached.
Simply I use Canence spectreRF PSS and PAC stimulator.

Detail results are attached~ Hi, steven,

If you refer page 5 in Lab2_mixer_tutorial, you can find there is a 50 Ohm resistor paralleling to the RF port and LO port. This makes the output voltage of the port is just the volage that corresponding to the power you set and 50 Ohm load. If you miss the 50 Ohm resistor, the output voltage of the port should be higher than what it should be. This may be the cause of your positive conversion gain. Then from your figure, the conversion gain should be -1.4dB (4.6-6).

Another way that you can get the correct convesion gain is to measure the voltage ratio of the output to input instead of just measure the output voltage. In this way, you needn't care the input signal level.

regards,
rfcn
Hi, rfcn. I parallel 50 Ohm resistor to RF and LO port.
Now Conversion gain is about -1.4dB.

But can this kind of mixer reach a conversion gain above 2/pi, or -3.9dB?
Many books say the maximum a conversion gain is 2/pi.

So problem still exists.

Another question:
You said: Another way that I can get the correct convesion gain is to measure the voltage ratio of the output to input instead of just measure the output voltage. In this way, you needn't care the input signal level.

How can I operate? Use AC analysis?
Please teach me
Thanks a lot !

In the plot form, select Voltage gain.
I upload a picture containing how I plot CG curve. ( see picture “Direct Plot Form” and “CG from Voltage Gain”)

In the picture “Direct Plot Form”:
Analysis: I choose pac
Function: Voltage Gain
Select: +- Output and +-Input Nets
Sweep: sideband
Modifer: dB20
Output Sideband: -1 150M---250M
Input Sideband: 0 550M---650M

Then I click MIXO_P MIXO_N RF_P RF_N

Finally I get the curve (see picture “CG from Voltage Gain”).
And the gain is abput -3.55dB.

Am I right?

And I use another way（See the pictures named “RF-port” “dft of output” and “dft of input”）

In the RF port : (See the picture named “RF-port”)

Source: I choose sine
PAC Magnitude: remove 1V

Then I use transient analysis:
Stop time: 1u
Accuracy Default: moderate

After run and stimulation:
I plot differential output signal (click MIXO_P and MIXO_N)

I plot differential input signal (click RF_P and RF_N)

I use dft analysis for both of the input and output differential signal.
Time is setting : from 500ns to 750ns
Number of Samples: 1024

You can see the file named “dft of output” and “dft of input”

In the dft of input, the mag of the input differential signal is 12.64mV @ 600MHz

In the dft of output, the mag of the output differential signal is 8.483mV @200MHz, which is the IF signal; another signal is 6.79mV @1400MHz, which will be filtered.

Sothe gain is 20 log(8.483mV /12.64mV)=-3.46dB. This result seems correspond to the picture “CG from Voltage Gain”.

However,the CG is still above -3.92dB which is the maximum CG in theory.
Why

And another question:

Assume the CG= -3.55dB is in reason, but why noise figure is different much from CG?

According to the paper “Highly Linear Mixer for On-Chip RF Test in 130nm CMOS” on page 30, NF should be equals CG.

But here I my design, it is not.
Why?

I upload the NF curve and paper”Highly Linear Mixer for On-Chip RF Test in 130nm CMOS