High Frequency Broadband Amplifier / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

High Frequency Broadband Amplifier

1 2 next

Move Thread

LAN_403

Dr. Who

Sat Dec 10 2011, 12:11AM

Registered Member #326 Joined: Sat Mar 18 2006, 01:12PM
Location: Glasgow, Scotland
Posts: 66

I'm attempting to design a broadband amp to drive piezoelectric transducers. Input voltage is 3V pk-pk from a 50 ohm generator. Frequency is 1-10 MHz and the transducers have impedances ranging 15 Kohm down to as little as 40 ohms, and can include up to 150 pF capacitance. I want a 30V pk=pk output voltage.

The best I've come up with after quite a bit of experimenting with component values in LTSpice that gives a clean sinusoidal output with no distortion, saturation, clipping, etc is this:

1323475371 326 FT0 Hf Broadband Amp Small

R2 and C2 represent a typical load.

The common-emitter 2N2222 provides voltage gain, and the IRF610 source follower provides current gain to drive the low-impedance load. The common-emitter gives 30V output unloaded, however driving the IRF610's significant gate capacitance drops it to 20V pk-pk. I'm not sure if my transistor choice is anywhere near optimal - these are low-cost high-frequency switching transistors favoured by radio amateurs building QRP transmitters and HF amplifiers.

I'm not sure where to go from here - whether to try other transistors, add another voltage gain stage or perhaps a push-pull to drive the MOSFET?

I want to keep the power supply at no more than 50V if possible, to allow it to run off a +/- 25V bench PSU.

Sulaiman

Sat Dec 10 2011, 12:22PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

I've experimented a little lately with hf/ham amplifiers etc.
and I'm not sure which way you want to go,
1) The easiest is to buy a hf PA (new or eBay/used)
2) a d.i.y. hf PA
I'd start with an irf510 in class-a with an inductor as the drain-supply load, part of a pi section to match the load.
Don't attempt broadband transformers unless absolutely necessary.
By turning down the bias for the pa transistor you can try class-c or class-e.
I expect the critical part of the circuitry will be the impedance matching between pa transistor and load.

the ifr510 is a good transistor for 1 to 10 MHz,
ok with 12V supply, good with 24 V supply.
(24V,500mA,up to 5W into 40 Ohms. much more class-c)
'rf transistors' are better but their unnecessary bandwidth needs precautions against instability, up to destruction.

Steve Conner

Sat Dec 10 2011, 03:21PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

So, Sulaiman says to avoid broadband transformers, but the thread title says "broadband", which is it to be? If the amp could be tuned, that would make things easier, but if it has to accept any frequency between 1 and 10MHz without adjustment, it has to be broadband.

I'd start with the PA board borrowed from a 100W HF rig, as these can normally handle 1.8 to 30MHz, and pad the transducer with some resistors both series and parallel, to take the edge off the VSWR. Maybe 100 ohms across the amp output, and 25 in series with the sensor, then the SWR can never be worse than about 2:1.

Because we started out with more power than needed, hopefully we can get 30V across the sensor. If not, maybe you could add an inductor somewhere to resonate it and boost the voltage towards 10MHz.

If that gave promising results, I'd build another one optimised for 1 to 10MHz. Plastic packaged MOSFETs should work OK at 10MHz, especially the smaller ones like the IRF530 or whatever it was.

Pinky's Brain

Sat Dec 10 2011, 07:49PM

Registered Member #2901 Joined: Thu Jun 03 2010, 01:25PM
Location:
Posts: 837

I know it's a bit gauche ... but isn't 10 MHz in the range of what you could now do with opamps?

Electra

Sat Dec 10 2011, 11:15PM

Registered Member #816 Joined: Sun Jun 03 2007, 07:29PM
Location:
Posts: 156

Well I make the real part of power needed for the load at about 12.2W for the highest frequency of 10Mhz if Iâ€™ve done my math right. (30v pk into 40ohms in parallel with 150pf).
Rf transformers for 1 â€“ 10Mhz might not be too difficult for a diy design if you donâ€™t overdo the turns too much, the standard approach being 2 ferrite tubes side by side with the turns threaded through the centres. At least you could use the familiar push pull design then.

I would have thought it easier to apply some automatic level control via feedback from the output voltage, to reduce the drive or pa stage voltage. As opposed to trying to make it very linear over the whole frequency range.

I like the idea of swamping the output with some resistive load like you show in your diagram and as Steve.C has suggested to keep the output impedance from being too high.
Perhaps keep the cable to the transducer short as needed too.

Pinky's Brain

Sat Dec 10 2011, 11:37PM

Registered Member #2901 Joined: Thu Jun 03 2010, 01:25PM
Location:
Posts: 837

Why couldn't you just use a high bandwidth/slewrate opamp with say +/- 18V supply to drive a source follower N/P MOSFET totem pole?

Electra

Sun Dec 11 2011, 12:22AM

Registered Member #816 Joined: Sun Jun 03 2007, 07:29PM
Location:
Posts: 156

I think youâ€™d need an op-amp with a gain/bandwidth product a lot over 10mhz otherwise youâ€™d get some crossover distortion, driving mosfet gate capacitance directly from an op-amp might be challenge, but I suppose if the right op-amp could be found. I had forgotten about using p fets too, Certainly this would stand a good chance of working at 1Mhz at least, phase shifts may tend to affect the feedback of the op-amp as frequency increases, up until what point I donâ€™t really know though.

I think there are op-amps for transducer driving probably low power less bandwidth Iâ€™d imagine.

Carbon_Rod

Sun Dec 11 2011, 12:53AM

Registered Member #65 Joined: Thu Feb 09 2006, 06:43AM
Location:
Posts: 1155

Noninvasive ultrasound surgery system:

Note that most transducers tend to be quite inefficient outside their rated frequency.

However, Steve makes a very valid point given large RF power amplifiers at one time were often tested using calibrated transducer disc shunt loads. In some ways these meters were quite accurate at characterizing true analog broadcast strength.

Cheers,

Dr. Slack

Sun Dec 11 2011, 08:46AM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

Check out a Texas Instruments ths3095, or several in parallel. It's what I used when I wanted a DC-50MHz low distortion 50 ohm output amplifier for an ARB. It won't wuite get to your 30v pp output as it uses +/- 15v rails, but it will come close.

Pinky's Brain

Sun Dec 11 2011, 01:05PM

Registered Member #2901 Joined: Thu Jun 03 2010, 01:25PM
Location:
Posts: 837

Electra wrote ...
Certainly this would stand a good chance of working at 1Mhz at least, phase shifts may tend to affect the feedback of the op-amp as frequency increases, up until what point I donâ€™t really know though.

Yeah you're right, rise time + turn on time is 50 ns ... 180 degrees of phase at 10 MHz, oscillation ahoy.

I found this :

He mentions that he didn't want to use a radio transmitter because of impedance matching concerns (Steve's suggestion burns a lot of power). He used an opamp buffer and class A/B BJT voltage followers in about the frequency range sought after. Apart from the voltage range his low voltage part of the amplifier seems sufficient, with 36 volt opamps the output voltage swing might be shy of 30 v p-p (loaded their output voltage sags quite a bit, and the transistors represent a significant load).

PS. as for putting opamps in parallel, the problem is power ... around 20 Watt takes a lot of DIP-8s to get rid of.

1 2 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob