Efficiency of RF signals from halfbridge using HEMT

[sourcecharge]

So, I've got a noob question regarding RF. 

I am interested in the EM drive and I would like to do some experiments regarding this new interesting technology that has been validated by researchers in china and in NASA.

Basically this all boils down to trying to make efficient 1 to 2 kw of RF AC square waves in the 1.5 to 6 GHz range with a duty cycle as close to 100% as possible.

I've been looking at HEMTs to basically make a half-bridge, but I have no experience in using them.

What I am expecting to be able to design is a variable frequency, variable duty cycle, variable amplitude type AC square wave driver using HEMTs or something similar.

I am in no way ready to even start working on this, but I have been thinking about it for awhile.

I would suspect that I would need to see the waveform on a scope, but those scopes are like $20k

so:

Would this even be plausible?

And second, is there a way of being able to observe the waveform using a 300MHz scope?

[sourcecharge]

Thx for your reply,

Do you have experience with HEMTs?

Would a class E (amp?) be able to directly drive a resonant cavity with a 1:1 efficiency with a AC square wave input?

[ejeffrey]

Would this even be plausible?

Honestly, not really.  High power RF is already a pretty tough place to start as a self professed n00b, and trying to make a GHz square wave generator with a HEMT half bridge is going to be... really hard.  Doing blind without a suitable instrument to actually see it working is going to be pretty much impossible.

Also, all the stuff about high duty cycle square waves is just voodoo nonsense (along with the rest of EmDrive).  With no principle of how it should work or clear demonstration of it doing anything other than generating a lot of heat, things like wacky drive waveforms (that make no difference to a cavity anyway), are just needless complications to confuse people.

[medical-nerd]

Hiya

Just a thought - could you use a microwave magnetron feeding into a tuned cavity?

Cheers

[sourcecharge]

Would this even be plausible?

Honestly, not really.  High power RF is already a pretty tough place to start as a self professed n00b, and trying to make a GHz square wave generator with a HEMT half bridge is going to be... really hard. Doing blind without a suitable instrument to actually see it working is going to be pretty much impossible.

Also, all the stuff about high duty cycle square waves is just voodoo nonsense (along with the rest of EmDrive).  With no principle of how it should work or clear demonstration of it doing anything other than generating a lot of heat, things like wacky drive waveforms (that make no difference to a cavity anyway), are just needless complications to confuse people.

Ya, no scope, that would be very hard.  There's no backyard hilbilly shadetree DIY workaround with a 300 MHz scope?  I kinda figured there wouldn't be.

If I were to start working on this, I probably would have to rent a scope that could handle it.

Square waves are actually the most efficient input waveforms for resonant converters.

As to the duty cycle, you are correct, I did not say the terminology right.  It's not duty cycle, it's duty factor, sorry.

Duty factor being the amount of time on(+&-) / total time of the period.

The higher the duty factor, the more efficient the output.

The voodoo of the EmDrive is another story.

There is no set understanding, but some,

2.4 Ghz and a frustum made out of copper.

Everyone seems to be using the method of creating RF power using Magnetrons, which are actually not that efficient, and have sinusoidal output. Magnetrons also only have one output frequency so the resonant cavity would have to change for the inputs signal.

If the input waveform could be changed to a AC square wave, using a lead follower orientation (half bridge), along with a variable duty factor, frequency, and amplitude, the efficiency would increase drastically simply by the driving circuits, along with a versatility of having variable frequency, the input signal could match the cavity instead of the other way around.  The variable duty factor is so that the bridge can be adjusted to limit the switching losses. 

Do you have experience with HEMTs?

[rfeecs]

Do you have experience with HEMTs?

Yes, but that won't help much.  Try learning a little about microwave power amplifiers.  They are not switching power supplies.  Audio amplifier techniques don't work up in the GHz range.

This might be someplace to start:
http://ecee.colorado.edu/~ecen5014/PA-overview-Raab-2003.pdf

Magnetrons can be pretty efficient.  They would be hard to beat with a solid state amplifier.  There are tunable magnetrons where you can vary the frequency.

You have a high Q resonant cavity.  What would you expect to see with an oscilloscope besides a sine wave?

It sounds like you have the wrong approach and want to use the wrong tools.  But by all means have fun and go for it and see what happens.

[sourcecharge]

Do you have experience with HEMTs?

Yes, but that won't help much.  Try learning a little about microwave power amplifiers.  They are not switching power supplies.  Audio amplifier techniques don't work up in the GHz range.

This might be someplace to start:
http://ecee.colorado.edu/~ecen5014/PA-overview-Raab-2003.pdf

Magnetrons can be pretty efficient.  They would be hard to beat with a solid state amplifier.  There are tunable magnetrons where you can vary the frequency.

You have a high Q resonant cavity.  What would you expect to see with an oscilloscope besides a sine wave?

It sounds like you have the wrong approach and want to use the wrong tools.  But by all means have fun and go for it and see what happens.

Cool, not alot of people have experience with HEMTs...

Magnetrons...
https://www.quora.com/What-is-the-magnetron-efficiency-in-microwaves-and-how-it-varies
<75%

https://en.wikipedia.org/wiki/Cavity_magnetron
"The modern magnetron is a fairly efficient device. In a microwave oven, for instance, a 1.1-kilowatt input will generally create about 700 watts of microwave power, an efficiency of around 65%"

They are not lead follower type drivers.  The current they pull is in parallel to the load, and the load must be matched in order to get 75% efficiency or higher.

The output of magnetrons are sinusoidal and will not test my theory.

The scope is to monitor the input waveform switching turn off and turn on timing and verify resonance of the resonant wave-guide chamber (frustum).

Since you have experience with HEMTs, do you mind if I ask you basic questions about them to gain a better understand what would be possible?

1. Starting off with just a P ch HEMT, with it's source connected to a + DC power source, and the drain connected to a 50 ohm resistive load, how realistic would it be to make a + DC bias square wave at 5 GHZ?

2. Can the same be done with a N ch HEMT, with it's source connected to a - DC power source, and the drain connected to a 50 ohm resistive load, to make a - DC bias square wave at 5 GHZ?

These two questions will basically give the plausibility of such a driver.

Not sure why you think lead follower type drivers are inefficient, but if there is something that I don't understand about the switching losses or the driver current requirement, that is why I'm asking questions here so that I can figure out if this might be possible.

Thanks so far for your help..

[rfeecs]

Magnetrons...
https://www.quora.com/What-is-the-magnetron-efficiency-in-microwaves-and-how-it-varies
<75%

https://en.wikipedia.org/wiki/Cavity_magnetron
"The modern magnetron is a fairly efficient device. In a microwave oven, for instance, a 1.1-kilowatt input will generally create about 700 watts of microwave power, an efficiency of around 65%"

They are not lead follower type drivers.  The current they pull is in parallel to the load, and the load must be matched in order to get 75% efficiency or higher.

Like I said, they are pretty efficient.

Some transistor manufacturers are trying to push the idea of replacing magnetrons with solid state devices for microwave ovens.  You might want to have a look at what they are doing.

Since you have experience with HEMTs, do you mind if I ask you basic questions about them to gain a better understand what would be possible?

1. Stating off with just a P ch HEMT, with it's source connected to a + DC power source, and the drain connected to a 50 ohm resistive load, how realistic would it be to make a + DC bias square wave at 5 GHZ?

You are not going to find a P channel HEMT.  Hole mobility for GaAs or GaN is more than 10 times worse than electron mobility.  There is no point in making a P channel "High Electron Mobility Transistor."

2. Can the same be done with a N ch HEMT, with it's source connected to a - DC power source, and the drain connected to a 50 ohm resistive load, to make a - DC bias square wave at 5 GHZ?

Sure.  But not at 1000W.  That would require more than 300V, and that is not happening even with GaN.

Also realize that a lot of smart people have been studying how to make high efficiency microwave power amplifiers for a long time and come up with good ways of doing it that work.  The literature is out there.

One other thing you might want to consider.  For power transistors you are looking at maybe around $1 per Watt of output power.  This gets expensive very quickly, not even considering the cost of test equipment.

[Mechatrommer]

This gets expensive very quickly, not even considering the cost of test equipment.

any GHz test equipment that can accept > 50dB input? the price should be fun...

[sourcecharge]

I'll do more reading on the subject and try to familiarize myself with the current tech in SS Microwave ovens.  That seems like a good place to start.

I did a search for "Pch" HEMTs and you are correct, there doesn't seem to be any available.

That's a problem for a half bridge driver.   



Thanks for your help, and I know this was an expensive idea.  Maybe in the future I might find a Pch HEMT someday.

[sourcecharge]

Is this a Pch HEMT or is it just a fast pch mosfet?

https://www.mouser.com/ProductDetail/NXP-Freescale/MMRF5014H-500MHZ?qs=u4fy%2fsgLU9NFxjJ7T8xUZQ%3d%3d

At 1300 dollars there's no way I'm getting that....

[rfeecs]

That is a GaN HEMT.  N channel.

RF HEMTs are depletion mode.  They behave like JFETs.  With no gate voltage they are turned on.  With enough negative voltage from gate to source they turn off.

The gate is a Schottky junction.  The gate to source looks like a diode, so if you forward bias the gate to source, it will draw gate current.  GaN has a high band gap, so the forward diode voltage is about 2V.

So it is normally biased with a negative gate voltage and the gate draws very little current.  When RF is applied the gate may draw forward or reverse DC current as well as RF current.

[CopperCone]

The EM drive needs a square wave? Can you link that?

I actually made a microwave oven magnetron to N waveguide-coaxial adapter for testing these out.. but you need alot of dough for the other parts of the equipment. I guess I missed the square wave bit?

[ECEdesign]

If you didn't need the power you could use a VCO to get your unable oscillation and pass output through an inverter to get a square wave.  Maybe put a PA on the output to give the power? 

[CopperCone]

above 50 watts a RF power amplifier with say, 10GHz bandwidht, will cost you like... 100k

[sourcecharge]

Ya, mouser looks like it classified this HEMT wrong.  Thanks for your in depth analysis of the function of depletion mode HEMTs.  This type of HEMT will not be able to do what I had hoped it would.

The EM drive is experimental, and it still is being researched by top govt and scientific agencies. 

The information regarding the EM drive is mostly coming out of NASA and DIY experimenters.

I wanted to try something different, with respect to how the actual signals were being generated.

It seems as if this is not just something I was thinking about for this one application.

Cost and design capabilities are going to be the most limiting factors for such experiments granted but I'm hoping over time, the industry will come out with new HEMTs or similar which give a equivalent P channel switch so that a lead follower type of signal can be made at realistic prices.

I'm theorizing that type of driver would not require a load impedance to be matched with the input signal because the load impedance would be in series, and not parallel.  Only the power requirement would have to be met.  This by itself would increase the power across the resonant load and therefor the efficiency of the EM drive.

Thanks again everyone, and I will keep my eye out..