Designing/building 2W RF amplifier for 40+/-5MHz fixed frequency sinewave

[JohnnyMalaria]

Hello,

I've been reading a lot about RF amplification and, frankly, am quite overwhelmed. There seem to be many variations depending on application and I don't want to start down the wrong tracks(s). My particularly application is (I think) relatively trivial as far as specifications are concerned.

Input signal:

-2.0dBm (0.25Vpeak @ 50ohm) sinewave at 39-41MHz (from DDS I already have). The amplitude and frequency will not change. The target is 40MHz.

Output signal:

+32dBm (12.5Vpeak @ 50ohm/1.6W). It doesn't matter if it is inverted with respect to the input.

My load is an acousto-optic modulator that has a functional bandwidth of +/-5MHz. I'm not concerned about harmonics because the modulator won't "see" them but flat amplitude is important (as provided by the input signal).

Supply voltage:

+20V DC (single supply).

I actually need two such amps - one operating at, say, 40.000000MHz and another at 40.050000MHz. I can already do this with two commercial amps I have but they are physically too big and designed for significantly higher amplification.

Thank you,

John.

[T3sl4co1l]

Funny, I made one of these,
https://www.seventransistorlabs.com/Images/Wideband_Amp_4W_50MHz_Sch.png
https://www.seventransistorlabs.com/Images/Wideband_Amp_4W_50MHz.jpg
which seems to do what you want.  It could be simplified for the narrower required bandwidth, and higher efficiency.  RF rather than power MOS types may be desirable as well, but not the crazy hot (1GHz+) RF types you get today; something MRF or the like, classic HAM/CB/etc. type would be best.  I suppose they really wouldn't be the best though, because they're always so stupidly expensive...

Tim

[ahbushnell]

Funny, I made one of these,
https://www.seventransistorlabs.com/Images/Wideband_Amp_4W_50MHz_Sch.png
https://www.seventransistorlabs.com/Images/Wideband_Amp_4W_50MHz.jpg
which seems to do what you want.  It could be simplified for the narrower required bandwidth, and higher efficiency.  RF rather than power MOS types may be desirable as well, but not the crazy hot (1GHz+) RF types you get today; something MRF or the like, classic HAM/CB/etc. type would be best.  I suppose they really wouldn't be the best though, because they're always so stupidly expensive...

Tim

Is it correct that one winding on L1 is grounded on both ends on the same winding? 

[T3sl4co1l]

It is.  It's not so much an inductor (the winding is more just for convenience taking up the length), as a delay line.

Not really necessary anyway, as the length is under two nanoseconds, hardly a problem at 50, even 100MHz.

I like going overboard with transmission line transformers, because it's the most theoretically satisfying construction method, and isn't much more difficult than a conventional windup.

As a result, the phase matching in this design is probably good to 200, even 500MHz, though the transistors aren't nearly as good so it doesn't matter.

IMHO, it's less effort to design TLTs, than a conventional design that definitely will meet requirements.  That is, a conventional one might poop out at 20 or 100MHz, and not have the right impedance, or have excess phase shift or imbalance or whatever.  Whereas using TLT concepts it's just geometrically correct to begin with.

The notation for TLTs is kind of an abuse of coupled inductors; if it helps, redraw each set of inductors as a [set of] transmission line [pairs], and consider the EM waves (and port impedances) entering and exiting them.

Tim

[ahbushnell]

It is.  It's not so much an inductor (the winding is more just for convenience taking up the length), as a delay line.

Not really necessary anyway, as the length is under two nanoseconds, hardly a problem at 50, even 100MHz.

I like going overboard with transmission line transformers, because it's the most theoretically satisfying construction method, and isn't much more difficult than a conventional windup.

As a result, the phase matching in this design is probably good to 200, even 500MHz, though the transistors aren't nearly as good so it doesn't matter.

IMHO, it's less effort to design TLTs, than a conventional design that definitely will meet requirements.  That is, a conventional one might poop out at 20 or 100MHz, and not have the right impedance, or have excess phase shift or imbalance or whatever.  Whereas using TLT concepts it's just geometrically correct to begin with.

The notation for TLTs is kind of an abuse of coupled inductors; if it helps, redraw each set of inductors as a [set of] transmission line [pairs], and consider the EM waves (and port impedances) entering and exiting them.

Tim

Is it a twisted pair or coaxial cable?

[JohnnyMalaria]

Thank you very much, Tim. Seeing how the end product should look is really helpful. It's certainly much smaller than the amps I use at the moment. I'll study the schematic closely - I'm sure there's a lot I won't understand which is a good thing since it'll force me to learn

John.

[mrtn]

A few years ago I designed and built an RF amplifier to drive AOMs.
Mine was for 100 MHz but it's sufficiently wideband that it would probably work from 50 MHz - 150 MHz.
It's class A, using a large RF MOSFET. It will easily do 5+ watts.

I also designed and built the 4 channel DDS generator with VME interface board  to go with it.

[JohnnyMalaria]

Very impressive - thank you!

I'm building a modern version of an instrument I developed 30 years ago for my PhD. At the time, Brimrose Corp designed me a custom SSB modulator/amp. It worked well but cost £5000 in 1987 money. This time around I got the correct amps for the AOMs off eBay for $350 (instead of the $1500 retail) and an amazing 4-channel DDS from China off eBay for $99. The performance is vastly superior. AOM manufacturers typically charge $4000 for a 2-channel DDS system. $450 for what would be pushing $8K - happy with that