[SOLVED] Fast phase changing for high power GaN transistors

[Andrew77]

Hello,

Working on a project of a high power (150W), wide band (VHF through UHF) amplifier using GaN hemt transistors I noticed failures in Gate (ie gate pad shorts toward ground/source) when used on saturated power environment in fast sweeping frequencies as a source
I did a lot of efforts to tackle this phenomenon: high current source for gate (Ig), compensation networks to increase the flatness through the whole band, etc..
I'm starting to infer that the rapid change in frequency source (100 Khz to 1 Mhz) such as sweeping frequency very similar to that used for jammers, could harm the device
I haven't found any research that prove that,
if anyone has any suggestions is welcome
tks

 

[mtwieg]

Have you looked carefully at the seep generator to make sure the frequency transitions are smooth and there are no discontinuities in the phase? Have you tried varying the sweep duration and dwell time at each frequency step?

 

[BigBoss]

Failures don't occur while the freuency is swept rapidly.There should be some amplitude peaks to create this failure.
Modern frequency sources use different switched signal sources to capture wideband and when the frequency is changed there shouldn't be any amplitude peaking while a signal source is switched to another.
There may be a signal source problem, it should be checked with a precise SA in time domain.A small peak can be caused this failure while switching period.

 

[Andrew77]

the frequency sweep is a DDS generator, looking at the spectrum analyzer I haven't seen an amplitude peak yet
in some frequencies the transistor gains more than others so it works in saturation...

 

[BigBoss]

the frequency sweep is a DDS generator, looking at the spectrum analyzer I haven't seen an amplitude peak yet
in some frequencies the transistor gains more than others so it works in saturation...

You might not see this amplitude jump on SA because it may occur very fast and it burns-out the gate ( failure ) instantenously.SA should be used as time domain mode ( zero span) to observe this fast amplitude jump.

 

[mtwieg]

You might not see this amplitude jump on SA because it may occur very fast and it burns-out the gate ( failure ) instantenously.SA should be used as time domain mode ( zero span) to observe this fast amplitude jump.

A DDS, when properly used, can give a nice frequency sweep without phase discontinuities. But even so if your circuit has any significant resonances, then a fast change in frequency can still cause a large peak in the envelope, which may damage the circuit. I would probe the gate and drain voltages during operation while looking for it. If you see peaking in the envelope then try using a higher resolution sweep.