eric.c-side@worldnet.att.net (Eric Thosteson) wrote:
>
> I am trying to find/design a driver circuit for use in a high
> frequency underwater ultrasonic transducer....The piezos I want to
> drive operate at 5 MHz with an impedance around 5 ohms
> at that frequency.
5 Ohms seems pretty low. Most transducers I have seen had some __ty
Ohms, but a few had internal matching networks. BTW, the actual
resonance frequency may deviate +-10% from the nominal frequency
for most heads and 10% away from resonance you get completely
different impedance values.
The transducers must have a quite good efficiency. If you measure
their S11 with a network analyzer you get a much smoother curve
if you press a finger against the active surface or if you wet the
surface with a drop of water.
> I would like to hit the crystals with
> around 100 volts with a short pulse width, around 10
> microseconds.
That's ok, but 10 usec seems unneccessarily long. It will
probably be best to approximate a 5 MHz halfwave i.e. 100 ns.
> This poses an interesting problem,
> because the power requirements should be minimal.
> (100 volts across 5 ohms sounds large, but because of
> the small pulse width, the majority of the time, the
> circuit is idle.) The precise pulse width provided by the
> circuit is of the utmost importance.
>
> I have seen the problem solved before by generating
> a 5 MHz waveform at low voltage, then using a hand
> wound toroid to bump up the voltage. (Commercially
> available transformers with these demands seem
> to be nonexistant - but if anyone knows of any...)
Just take some Amidon cores and have someone wrap them.
The transformer itself is probably less of a problem than
the source that has to drive it. If you choose a voltage
step-up ratio os, say, 1:10, the impedance that has to be
driven will sink to 5/3.3 Ohms.
> I would like to find a more elegant method of doing this,
> as hand wound toroids at this frequency have very
> few wraps,
That's an advantage if you have to wrap them. :-)
> and bumping them totally changes their characteristics.
> If anyone can suggest an alternative, knows anyone
> who has attempted this, knows of a source for any
You may want to try an avalanche transistor. That's a
normal low-power transistor, operated under unusual
conditions. (see the text books)
250 V (Yes!) --------------------*--------
|
\ 100K
/
\
| ----------------------
*----------------------------
| ----------------------
50pF 100R C | some meters of
TTL-gate----||----\/\/----*----B GND coax cable
| E
/ |
100R \ |
/ |
\------*-------- out
|
/
\ 50R
/
|
GND
(first time i used netscape as a CAD tool.)
The low cost 2N3904 is reported to work well as an avalanche
transistor. In the quiescent state, the collector will be at
250 V and the coax line will be charged through the 100k resistor.
A rising edge at the ttl input will bring the transistor to
immediate breakdown. The charge contained in the cable will
flow to the output. You will see there a positive pulse with
150 V amplitude and a risetime of 1 ns. The pulse duration
will be 2 times the propagation time through the coax cable.
When the cable has lost it's charge, the transistor will switch
off again and the cable will recharge slowly through the 100K-R.
If you don't need a rectangular waveform you can substitute a
capacitor for the coax cable. If you make the capacitor too
large you may burn the transistor.( > 5nF is risky.) The
capacitor must have good rf properties.
There are way to cascade several avalanche transistors if you
want even more pulse power:
W. Pfeiffer: Erzeugung hoher Impulsspannungen mit
Avalanche-Transistoren in Kaskadenschaltung
Internationale Elektronische Rundschau, November 1973
It's written in german, sorry.
Mister Pfeiffer produced pulses of upto 1.6 KV into a 125 Ohm load
and upto 4.2 KV into 20 pF. Risetime was abt. 5ns.
The pulse repetition rate is limited, but this won't hurt in
your application.
Gerhard
Date: 7 Jul 1996 17:38:40 GMT
Original Subject: Re: Ultrasonic transmitter