new question about bias in LNA

When i design a bandgap reference for LNA's bias, the bias voltage is constant with changing of temperature, but S21, NF already change with temperature, for example, S21 reduce 3dB with the rising of temp., from -15 to 85, How can settle this problem.

Maybe you should check the reason of temperature depedence for your LNA design. For NF, it always porpotional to temperature without special effert.

Yibin.

the transconductance will decrease as temperature increasse, so is the gain. So some people design PTAT bias circuit rather than bandgap.

Absolutely agree..

Vt (thermal voltage ) which is equal to kT/q has a behaviour -2mV/T and corresponding with this gm=Ie/Vt has the same characteristic even tough Ie is constant. (in fact Ie is not constant, is increased by temperature but slopes are not equal, indeed Vt drops rapidly than Ie )

To over this phenomena a PTAT current source is used to compansate this decreasing in Vt, so gm will be practically constant.

PTAT ( Proportional To Absolute Temperature ) current source is obtained in many ways. To find a proper one, it should be examined in such VLSI or Analog Design Text books..

It's many years that, in radiotelescopes, FET based LNA, are driven with cosnstant current rather than constant voltage.
This make spar mutch more stable vs temperature changes.
To drive FET with constant Id, special power supply is used, but for the case of single supply FET LNA, you may consider to replace Source Resistor with a diode (or network) of the same chemistry (same bandGap) of the FET.

Note: for FET i means, Si FET, GaAs FET, HEMTS, PHEMTS etc.