TX amp temperature stability

[mtwieg]

I added the current regulation circuitry, which seems to regulate the current just fine, but as I feared my gain is still drifting by almost -2dB, even with all the bias conditions fixed. So the S11 at least of the transistors is temperature dependent. I've found that removing my matching networks and just connecting everything directly improves things so I get a little less than -1dB of drift, but at the cost of about 10dB in overall gain...

I'm thinking that maybe if I use matching networks with broader bandwidth and more poles, the drift of the transistors' S parameters won't impact gain so much... does that even make sense?

 

[tony_lth]

The FET or transistor S21 degrades about 0.03dB/deg. Sorry, I can't remember the exact data.

 

[mtwieg]

The FET or transistor S21 degrades about 0.03dB/deg. Sorry, I can't remember the exact data.

Do you have any literature on this? The only thing I get when searching for temperature drift on RF FETs is drain current regulation/compensation, which isn't my problem. Nothing on S parameter drift. And I find it pretty strange that S21 decreases as temp increases, rather than increase. But that 0.03dB/deg figure seems to match up somewhat to my measurements.

 

[mtwieg]

Update: I've somewhat solved the issue. I changed my transistors to MRF173s (higher power) and dropped a large heatsink on them, and that, along with the current regulation, gave gain drift of 0.2dB or less, which was great. However I then ran into an issue where under certain circumstances my bias feedback loop would become unstable. Strangely it only occurred with the RF input at certain power levels (around the IP1 level of the amplifier). I can't seem to get rid of it without slowing down the loop response so much that it takes more than 100us to stabilize, which is unacceptable for my application. So I've reverted back to using no feedback or regulation at all, and just using the bigger transistors and heatsink. I also put larger source resistors (about 3ohms) in. Seems to still work pretty well; seeing maybe 0.3dB of drift, even when it gets very hot. I'm not a big fan of just strong arming a circuit into working, but it's good enough for me...

 

[vfone]

As I mentioned previously, MRF136 (15W output) it will not work with your requirements, and of course a bigger one (80W out) which you are using now is fine.
Always bigger is better. Bigger car, bigger TV, bigger house...only wife should be thinner and thinner