oscillation in a Power Amplifier
We are testing our 1.5 GHz Power amplifier. In our first attempt, we have faced an oscillation. Our transistor is CGHV14240 CREE (wolfspeed). We utilized a resistor in series with DC feed of input and also a resonator at 1.5 GHz in parallel with a resistor. If we measure S parameters of PA, in a range of 1-1.5 GHz we will have S11 higher than 0 dB. The original source of oscillation was determined with the aid of spectrum analyzed which is from 90 to 150 MHz. Thermal instability was inspected using infrared thermal camera. no oscillation due to thermal instability.
Any comments?
If S11 is higher than 0dB, the oscillation declares itself.
Follow the stabilization techniques
Your input is not the only "port" through which signals
can be injected to close an oscillation loop. Ground
is a nice "back door". Inductance from PA device to
ground can separate the "chip ground" from the
"ground plane" and push signal back through the input
port-pair.
Have you attempted to quantify the assembly and
test setup parasitics and push them back onto a version
of your design testbenches, and see what you should be
expecting (not what you thought at design review, from
the inside looking out - but here and now with hardware
in hand to inspect, to "close the loop" on design and
analysis). The process of so doing, may show you things
that want improved on the bench setup or the test article.
First I have no info on that transistor described, so I do not know power, etc.
In any case first try with a piece of ferrite on the input trace, to reduce the Q of circuit, usually it helps. In this case use some resistors to reduce Q and stabilize circuit, otherwise make wider the input bandwidth or retroaction between collector and base.
The oscillation is 90-150MHz on a 1,500MHz design? That's seems pretty low- large parasitic or coming in the "backdoor" as dick_freebird says.
You cannot easily predict that.The amplifier may work @ 1MHz, the oscillation may occur @10GHz.Totally depends on "intrinsic" and "extrinsic" components of the active device(s).
The essential is to find that the oscillation is output related or input related..
Bias network decoupling (and, to what?) is another
point to consider. There would be some filter on the
gate DC network (?) and maybe it is a place for a ground
loop / excess inductance problem to manifest.
last time I played with a cree...it needed a small series resistor in one of the bias lines...
Yep, I did not find the 14240 data sheet, but the 14250 data sheet...clearly shows a 5.1 ohm gate bias network resistor. It is there for a GOOD REASON!
