Overheating of CGH40010F transistor in power amplifier testing

[Mabrok]

Overheating just to consider during power measurements. But for S-parameter measurements no need to mointor the transistor temperature, Am I right?

 

[crutschow]

I have attached the transistor on heatsink directly without any interface (like indium foil or thermal paste)

You need to use a thin layer of thermal paste to get a good thermal contact between the transistor case and heat-sink, otherwise the small air gaps between the two will significantly increase the thermal resistance (and the transistor temperature).

During testing I suggest you attach some type of thermal sensor directly to the transistor so you can monitor its operating temperature.

 

[albbg]

Overheating just to consider during power measurements. But for S-parameter measurements no need to mointor the transistor temperature, Am I right?

Yes, but when you want to measure the S-parameter be very careful to connect the network analyzer to the output of the amplifier: you risk to damage the instrument. Even if you input power is low, unless you are very very sure the amplifier is stable, an instability can generate an unwanted full power (10W or more) signal. I think you could use an attenuator (roughly 40 dB 20W) from the output of the amplifier to the port 2 of the NA, that you have to insert during the SOLT calibration. In this way you will loose the information about S22 (it will appears as perfectly matched) but the other three params will be valid.

 

[Mabrok]

Yes, but when you want to measure the S-parameter be very careful to connect the network analyzer to the output of the amplifier: you risk to damage the instrument. Even if you input power is low, unless you are very very sure the amplifier is stable, an instability can generate an unwanted full power (10W or more) signal. I think you could use an attenuator (roughly 40 dB 20W) from the output of the amplifier to the port 2 of the NA, that you have to insert during the SOLT calibration. In this way you will loose the information about S22 (it will appears as perfectly matched) but the other three params will be valid.

I have simualted the stability using K-delta test and it was stable during the simulation. Even the out of band was stable. But i am not sure if the simulation results accurate or no. I simulated using ADS. For the attenuator, I have only 20 dB one. As i will loose information about S22, then how i can get valid S22 for comparsion with simulated one?

 

[albbg]

Don't trust too much in the simulations. I will start with comparing S11, S12 and possibily S12 with the simulation leaving S22 for further investigations if needed.

 

[Mabrok]

Don't trust too much in the simulations. I will start with comparing S11, S12 and possibily S12 with the simulation leaving S22 for further investigations if needed.

Furthermore, if accurate S22 is missed from the measurement data due to attenuator. We can not calculate the stability factor as it is function of all S22 and other S-parameters. Thus, we can not know if amplifier stable or no. Any suggestions regarding this issue?

 

[albbg]

I'd test before the amplifier with generator and spectrum analyzer for different loads other than 50 ohm to be sure about the stability. If always
is OK then measure only the S22 parameter with the input of the amplifier closed on 50 ohm. At this point in the S-parameter taken with the output attenuator just substitute the original S22 with that measured in this way.

 

[Mabrok]

I'd test before the amplifier with generator and spectrum analyzer for different loads other than 50 ohm to be sure about the stability. If always
is OK then measure only the S22 parameter with the input of the amplifier closed on 50 ohm. At this point in the S-parameter taken with the output attenuator just substitute the original S22 with that measured in this way.

In the case of amplifier unstable. Are there any risk on PA or VNA during measurements(s-parameter meas)?

 

[albbg]

Yes, there is the risk that an unwanted RF signal at the output of the amplifier can reach the port of the VNA and damage it.

 

[Mabrok]

Yes, there is the risk that an unwanted RF signal at the output of the amplifier can reach the port of the VNA and damage it.

So, Is 20 dB attenuator sufficient to protect VNA?

 

[albbg]

It dispensa form you vna.

 

[Mabrok]

It dispensa form you vna.

I did not understand what you mean?
VNA input max is 26 dBm, my PA expected to give 40 dBm, with attenuator of 20 dB. Means 20 dB will be out from attenuator to VNA port 2. Which means VNA protected. This theoritically, i do not know practically as do not have industry experience.

 

[albbg]

Sorry I wanted to write "It depends from your VNA" but my cellphone automatically corrected (wrongly) my sentence.
However, yes your calculation is correct.

 

[dick_freebird]

Your calculation is as good as the embedded assumptions.
Such as the max Pout. Given that you've seen uncontrolled
oscillation (?) I say all bets are off; the circuit is not what
you think it is, or else it would not oscillate and burn up
devices.

You might like to begin with lower drive power and more
inline output attenuators until you have confirmed that
the VNA is indeed safe.