Update on capacitors ?

[W4NEQ]

Hi folks,

I'm new to the forum and new to saturated fet amplifiers. I've never done any switching ps design. After emailing Steve for some advice and PWM kit availability, I've decided to build an 8 fet 400 watt rig for 80 & 160 - very close to Steve's published design - and am checking out availabilities of key parts. All seemed well until I tried to source these exotic ATC type caps. I've done limited surface mount work, and the RF caps I'm use to are either the fist-size mica caps - or the "unelco" style.

I'm gathering that the voltage rating should be similar to the fet - about a Kv? And really low esr is important?

What are folks using these days? The closest I could find were some that were almost $40 each.

Also, I see that Steve bandswitches the drain-to-ground caps with relays? Series inductance isn't an issue?

Could I use fewer drain-to-ground caps, and then resonate the push-pull output with a floating 1000 pf vac var between opposing drains?

If anyone has the exotic caps for sale I'm interested.

Chris

[steve_qix]

Hi !

The ATC caps are out there, but they're not in common distribution. There are some other caps some folks have been trying, and they seem to work just as well. Will report back on this.

Low ESR is the key. They caps have to be low ESR.

For bandswitching the shunt capacitors, use a good, low inductance relay. I use the 30 Amp automotive relays available from Radio Shack. These relays have fairly low inductance on their own, and you can solder directly to the relay itself, and not use the connecting tabs at all. Doing it this way really keeps the stray inductance down. It's not THAT critical switching between 160 and 75 meters. With one rig I built, I used a piece of flat stock about 1.2 inches long, and this was connected to the 160 meter shunt cap (one for each module), the other end of the shunt cap went to a relay to ground.

Hope this helps!

Regards,

Steve

[W4NEQ]

Thanks Steve, and please forgive me if I'm having trouble with the class D versus E distinction.

So a class D amp like this:

http://www.w1vd.com/375WclassD.pdf

Which uses a more recognizable (to me) output network, is inferior in terms of efficiency ?

After reading this lucid explanation:

http://rfic.eecs.berkeley.edu/~niknejad ... EF_PAs.pdf

I gather that the primary difference between the two classes is essentially the way the output network appears to harmonic energy - with class E presenting a higher impedance, thus allowing faster rise and fall times, approaching squarewave voltage during the off cycle, which rings in the output, ultimately being flywheel'd back to fundamental frequency energy through the series LC? Is this correct?

What kind of efficiency differences would be expected between D and E in our application?

Chris

PS. I still haven't located any shunt capacitors - have folks obtained them directly from the manufacturer by ordering quantity?

[W4NEQ]

Being new to saturated mode RF generation, let me reduce the question in my last post:

1. Is the only difference between class D and E the nature of the output network - that is with Class E harmonics must see a higher impedance?


2. In 160 / 80 meter AM service, what kind of efficiency difference is typical between D and E ?

Chris

[steve_qix]

Everything that follows assumes real power with large MOSFETs like the FQA11N90, etc.

Class E and class D are virtually identical up to and including the gates of the MOSFETs.

At lower frequencies (like 160 meters) there is probably little, if any difference in efficiency, assuming you don't have "on time" overlap and things like that.

As the frequency is increased, it becomes a bit trickier to get class D to work perfectly, and you are starting to get into switching losses involving the output capacitance of the MOSFET. But, things can still work reasonably up into the lower HF range if the output network is properly designed, the layout is right, and you get the drive to work correctly. The higher the frequency, the harder it gets.

You can be a lot sloppier with class E which is probably why I use it The tuned circuit lets you match into a widely varying load, and the smooth class E waveform is fairly easy to work with from a combining and distribution standpoint within the RF amplifier, without adding ringing and causing other problems. Stray inductance in the output network is generally not an issue. The only real critical path from a stray inductance standpoint is that of the shunt capacitor to ground.

At lower frequencies where stray inductance is not a problem, and switching losses are low, you should be able to run more power with class D, assuming the same devices. In theory, the RF amplifier can be smaller and less expensive to build because there are no tuned circuits or variable capacitors, which require space and cost. However, the tuning/matching function is generally required somewhere in the chain, so the overall size/cost advantage will usually be offset elsewhere in the system.

So, both systems have their advantages. I find class E easier to get working, and less critical with respect to layout, and very easy to combine modules to get a lot of power. And for me personally using coax feed and not using an antenna tuner, the variable output network is necessary.