I just found this group and noticed that there seem to be a lot of people that know much about amplifiers here. Thought I'd pester you with some question and see if you could give me some design hints.
I'd like to build a power amplifier that works up to 50 MHz and can deliver maybe 50 Watts. The input will be more or less a pure sinusoidal signal and no spikes or distorsions are allowed.
I've not very experienced in amplifiers so I'm not even sure where to start looking and what kind of techniques are available that might be interesting for me.
If anyone could please give me som clues or good places to find more information I'd be very grateful : )
Wow, that's fast input from you. thanks for your answers and opinions. If I'm not being very specific it's intentional : )
The application is in ultrasound but the frequencies it will be used can vary a bit. This is one of the reasons I wanted to post the question here, I know a bit about ultrasound but when it comes to radio I'd rather not say too much : ) Wouldn' it be correct to say that it is ultrasound as long as it's a longitudinal pressure wave above hearing range? 2 GHz sound pretty far up though : )
What I want to do is to amplify a signal from a standard waveform generator in the range of between 2 MHz up to as of now a maximum of 14 MHz. The 50 MHz maximum limit is probably a bit higher than I'll ever go but it would be nice to be able to go a bit further than 20 MHz just to be on the safe side.
Concering the load it's a bit trickier. I'll be using the amplifier with different piezomaterials and sometimes you drive at a low impedance, around 2 Ohms, and sometimes you're working more around 50 Ohms. We're trying to keep the piezos as close to 50 Ohms as we can but it's not always controllable. The voltages I guess would be between
10-50 V.
As you say, it might be that I should be looking for a RF-amplifier rather than searching for amplifier designs suitable for ultrasound ...
Hi J=F6rg, assuming that's the correct spelling ; )
2 Ohms is very low and that's maybe another problem that should be adressed separately. I think it would be ok to assume an impedance range of may 30-50 Ohms. I'm usually working with rather small transducers, around 600 =B5m square but probably not anywhere near the sizes you seem to have used : )
I guess that it wouldn't have to be linear. It's always easier but it really shouldn't matter. The only ultrasonic amplifiers I've seen are huge, expensive, unstable and maybe even more important very old. One of them were described as a "time capsule" by a guy that opened it and found 10 years of collected dust in it : )
Assuming that it doesn't have to be linear, what should I have a look at apart from fET switches?
Thanks for all the info! Actually I'm only interested in transmitting and not receiving anything at all. To make it even better I'm only running continous ultraound. It might sound a bit strange but it's quite nice : ) We're using the ultrasound to position cells and particles within fluids and are only using the transmitted wave to create a standing wave. Thankfully this makes the electronics a bit simpler as well!
Due to the low polarising voltage required on the transducers I'll have to use bipolar driving. The FETs I've looked at before had problems in that the either weren't fast enough or they simply couldn't deliver enough current.
Since it's time to stop working for today I'll have to wait till tomorrow to start looking for more FETs. Thank you very much for all the input, you've been very helpful!
The first thing I worked on in grad school was a 2 GHz acoustic microscope. The group down the hall was trying to do it at 100 GHz in superfluid helium.
Look for a "linear amplifier" for amateur radio work. Many of those would go from 160 meters to 10 meters, which is about 2 MHz to 30 MHz, in power levels up to a KW. You will want a pi matching section between the amplifier and your load, like the old Johnson Matchbox.
You need to specify a couple of other things before a sensible reply can exist. 50W, into _what_ load?. This changes massively the sort of voltages and currents involved. Also does the range go 'down' to DC?.
50MHz, if that's what you mean, is not "ultrasound". It's radio frequencies, and is in the low VHF, ie Very High Frequency range.
And you've not really specified what you want, because you say "works up to" but gave no indication of where else you expect it to work.
And, as is too often the case, you are asking for a solution you think you need, while revealing that you don't know much about that end solution. If you don't reveal the real purpose of this amplifier, the solution will be long in coming. Because it may turn out you don't need the amplifier, it may turn out that you've missed some detail and it's not 50MHz but 50KHz (a 1000:1 error), it may turn out some other solution will come into play based on the end game. Or, if this is really a DC to 50MHz amplifier you really need, it may turn out the solution is a collection of amplifiers, covering portions of that range.
50MHz can be ultrasound. The highest frequency in ultrasound I designed stuff for so far was 40MHz but the trend is upwards, for more resolution. 20-30MHz is nowadays kind of a 'normal' ultrasound frequency. The days that it was all under 7.5MHz are long gone.
2ohms would be rather low unless you are working with huge transducer disks. For amplifier design you need to narrow that range down somewhat. Once you have picked a certain transducer design and have it made at the usual places like TRS or Parallel Design they won't stray that much.
+/-40% maybe, at the most. Unless you do what we did with elements that are barely wider than the crystal sizes ;-)
Does the amp really have to be linear? I have never driven any PZT with a linear amp in about 20 years except for R&D testing. If it doesn't have to be linear it'll be much easier, FET switches etc. In case it must be linear be prepared for a larger amp and lots of heat dissipation. You can do PWM and stuff like that if it has to be small but the engineering efforts for that only pay off if it all goes into volume production or where cost doesn't matter.
ENI makes nice RF power amps. 100W is not big deal but they are huge and heavy. At that power level it would be about the size of carry-on luggage. You could rent one to try it out. These are normally used for EMI susceptibility measurements.
Yes! Nice to see that umlaut back in there once in a while.
Oh, some of ours were a lot smaller. Like an array of 64 elements in a cirular arrangement and the whole thing being under 1.5mm in diameter.
That sounds like old NDT stuff or amps designed for scientific projects.
You'll have to make a decision between unipolar or bipolar pulsers. Unipolar only requires one FET per channel, inductors, some diodes. Bipolar requires push-pull and that usually comes with a pulse transformer. In your case probably something on a #43 ferrite toroid core. The driver is often a challenge because FETs need a good 10V and lots of pulse current into the gates to be snappy enough, no matter what the ads say.
No idea what you want to do after the pulse. But if you have to detect weak echoes or even Doppler make sure the supplies for the pulsers and their drive logic are clean like a whistle. There is an old saying that truly applies to ultrasound: What you shout into the forest will come back to your ears. IOW, any noise that's on the TX pulse is tough to get rid of afterwards.
The drive logic should be free of jitter which usually means that the uC, FPGA or whatever is triggering the pulse should not be shouldering any other realtime tasks that could cause its substrate and busses to become modulated.
Then there is the T/R switch but that shouldn't be rocket science. It just has to be squeaky clean as well on the supply rail. The receive amp must be able to cope with a huge overload peak without doing any weird stuff such as ringing.
Well, that about sums it up. Sounds easy but it isn't, at least not when your Doppler has to be 2-3dB more sensitive than the competition's ;-)
Check IRF. They have one of the largest selections for that purpose. You can always parallel some if they are on the wimpy side. I have even used the rather tiny BSS123 and BSS84 in some smaller applications. Old as Methusaleh but hard to beat at five Cents a pop (at our quantities...).
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.