Power control of Class D Rf AMP

[dnyberg2]

I need to design a class D low power RF amp that I can control the RF output of without tampering with the input signal. I need about 200mW out @ 50 MHz but the input drive is a digital coded pulse train I can't tamper with in any way. I need class D for best efficiency and battery life. How does one go about controlling the output power of such a design?

 

[Greg-ulate]

A class D amplifier uses pulse width modulation to average out into a signal with a variable amplitude. This means that the PWM frequency needs to be a lot higher than the output frequency. You'd need some fancy transistors to switch at >500MHz... Class D amps are efficient because the losses only occur while the transistors are switching, so when the switching time is small compared to the on and off time, there is less loss than with an analog amp. If you switch at 500MHz won't that benefit be lost? I don't know anything about RF amplifiers.

 

[Enthalpy]

Class D is not used at 50MHz. Class C for CW especially FM, B and variants is linearity is needed in AM and APM, and some exotic ones whose name was forgotten as soon as created, like F1, F2, G...

 

[dnyberg2]

Okay, I get it. What everyone is telling me is basically you can't change the output RF level of a Class D RF amp without pulse width modulating it or varying the supply voltage. I presently vary the supply voltage but that costs me efficiency because of the CKT used to vary the supply to the present amp. I want to get rid of that.

I don't have the luxury of pulse width modulating. It would mess up my pristine digitally modulated data stream by making more complex than I can afford. What I have is a 50 MHz system that is very efficient because it uses a Class D RF approach to creating a 200mW digitally modulated signal.

What I need to do is make that source a driving level only RF signal that I can then amplify up to the 200mW I need using an RF amplifier topology that is as efficient as possible outside of Class D. This 2nd RF amplifier technology is the one I get lost on.

What is the best RF amplifier topology one can use for the application above that has low quiescent power waste?

 

[Averagesupernova]

Ummmmm, not sure what you really have. Class D by definition is pulse width modulation. But then you say you don't have the luxury of pulse width modulating. Just what do you have? Enthalpy already told you class D is not used at radio frequencies.

 

[dnyberg2]

Regardless, I am working with a Class D system @ 50 Mhz. The modulation is used is a 50MHz square wave on off keyed at the data rate. The output is pumped through a nice BPF so the output is useful RF. My dilemma is trying to create a new RF amp system that I can control the output RF PWR without having to tamper with the working source I have in any way. Let's just say I have a low PWR modulated RF source I need to amp up to ~200mW.

What is the best RF class to use that wastes as little DC PWR as possible?

 

[Averagesupernova]

I'm still not sure what you have. It makes no sense to me. I suspect you actually have class C but until I see a schematic I can't say. I suspect you already have the most efficient class possible. My first inclination is to lower the power supply voltage to lower the power. Can't do that because you will waste power in the voltage regulator? Use a switching regulator instead. You really are not giving us enough information.

 

[dnyberg2]

Averagesupernova, maybe I'm mis-speaking. Maybe what I have is better classified as Class D RF PWR OSC. And your right, we use a programmable switch-mode power supply now to control the output PWR. I'm hoping there is a cheaper PWR cost method to do this.

 

[Averagesupernova]

Cheaper is whatever it is. Do you understand where power is wasted as heat in various regulators and classes of amplifiers? I suspect you maybe do but haven't completely wrapped your head around it. It can be summed up this way: Suppose you have a switch in series with a 100 watt light bulb. When the switch is open, no power is dissipated in the switch. When the switch is closed, no power is dissipated in the switch (some will be due to imperfections in the switch but we will consider it zero). Anywhere in between open and closed power will be dissipated such as replacing the switch with a resistor. It is no different with transistors. Class C and class D amplifiers are on and off amplifiers.

 

[dnyberg2]

So, you're saying regardless of how I generate a driving RF signal, there is no benefit in power savings to be had by various classes of operation in RF amplifiers such as B, E or F as far as efficiency is concerned? If that's true, then why are there so many different classes to choose from aside from linearity and conduction issues??

 

[Averagesupernova]

No that is not at all what I am saying. A class B amplifier does not at all behave the same as a class C.

 

[dnyberg2]

Right, so my question to the forum is still...

What is the most efficient RF amplifier I can build to amplify this RF signal?
Class D is out of the running.

I need to choose the topology that wastes the least amount of energy to achieve ~ 200mW in my case.

Anybody know what that is?

 

[Averagesupernova]

I don't think anyone knows what the signal is yet. What is it? Something phase or frequency modulated? Class C is the most efficient RF amplifier but it is not a linear amplifier such as class A, B or A/B. We know you need about 200 mW but don't know what you are starting out with. So what do you need?

 

[dnyberg2]

Allow me to try once more. My problem begins with a coded telemetry data stream at over a megabit per second.

A 50 MHz OSC is on off keyed by this data stream. This keyed output is passed through a BPF and on out to a load, let's call it 50 ohms for argument sake.

This is the signal I need to amplify up to ~ 200mW. A Class D power oscillator was used and employed a programmable switch-mode power supply to control the output level by varying the power supply to the Class D oscillator. That method of using a programmable switch-mode power supply causes other system issues and I'd like to remove that as the main control method for controlling the RF power output. I cannot change how the RF signal is created but I know darn well, there are many ways to amplify RF. What does it matter that my RF source happens to be modulated with a digital codded data stream?

What is the most efficient RF amplifier topology to use in order to amplify this RF signal I have, up to the 200mW I need?

 

[Averagesupernova]

The difference is that you cannot amplify without major distortion an AM or SSB signal for instance with a class C amplifer. You CAN amplify an FM or CW signal with a class C amplifier. What you are describing is CW (continuous wave). Up until now it was unclear what your signal was. If you are SURE that the signal is CW with definite on and definite off times class C is what you want. If it is a type of AM where the signal doesn't go completely away then class C will not necessarily work. What is the data rate? It's hard to believe that anything very fast would use CW.

 

[dnyberg2]

The data rate is over 1mb/sec

 

[Averagesupernova]

I guarantee it is not CW. Probably FSK.

 

[dnyberg2]

It is more like ask but the coding scheme keeps the duty cycle at 50 percent minimum all the time

 

[Averagesupernova]

It is more like ask but the coding scheme keeps the duty cycle at 50 percent minimum all the time

I don't understand what you are saying here. What is ASK?

 

[dnyberg2]

Amplitude Shift Keying.

 

[dnyberg2]

Because the data keys the RF square wave on and off at the data rate it acts more like ASKing

 

[Averagesupernova]

I still don't believe that the RF is turned completely off. I suspect it is AM. I have never heard of anything referred to as Amplitude Shift Keying. Can you see on a scope what this signal looks like after its filtered?

 

[dnyberg2]

Looks like modulated RF with a peak at plus and minus 1 MHz around the carrier on a specan...

 

[Averagesupernova]

A CW signal would look similar on a spec an but it would be a lot noisier than AM. That is why I don't think it is CW. I don't think a class C amplifier will work for you. When you look at it on an OSCILLOSCOPE, not a spectrum analyzer, does the RF go to zero at any point or does the amplitude just get lower?

 

[dnyberg2]

Let me run that test in the morning. I'll get back to you. And thanks for your patience and help!

 

[Averagesupernova]

Not sure if you have fully understood what CW (continuous wave) is. It SOUNDS like a carrier on all the time with no change. It is not. It is a carrier that is switched on and off like morse code. Generally not fast. Every time the carrier is turned off or on it generates extra noise in the spectrum around the carrier. Anyway, just thought I'd throw that in. I'll be waiting for your results.

 

[dnyberg2]

Averagesupernova, As a one time ham, I know CW intimately. Attached are some screenshots from my RF source. The output is turned way down and I used a 100 MHz digital scope to capture these with the 50 ohm termination turned on. There is some proprietary info in here so the scales have been omitted on purpose, but trust me, this is ~50 MHz.

 

[Averagesupernova]

As a one time ham I am surprised you needed to ask this question. Looks to me like class C will work for you.

 

[dnyberg2]

Hey! I was a kid! ;-) Now, trouble is, correct me if I'm wrong, but doesn't class C have bad efficiency? Maybe a class D with the switcher for the supply is better?

 

[Averagesupernova]

Sigh...
No, class C is the most efficient of RF amplifiers. Class D is NOT used at radio frequencies. I am not sure you understand what a class D amp does. A class D amplifier turns the output either full on or fully off so very little power is ever dissipated in the transistor. Suppose we have a switching frequency of 100 Khz and the duty cycle is varied and the output is low-pass filtered. If the duty cycle is varied at 100 Hz, a signal of 100 Hz comes out of the low pass filter. Class D can be used in switch mode power supplies, but you still need the RF amplifier in your case. The RF amp and the class D switcher are two separate parts.

 

[dnyberg2]

Okay, so why do the architects of the system I'm suffering with use the term class D?
There is a fixed 50 MHz source that is being turned on and off (OOK) at the data rate. I have seen the schematic on a white board. Its two or four fets in series on a die whose power supply is controlled by the programmable switcher I told you about... I've heard the term Class D power osc flung around as well...

 

[Averagesupernova]

Probably because they started with a square wave generator which technically would be class D and filtered the hell out of it. Class C is closely related to class D but C is made to run at a resonant frequency. There is a parallel tank circuit in place of the collector resistor. I have no idea why they would run all the way out to the last amplifier stage with a square wave. Good engineering practice would be to filter as soon as possible in the signal chain. I guess I wouldn't be too surprised if what you have is actually class C but for whatever reason is mis-named.

 

[Greg-ulate]

So you have a On-Off-Keyed 50MHz signal switched on and off at 1MHz and you want a adjustable gain amplifier that is as efficient as possible to step up the power?

A fixed gain amplifier won't work for you because you have found it necessary to vary the output of your current amplifier by adjusting its power supply voltage, correct?

I'm looking at the diagram for the Class C amplifier on wikipedia. Would varying the value of the collector resistor change the gain of that circuit?