High power RF amplifier at fixed frequency

[BillKet]

Hello! I have a signal at fixed frequency of 4 MHz and amplitude of 0.075 V. I need to amplify it to ~75 V with the same frequency. I checked online a bit and I found either cheap amplifier that can't amplify this much, or very expensive ones that (almost) can (I might need to use a smaller amplifier beforehand). I am still trying to find the best option, but for basically everything I found online, they claim large bandwidth, which I assume it's a contributing factor to the high price, too. However, I just need it at 4 MHz. Naively, I imagine one can build a simple RLC circuit for this. Of course in practice there are other factors to account for, so it is not really trivial, but I assume it would be a lot easier/cheaper to have an amplifier at a fixed frequency, rather than one with a broad range. Can anyone advice me on what is the best way to go (or even an amplifier model)? Are there comercially available amplifier with a fixed (or narrow band) frequency? Thank you!

 

[DaveE]

At only 4MHz, I wouldn't think wide band is too much harder (more expensive) than narrow band. I think the cost is due to the high voltage/power. That's over 100W (@ 50Ω) plus there are safety concerns that (should) make the construction a bit harder.

Also, no, you can't get that much gain (60dB) from an LC tank, especially at high power levels.

I would imagine that the cheapest stuff is in the HAM world. Maybe something like this:
https://www.ebay.com/itm/1241173116...gk6HoGKTGQMwjPsoy8O0ar49s9opgazAaAi_VEALw_wcB

 

[berkeman]

I need to amplify it to ~75 V

Can you say what the application is? Hopefully the application will keep all of that RF energy contained, so none can get out and radiate away to cause problems with local comms in the area...

 

[BillKet]

Can you say what the application is? Hopefully the application will keep all of that RF energy contained, so none can get out and radiate away to cause problems with local comms in the area...

It is used to drive an electric optical modulator. It is this one: https://www.thorlabs.com/thorproduct.cfm?partnumber=EO-PM-NR-C2. If you look at the data sheet and do the math, for my application, I need ~70-80 Volts.

 

[DaveE]

It is used to drive an electric optical modulator. It is this one: https://www.thorlabs.com/thorproduct.cfm?partnumber=EO-PM-NR-C2. If you look at the data sheet and do the math, for my application, I need ~70-80 Volts.

Make sure to check out the load impedance. These things often look like a capacitor and may not be compatible with cheap RF amps. I see a smith chart in your future, LOL.

 

[berkeman]

It is this one: https://www.thorlabs.com/thorproduct.cfm?partnumber=EO-PM-NR-C2.

Which of the 2 versions? If the non-resonant one, you will likely need a -6dB matching pad between your amp and the EO device, so that increases the output voltage required from your amp...

 

[BillKet]

Make sure to check out the load impedance. These things often look like a capacitor and may not be compatible with cheap RF amps. I see a smith chart in your future, LOL.

It seems like this one is a purely capacitive load (12-14pF)

 

[BillKet]

Which of the 2 versions? If the non-resonant one, you will likely need a -6dB matching pad between your amp and the EO device, so that increases the output voltage required from your amp...

It is the non-resonant one. What is a matching pad (I am just getting started with working with RF for this project)?

 

[berkeman]

What is a matching pad?

When a 50 Ohm output matched amplifier drives a non-50 Ohm load, it's important to use a matching network ("pad") between the amp and the load, to avoid damaging the amplifier. For example, in some RF immunity testing that I have set up in my work lab right now, I use a 75W power amp to drive a coupling coil with signals from 150kHz to 80MHz to test the immunity of IoT devices that we design and build. The input impedance of the coil varies with frequency, so a -6dB matching pad is placed between the amp and the coil. That reduces the drive available to the coil, but is necessary (and a part of the standard test setup definition) in order not to damage the amp.

 

[Baluncore]

I would consider making the high capacitance load part of a power oscillator, then phase lock that oscillator to the input signal. That will handle the load impedance matching, and increase the efficiency of the system. Unfortunately, that gets more involved with components.

Whichever way you go, the high capacitance load will require a tuned parallel inductor to neutralise the load presented to the driving amplifier. Making that inductor a tapped auto-transformer, could significantly reduce the amplifier specifications.

To neutralise 13 pF, at a frequency of 4 MHz, will require an inductance of 122 uH. The reactance is about 3k ohms, so it can be wound as quite a high-Q matching transformer, to be driven by a lower voltage amplifier with Zo = 50 ohms.

 

[DaveE]

Unfortunately, that gets more involved with components.

For someone that doesn't design amplifiers, this is a problem. Easier (but not better) to just buy an amp and put in a matching network. Particularly since he only cares about one frequency.

There is a lingering question of amplifier stability at other frequencies due to the reactive load. But that may just be a "try it and hope" situation. That gets into whether this is an experiment, or a real product. At the laser company, we would have to check that stuff to avoid pissed off customers later after we thought we were done and had moved onto a new project. Usually it wasn't hard to show that stuff wasn't likely; like, just give it a step input and see if it's well behaved.

 

[DaveE]

What is a matching pad

It is a network of additional electrical components added to the load (in this case) to change the impedance "seen" by the amplifier. In your case it could be as simple as a series inductor that resonates with the load capacitor at 4MHz to make it appear resistive. You'll need to know what the amplifier requires, often specified as VSWR.

https://www.analog.com/en/technical...d-smith-chart-impedance-maxim-integrated.html

 

[DaveE]

BTW, you'll be reading a lot about transmission line length and such. At 4MHz, you can mostly ignore that. A wavelength is going to be something like 50m of cable, so your's are probably close enough to 0. You can treat this as a "lumped element" circuit and be OK.