Andy, that is a whole lot of writing there, and I'm not really getting what you're trying to say here.
The diode to ground in the bias circuit of a bipolar transistor isn't there to define the class of operation, it's there to provide rudimentary thermal tracking for the bias current. The base-emitter junction of the transistor and the p-n junction in the diode both have a negative temperature coefficient (NTC) and as they both heat up the voltage drop across the p-n junction drops causing more current draw. In theory, the diode will pull current away from the b-e junction as they both heat up and prevent the b-e junction from drawing more current and causing a thermal runaway. This is why those diodes are held to the transistor with thermally conductive glue. Bipolar operation isn't really relevant to the problem with these mosfets though.
I get that the impedance and RF isolation of the gate biasing circuit can be an important factor when using Mosfets, especially when the gate has a non-symmetrical (distorted) drive signal. If the impedance of the bias supply is too high, or not properly decoupled from the RF, a non-symmetrical drive signal can cause the gate bias voltage to drift away from it's steady-state point and can change the operating class of the FET, sometimes this is minimal, but I've seen it both collapse the bias voltage and push it to the point where the FET never turns off. In the Qixiang based/derived radios (Strykers, Anytones, ect.), there is a reversed biased (in relation to the bias voltage) 1N4148 diode in series with a low value resistor (330R for the finals and 820R for the drivers I think) across the gate-source connection. I think this is supposed to provide a small amount of DC voltage offset across the capacitor that is driving the gate, essentially providing a slight increase in bias point when RF is present vs. when there is no signal at the gate. I've never actually tried scoping the gate with and without this D-R in place to see if there is an offset change, so this is purely speculation. The biasing setup still isn't the issue I'm/we're facing with driving these 2030+ FETs though.
When you drive a mosfet gate with a really fast rising edge, which a 27MHz sine wave has, you get a step in the gate voltage as the FET begins to turn on significantly. This is caused by the miller effect of the gate-drain capacitance. With the gate voltage rising so fast, this will cause the drain voltage to fall very fast as the FET turns on, the Cgd will cause the falling drain voltage to pull charge away from the gate and causes a step in the rising edge of the gate voltage. Now, mosfets are voltage controlled devices, but due to the capacitance of the gate, combined with the miller effect, driving the gate very fast requires a significant amount of current to get through that stepping point without having a step occur. This is something I've dealt with when building switch-mode power supplies. In my Connex 4400 Turbo, I was able to provide this current in the amp section by adding an additional turn to the primary of the input transformer (total of 3 turns). This allowed me to drive the 2030+ gates hard enough to eliminate the step in the gate signal, minimizing distortion, and allowed for 80W PEP with only 4W PEP input signal. Getting this type of drive without the use of a transformer in a single ended configuration has been out of my reach though. I'm not too super concerned, as I do have other options to use in my radios. My posting here was merely a curiosity if anyone ever figured this out. As I said earlier, I'm absolutely no expert, and there is obviously something I'm missing, but it's nice to know that I'm not the only one having the same exact problem with these things.
Tallman: The thermal runaway on FETs is kind of hit or miss these days. It depends on what FET technology you're working with. Older FET designs like the IRF510/520/530 were like you described with PTC, allowing easy current sharing and gate biasing for RF applications. Many of the newer switching FET technologies actually have NTC and can suffer from thermal runaway just like a bipolar transistor. I was just reading about this the other day and I think it had something to do with the evolution of how the gate was designed and what materials were used in the gate's construction. It's pretty easy to tell what kind of FET you have. Put it in a test jig and apply a constant biasing voltage to the gate while measuring the drain current. As the FET heats up, the drain current will either fall or rise depending on the generation of FET tech that is used inside that device. I believe these 2030+ FETs have the NTC type gate design, but I haven't set them up and measured them.
Those cheap component testers are pretty nice for quick device measurements. I used a similar one at a friend's house and I'm awaiting mine to be delivered.
Eldorado: They were cheap, so I'm not super worried about it if I never make use of them in another radio. The 4 in my Connex 4400T work fine, but it took more work to get them there than it would have taken me to match out a set of IRF520s and go that route. I've built a couple homebrew amateur radio rigs and amplifiers, but never anything that was presentable or could be used in a mobile application. I mentioned earlier about maybe prototyping an RDxxHHF1 TX chain into one of my CBs just to see if I could get it to work and keep it cool. Sort of like the old Connex 4300HP with the 2SC2290 as a final. In addition to brainstorming that prototype, I've also been brainstorming designing my own mobile 10/11m mobile rig with a 2 x RD100HHF1 amp section built into it. Not a strap-on like the RCI based radios, but built into the main board like a Cobra 200GTL with the 2 x 2SC2290. There is a lot of stuff I'd have to learn to build this, such as ANL and Noise Blanker circuitry designs, and other things that are great in a mobile, but not really necessary in a base rig. With the cheap and easy availability of professional custom circuit boards from China, and the ease of frequency generation you can get with a simple micro-controller and something like a Si5351, I think I could roll my own experimental prototype mobile rig.
I guess my ass better get to work. If I come up with anything worthy of sharing, I'll post it up on the forums here. It seems like you guys have a pretty cool community here.
- Brad, KC3MOP
