The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> RF Design >> Linearity of CMOS active Gilbert Mixer
https://designers-guide.org/forum/YaBB.pl?num=1194049426

Message started by nxing on Nov 2^nd, 2007, 5:23pm

Title: Linearity of CMOS active Gilbert Mixer
Post by nxing on Nov 2^nd, 2007, 5:23pm

Hello everybody,
I am designing a CMOS active Gilbert Mixer and running into the linearity problem. The problem is that the switching pair degrade my linearity quite a lot (for iip3 roughly about 10dB). I can not understand how this happens? Is it because of the non-linearity of the on-resistance? If so, how to improve that?

Thanks

Title: Re: Linearity of CMOS active Gilbert Mixer
Post by pancho_hideboo on Nov 3^rd, 2007, 12:06am

What value of conversion gain is achieved ?

Check transistor operation point as well as local signal amplitude for gilbert quad MOS-FET.

Title: Re: Linearity of CMOS active Gilbert Mixer
Post by nxing on Nov 3^rd, 2007, 9:05am

The mixer has the gain of 20dB .my understading is that you still want to keep the switching pair in saturation for a active Gilbert cell, in CMOS. by doing this, we probably can not have a very big LO swing, is that right?

Thanks

Title: Re: Linearity of CMOS active Gilbert Mixer
Post by RFICDUDE on Nov 3^rd, 2007, 8:16pm

Yes, you want to keep everything in saturation, but a better way of thinking about the mixer is that when the switching core is fully switched it looks like a cascode amplifier. You want to make sure the cascode amplifier is in saturation and that you keep the switch transition as quick as possible. There can be substantial nonlinearity and noise during the switch transition, so keeping the transition short will minimize the effect. However, the signal swings at the output should not be large enough to cause the switching devices to go into triode (or else there will be substantial nonlinearity).

The other source of nonlinearity is the input differential pair itself. If the output swings are not causing the switches to go into triode then check to make sure the swing at the input of the differential pair is not so high as to cause the input pair to generate nonlinearity.