What is the control linearity in mixer?

Linearity in Mixer..?

hi
can any one tell me what control linearity in mixer specially the folded gilbert cell

what parameter i have to improve IIP3..?

is there another thing rather than current ?if so what current should increase
in transconductance stage or in switching pairs..?

please, i need any reference or paper about improving linearity in CMOS mixers?

thanks in advance,

Re: Linearity in Mixer..?

I thinl that to increase linearity, you need to increase input and output swings of the mixer ( besides of course increasing the device linearity ). I think that some of the important factors for determining linearity would be:

1. Overdrive of input pair; the higher the overdrive the better the linearity.

2. The output swing, the higher the better. This includes a good choice of bias points, LO level and load resistors.

3. Capacitance of the switching quad. this a non-linear capacitance that shunts the current going to the output. The larger this capacitance the worse the linearity.

I think also that in a folded mixer, yu would better increase current in the input stage and decrease it in the folded part to improve gain/linearity. Another thing to take care of is that higher gain will always mean a worse linearity so you can trade-off gain for linearity if you have this option.

Another thing that can be done is to use inductive degeneration of the input pair to improve linearit at the cost of lower gain and larger area.

I think Razavi's book "RF microelectronics" discusses linearity in mixers. Also, "Radio Frequency integrated circuits" by John Rogers is a good resource.

Re: Linearity in Mixer..?

thank you very much
i have another issue in folded Mixer the transconductance RF transistors
when has multiplier=m=1 & large W& nf=16 ---->give certain linearity
when i try to split it m=2 & W=W/2& nf=16--->Linearity go worse...Why?
thanks

Re: Linearity in Mixer..?

Hello,

Make sure that the multiplyer is working correctly, e.g but a simple transistor with bias, change m and check current.

using multiplyer instead of large w may have some effect on circuit performance, but not really that much, how much does the linearity degrade anyway ? 0.2 dBm is nothing, espicially if you are "which is most probable" using moderate or loose accuricy.

Re: Linearity in Mixer..?

AA
really every small parameter changed some specs goes up &others down

for noise measurments PSS & Pnoise
all real components& i do matching with method of port (R=50)&
another R=50 // & VCVS then ideal balun then i/p differential

NF differs
when i connect Port directly to ideal balun
& make port source type =DC
white NF ~6dB ......do u believe 6dB as white noise figure ..?

is the green word is true..?port directly to ideal balun or keep same set up as linearity measurment i.e:R=50 //to port & VCVS..?
thanks , best wishes

Re: Linearity in Mixer..?

I think in noise measurement you do not need matching resistors . But I am not quite ( the port is avery tricky element ).

6 dB is a very good Noise figure . How many sidebands do u specify for pss/pnoise analysis? Are you using a filter at the mixer's output? What is the shape of the LO signal?

Re: Linearity in Mixer..?

elbadry said:

I think in noise measurement you do not need matching resistors . But I am not quite ( the port is avery tricky element ).

Click to expand...

i 've port(has 50 ohm)

elbadry said:

6 dB is a very good Noise figure . How many sidebands do u specify for pss/pnoise analysis? Are you using a filter at the mixer's output? What is the shape of the LO signal?

Click to expand...

6 dB from CMOS mixer has 8 noisy mosfets ....!
in PSS only LO is active & i made no. of harmonics=0.
in pnoise :sweep 1kHz:3GHz logarithimic sweep
Maximum side bands=30..?

elbadry said:

Are you using a filter at the mixer's output? What is the shape of the LO signal?

Click to expand...

yes i connect 1pF between the differential o\p of the mixer(resistive loads)
LO is sin wave 500mV peak i.e 1V p-p....Do you think it's reasonable..?
thanks for your help

Re: Linearity in Mixer..?

The noise figure may be reasonable ( knowing that you filter the output ). Try removing the cap and see how the noise figure changes ( should increase significantly I think ).

The port is 50 ohms, but do you place a resistor at the input of your mixer for matching?

The sine is fine ( however, whether this amplitude is feasible or not depends on the VCO ).

Re: Linearity in Mixer..?

i removed cap, no effect on noise figure
yes i place another resistor=50 // to port and VCVS then ideal balun at the input of mixer for matching
only at noise analysis i remove only vcvs is it right or suggest another method for noise..
i want to make IIP3
fLO=2.450G
do you think choosing IF will differ b/c i use IF=50MHz for fast simulation(however,
IF =2M in the system )
i.e fRF=2.400G & 2nd tone via PAC =2.401GHz
does the IP3 changed according IF or difference between 2RF tones
or it's const for certain design

i'm now tring fRF=2.448G(i.e. IF=2M) & fRF2=2.4478G(i.e diference=200k)
from more than 35 min via PSS & PAC not finished yet ---i feel the computer is cring now in your opinion when should i restart my computer :|
thanks

Re: Linearity in Mixer..?

6 dB is very low noise figure !!!

As I understand you are using input matching, this adds roughly 3 dB to NF, then your unmatched noise is 3 dB ???

I believe something is missing here...

NF settings : Change RF port into DC.
Remove input match, keep Rs of port 50Ω
For PSS , zero harmonics.
For PNosie, ref sideband = -1, and use good number of
harmonics 10-20

For linearity, try 1dB compression, it's always faster than two tone, if you still have no convergence use qpss "Just same as pss with less aquricy.

Re: Linearity in Mixer..?

Check the link below for a very fast simulation of IP3

https://www.designers-guide.org/Analysis/intercept-point.pdf

Re: Linearity in Mixer..?

elbadry said:

Check the link below for a very fast simulation of IP3
h**p://www.designers-guide.org/Analysis/intercept-point.pdf

Click to expand...

thank you very much i will try qpss isA

here is my i\p port setup for RF & LO i used in gain ,linearity analysis

please, tell me specifically which component keep ,which delete
in Noise PSS,Pnoise analysis
note: i made NF settings what you said every time in noise analysis
" Change RF port into DC.
Remove input match, keep Rs of port 50Ω
For PSS , zero harmonics.
For PNosie, ref sideband = -1, and use good number of
harmonics 10-20 "
thank you all for your help ,
Best regards