Who has the experience on cryogenic LNA?

[g94426]

noise figure of cryogenic lna

I have some trouble in my cryogenic LNA design:

The three-stage LNA using NE334S01, whose freq. is 4.65~5.15GHz, seems quite good at room temperature, but fails at 15K ambient temperature. The gain drop down and the noise is higher than my estimation. And there is no oscillation occurs according to the monitor by a spectral analyzer.
I also measured some sigle NEC HEMT samples, they work well when they are cooled down and the gain is even higher than at room temperature.

Who has any experience on cryogenic LNA? Could you tell me your opinion to this phenomena?

 

[marcomdd]

cryogenic lna

Normally, at cryo temperatures the noise is lower and the gain is higher than that at ambient temperature.


How much lower is the gain ? Perhaps, the temperature decrease makes some bonding to shunt somewhere.

Are you sure no oscillation occurs ?
Are you able to monitor the gate current of each single stage ?

 

[g94426]

Normally, at cryo temperatures the noise is lower and the gain is higher than that at ambient temperature.


How much lower is the gain ? Perhaps, the temperature decrease makes some bonding to shunt somewhere.

Are you sure no oscillation occurs ?
Are you able to monitor the gate current of each single stage ?

Thank you!
I think the transistor doesn't work at proper status. Although the noise(10~14K) is lower than that at room temperature, but I estimated it should be 5~10K.
And the biggest problem is the gain. The gain is depend on the drain current. According to the result from the single HEMT tests, at 15K, the gain of a 2V/4mA biased HEMT is close to a room temperature, 2V/15mA one. But actually, the gain of the cooled three-stage LNA is at least 6dB lower than the gain at room temperature.
I've used a spectrum analyzer to watch the output of the LNA, and I'm sure no oscillation lower than 13.6GHz ------ even the input was open or short. But when I increase the drain current, oscilation occurs at 2~3GHz.
The gate current is too weak to be measured accurately. I can only monitor the drain current.

 

[flatulent]

try other temperature

Try some measurements at a higher temperature like 70 K from nitrogen. The mobility of holes and electrons in silicon have their best values at this temperature and at lower temperatures the number of ionized carriers declines. Other materials may have the same effects.

 

[nandopg]

I had the same problem as yours when I characterized some FETs at -40C. In my case what happened was a changing in the operating point due to an increase in the gate current (what is expected at low temperatures). This changing in the gate current decreased the gain of the device and increased its noise figure, exactly what is happening in your case.
To check that, connect a 50Ohms resistor in series with the gate bias voltage and measure the gate current over temperature.

NandoPG

 

[g94426]

Thanks!
BTW, are there any reports or articles about this effect?

 

[tungchang]

In my experience, when the temperature down to 77K, the gain will increase more than 3 dB. I had used NE321000 and NE329000 for years. The PHEMT transistor can be work but FET. I wish these will reduce your problems. By the way, may I ask you how to figout out the matching point in smith chart in the lower temperature?
hao.jung@msa.hinet.net

 

[flatulent]

nitrogen

The information about low temperature effects were in most of the semiconductor textbooks from 30-50 years ago. For some reason many of these useful items are left out of modern ones.


For the matching networks, you will have to run a s parameter test set with the transistor cooled to measure the values. Perhaps the manufacturers will have such information. The only problem is that the s parameter test set will not give you the optimum source impedance for minimum noise figure. You will have to do some experimenting.