Continue to Site

Eng-Tips is the largest engineering community on the Internet

Intelligent Work Forums for Engineering Professionals

  • Congratulations IDS on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!

RF power amp that can withstand infinite VSWR 2

Status
Not open for further replies.

Renovator1

Industrial
Mar 14, 2003
72
Hey all. I got roped into building an RF power amplifier for a physics professor friend and I wanted some second opinions on which way to go with the design. The amplifier needs to deliver up to 300Vpp and 100W across the frequency range of 0.1 to 1 MHz into a purely capacitive load (~200-1000pF), preferably without tuning...

Much as I'd like to, I don't believe I can get away with whipping up a quick-and-dirty push-pull Class B out of some spare SMPS parts. It seems like I've only got two choices here, but I'd love to hear otherwise:

1) Single-ended Class A (with either a choke or a cascode current source feeding the drain to at least get 25% eff.)

2) high-speed current feedback op-amp driving a FET voltage gain stage then a FET current gain stage (ala EDN's Design Ideas from April 26, 2001).

The distortion requirement isn't too critical, so that's a plus, at least. Any ideas or comments would be most welcome.

-Jeff
 
Replies continue below

Recommended for you

The Book "The Art of Electronics" has a really good HV drive circuit-FET output. Used to drive things like ceramic tranducers.
 
I'm guessing you are referring to Fig. 3.75? That's definitely a clever circuit (love the "stealth" pulldown) but the upper frequency limit is only 1kHz. My feeling is that Miller effect in the common source driver stage makes speeding this circuit up to the 4000V/uS or so I need an exercise in futility...

 
That's the circuit. I havn't looked at it so I don't know what limits the bandwidth.

Power MOSFETs can get to the speed, voltage and power you need. Look at the offerings from IXYS.


I think what you need to build is a higher frequency audio amplifier without needing to consider crossover distortion. I have designed a couple for linear servo amplifier applications with a p-p amplitude of 100V and a 1 MHz bandwidth.
 
That sounds a lot like what I am aiming for - what kind of output stage did you use for the servo amp?

 
The first one had an N channel-P channel complementary symmetry (CS) output. A later version used N channel devices top and bottom.A PNP transistor was used to invert the lower gate drive signal snd translate the signal to the B- power supply rail. The CS is easier to design. I used a floating +/- 15 VDC power supply, common to the amplifier output) to make driving th gates easy.
 
Another thought, if your load doesn't need to be grounded, a bridge tied load configuration will ease some of the voltage requirements, though you'll need two amplifiers.
 
Thanks for the reply, jimkirk. I am familiar with Apex uTech's product, and even have several of their venerable products in stock, but, unfortunately, nothing they offer has the right combination of slew rate and voltage/current output. Or, at least, not according to my seat-o'-the-pants calculations.

I believe a BTL configuration would be ideal, but a sanity check would be appreciated. The load I'm trying to drive is an "ion guide", which is basically a stack of plates where each is driven 180 deg. out of phase to the next. It seems, then, that it would be logical to drive all of the even plates, say, with the non-inverting half of a BTL amp while all of the odd plates are driven by the inverting half. Sound reasonable?

Keeping in mind, though, that the amp will need a slew rate of ~2kV/uS to be able to deliver 300Vpp at 1MHz... Hence why I was thinking of using CFB op-amps in the feedback loop.

 
Hi Renovator, yes, that sounds reasonable. You might also try driving single plates, or groups of plates, with separate output stages, so each output is driving a fraction of the total capacitance. More complicated, but the demands on each stage will be considerably less.
 
You could also use a broadband matching transformer to either increase or decrease the driving impedance, a 10:1 untuned frequency operating range should be dead easy. It also makes push/pull operation feasible.

That opens up a few more possibilities for you. They may be long out of fashion, but vacuum tubes can still be useful sometimes for some unusual applications. They make quite good high frequency current sources with few worries about voltage transients destroying things.
 
jimkirk - I'll ask my friend if dividing the stack of plates into two or more sections is feasible. That's an otherwise excellent idea, which means it will probably be torpedoed - if you've ever been involved in a physics vs. engineers debate you know what I mean...

Warpspeed - This was precisely my initial plan: whip up a simple broadband transformer coupled push-pull amplifier with cheap off-the-shelf SMPS mosfets. Unfortunately, along the way I read that Class B and C biased amplifiers hate driving pure reactance loads because all of the output power is reflected back to the active devices where it results in either 2x or 4x the normal dissipation (depending on the circuit, etc.).

The one thing that really has me stumped here is that I can't determine at what frequency/power range one should abandon an "audio amplifier" style circuit and go with an "RF" one. Clearly, an audio amplifier could care less if it is asked to develop maximum output voltage with no load attached whereas this is not often true of the RF amp, yet, one need not be concerned with slew rate in an RF amp as long as the proper impedance matching network is used at the input and output... It's enough to drive a man to "drink the kool-aid", if you know what I mean.

IF the solution to using a Class B RF design is to simply up-size the heatsinking and double (quadruple) the voltage and/or current margins for the active devices then I'd consider such excellent trade-offs compared to either requiring 4x the power supply wattage or the vastly more complicated circuitry required to achieve 2kV/uS with an audio amplifier style circuit.

Once again, comments appreciated.
 
As you say, 1 MHz is either very high audio or very low RF. From the stand point of driving a capacitive load it doesn't really make any difference. What one is really interested in is "How much power do the drive transistors have to dissipate?" After a few wrong turns (never having thoutght about driving a pure capacitice load in my life), I came to the simple answer; for a capacitive load, no power is being consumed by the load (Duh)so the power dissipated by the output transistors is equal to the input power from the power supply(s). So assume +/-150 VDC supplies, at +/-1 Amp peak current (roughly 1000 pF at 1 MHz, 300 Vp-p), class B output.

The input power is the supply voltage times the average current. The average current is 1/pi x Ipeak.

Iave = 1/1.57 x 1 = 0.637 Amps

Pin = 150 VDC x 0.637 = 95.5 Watts

This is split between the two output transistors so each has to dissipate 48 Watts.

The peak power dissipationis is also high at about 200 Watts for each transistor but at 0.1 to 1 MHz the transistors will probably only care about the continous power.
 
I cannot help feeling that a conventional high powered audio amplifier will not be at all happy with a completely capacitive load, especially right at the upper limit of it's frequency range.

One application that has just come to mind is the driving of large electrostatic loudspeakers to several hundred volts peak to peak. While your drive current will be much greater, the concept is probably not that much different.

In the past this has sometimes been done by connecting the speaker directly to the anodes of a tube type push pull amplifier, where the existing center tapped output transformer was terminated in a nominal high power load resistor. Negative feedback from the transformer secondary kept distortion reasonably low, and the load resistor damped the transformer inductance and any resonance. It may not have been ideal, but it was practical.





 
sreid - Your analysis agrees with mine, but one question that remains is at what point does VSWR enter into the picture? In an RF amplifier, 100% mismatch between amplifier and load occurs whenever the load is one of the following -

* a short
* an open
* a pure capacitance
* a pure inductance

- and this mismatch results in all of the forward power being reflected back to the active devices (less losses in cables, etc...) where it results in twice as much dissipation for a Class B amplifier but no change in dissipation for a Class A amplifier (according to vendors of amplifiers for EMC compliance testing)

Hence, why there is much pulling of hairs and gnashing of teeths around here...


Warpspeed - you are truly a master of understatement... ;)

Designing the feedback loop for an amplifier that is unconditionally stable at up to 1MHz while pushing significant amounts of power around will not be trivial, hence the near-resignation to just live with wasting 3W of power for every 1W delivered to (then immediately returned by) the load...

The unbelievably dedicated people/nutcases at DIYAudio.com attempt to design an amplifier capable of directly driving EL speakers every so often but my last search of their archives didn't turn up anything promising. Seems they all end up resorting to a transformer coupled output - which is fine, mind you - but even then their designs tend to blow up with distressing frequency (probably from not being compensated at all...).

But hey, if this were an easy problem to solve I wouldn't be begging for help on forums such as this one!?! :)

 
I believe your original concept of building a high power complimentary class B active current source/sink, and just live with the abysmal power dissipation and inefficiency.

A whole bunch of high voltage mosfets might work, to physically distribute the horrific power dissipation. Your idea of a cascode (common gate) output circuit, would certainly help with the Miller feedback problem.
 
You may be agonizing too much about the difficulities of this design project.

1) A Class A amplifier should be rejected due to power dissipation in the drive transistor.

2) A Class C amplifier only works with a tuned circuit at one frequency.

3) You are left with a "Push-Pull" class B amplifier as the only viable alternative.

4) Driving a capacitive load may have some unique problems;

a) Heat dissipation; you have recognized this and are
dealing with the problem.

b) Instability; caused by too much phase shift from
feedback at the amplifier output. If this is a
problem it can probably be solve with a small
resistor in series with the output.

I've taken this week off work. If you send me your email address to stevelreid-at-aol-dot-com I'll try to draw up a rough schematic of what I think will work.

 
There will be no problems with feedback phase shift (or instability) if the current passing through the load is monitored with a current shunt, and the resulting voltage fed back.

This will be a high frequency current source. What the voltage across the load does is of no real interest beyond any required peak voltage clamping. A current driver like this should be able to work equally well into a dead short.
 
If controlling the exact amplitude of the voltage across the load is important, that could be done with amplitude feedback, and some type of automatic gain control system, such as an analog multiplier, ahead of the power amplifier.
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor