Sonar : Problem with a resonant circuitry

[Simon]

Hi,
I'm working on a sonar and when I transmit a signal in water, it goes
through a transformer: two MOSFET are switching on the primary with a
TAP on a DC Voltage and the transducer is on the secondary.

The Problem is, when I finish transmitting (the MOSFET are no longer
switching) I cannot listen the reflected signal on an object for a
moment because the transducer is resonating with the transformer. The
idea would probably to short this resonance to ground but it has to be
fast enough (in µ-second). The signal on the secondary when
transmitting is +/- 400V for ~30 µ-second.

Any ideas for this situation? What kind of circuitry to use?

Thanks

Simon

 

[Joerg]

Hello Simon,

I'm working on a sonar and when I transmit a signal in water, it goes
through a transformer: two MOSFET are switching on the primary with a
TAP on a DC Voltage and the transducer is on the secondary.

The Problem is, when I finish transmitting (the MOSFET are no longer
switching) I cannot listen the reflected signal on an object for a
moment because the transducer is resonating with the transformer. The
idea would probably to short this resonance to ground but it has to be
fast enough (in µ-second). The signal on the secondary when
transmitting is +/- 400V for ~30 µ-second.

Any ideas for this situation? What kind of circuitry to use?

Ok, I am from the medical side of ultrasound but we have the same
problem, the need to see tissue that's right under the scanhead. What
you can try is to arrange your TGC amp so it loads down the transducer
heavily, at least for the first few usec (after the T/R switch).

However, the real issue is the backing material of the transducer.
That's a science in itself and the recipes are closely guarded. You
could talk with the folks that process PZT at companies such as Parallel
Design or TRS. Or with experts at universities like Penn State.

Regards, Joerg

 

[Phil Hobbs]

However, the real issue is the backing material of the transducer.
That's a science in itself and the recipes are closely guarded. You
could talk with the folks that process PZT at companies such as Parallel
Design or TRS. Or with experts at universities like Penn State.

At Stanford in the 80s they used epoxy loaded with tungsten powder for a
nice absorber that impedance-matched to zinc oxide.

Cheers,

Phil Hobbs

 

[qrk]

Hi,
I'm working on a sonar and when I transmit a signal in water, it goes
through a transformer: two MOSFET are switching on the primary with a
TAP on a DC Voltage and the transducer is on the secondary.

The Problem is, when I finish transmitting (the MOSFET are no longer
switching) I cannot listen the reflected signal on an object for a
moment because the transducer is resonating with the transformer. The
idea would probably to short this resonance to ground but it has to be
fast enough (in µ-second). The signal on the secondary when
transmitting is +/- 400V for ~30 µ-second.

Any ideas for this situation? What kind of circuitry to use?

Thanks

Simon

You might be looking at the natural ringing of the transducer. Your
transducer may not be appropriate for your task if it isn't damped
enough for your application. What sort of minimum range do you need
and at what frequency? If the transformer is resonating with the xdcr,
your transformer may be improperly designed (resonance too close to
the operating freq). If shorting the primary is helpful in reducing
the ringing, try using a H-bridge instead of a push-pull or
half-bridge. In a H-bridge, you can leave either the top or bottom
switches (FET's) on during the receive period which effectively shorts
the transformer primary.

 

[Joerg]

Hello Phil,

At Stanford in the 80s they used epoxy loaded with tungsten powder for a
nice absorber that impedance-matched to zinc oxide.

Yep, that's a start. There is also a lot of black magic to processing
that, for example making sure there aren't the tiniest pockets.

Regards, Joerg

 

[Joerg]

Hello Mark,

... If shorting the primary is helpful in reducing
the ringing, try using a H-bridge instead of a push-pull or
half-bridge. In a H-bridge, you can leave either the top or bottom
switches (FET's) on during the receive period which effectively shorts
the transformer primary.

Shorting isn't always going to cut it. The energy from the ringdown
needs some place to dissipate, ideally a resistance. It's like the shock
breakers in a car which transfer vertical ringing of the suspension into
heat after driving through a pothole. If you'd 'short' them all the
stuff the trunk and the passengers would go flying about.

Regards, Joerg

 

[[email protected]]

Hello Simon,



Ok, I am from the medical side of ultrasound but we have the same
problem, the need to see tissue that's right under the scanhead. What
you can try is to arrange your TGC amp so it loads down the transducer
heavily, at least for the first few usec (after the T/R switch).

However, the real issue is the backing material of the transducer.
That's a science in itself and the recipes are closely guarded. You
could talk with the folks that process PZT at companies such as Parallel
Design or TRS. Or with experts at universities like Penn State.

At EMI we just used resin-bonded tungsten powder - mostly bonding it
with regular epoxy resin. You'd want to de-air it (stick the mixture in
a vacuum desicator (or some other chamber tht you can see into and pump
down) and evacuate it until the mix starts frothing, the evacuate more
slowly so that the froth doesn't overflow your container before it
breaks.

 

[[email protected]]

Hello Phil,


Yep, that's a start. There is also a lot of black magic to processing
that, for example making sure there aren't the tiniest pockets.

Another trick we used at EMI was to cast a quarter-wavellength-thick
layer of resin-bonded tungsten powder and glue that on the front of the
transducer as an impedance matching device. Apparently that helped too.

 

[Joerg]

Hello Bill,

Another trick we used at EMI was to cast a quarter-wavellength-thick
layer of resin-bonded tungsten powder and glue that on the front of the
transducer as an impedance matching device. Apparently that helped too.

Yep, that's another trick. But can cause trouble in very wideband PZT-5H
apps. We used to call that acoustic lens, it doesn't have to be a
constant thickness layer.

Regards, Joerg

 

[[email protected]]

Hello Bill,


Yep, that's another trick. But can cause trouble in very wideband PZT-5H
apps. We used to call that acoustic lens,

You shouldn't have, it is strictly an impedance matching device,
equivalent to the anti-reflection layer on the face of an optical lens.

It doesn't have to be a constant thickness layer.

It does if you want to improve the coupling between the transducer and
the fluid in which you are doing your imaging. An acoustic lens doies
something rather different.

 

[Mark]

Simon,

the fastest way to stop the transducer ringing is to drive it with
another signal that is 180 deg out of phase with the original
signal...for a very short time....just until it stops ringing...

Mark

 

[Joerg]

Hello Bill,

You shouldn't have, it is strictly an impedance matching device,
equivalent to the anti-reflection layer on the face of an optical lens.

Reason is that it isn't always a flat surface but.

It does if you want to improve the coupling between the transducer and
the fluid in which you are doing your imaging. An acoustic lens doies
something rather different.

Yes, but there is always the trade-off between resolution, focus and
matching. In the near field matching isn't too important (provided you
have a good receiver circuit) but focus is. Imaging in our case is
mostly through tissue with very little in fluids. Except to some extent
in coronary imaging and Ob-Gyn during pregnancy.

Regards, Joerg

 

[[email protected]]

Simon,

the fastest way to stop the transducer ringing is to drive it with
another signal that is 180 deg out of phase with the original
signal...for a very short time....just until it stops ringing...

Interesting theory. Have you evr tried to reduce it to practice?

 

[Phil Hobbs]

Interesting theory. Have you evr tried to reduce it to practice?

*Any* change in the drive will translate into a change in the motion via
convolution with the impulse response of the transducer. You can
conceivably drive the transducer with an antiphase signal to zero out
the terminal voltage, but not the actual acoustic wave.

There's the acoustic transit time, at a minimum.

In general it's probably better to subtract the ringing out in
software--but that requires really good control, i.e. keeping the
transmit pulse exactly the same each time--same phase, same length, same
transients.

Cheers,

Phil Hobbs

 

[Joerg]

Hello Phil,

*Any* change in the drive will translate into a change in the motion via
convolution with the impulse response of the transducer. You can
conceivably drive the transducer with an antiphase signal to zero out
the terminal voltage, but not the actual acoustic wave.

The best result I achieved was with some nice resistive damping right
before the first receive amp which is then gradually eased off towards
the long range echoes.

Look at it this way: You can't really control the oscillations of a car
suspension after hitting a pothole via slamming an 'inverse pothole'
into it. To some extent you can regulate like it was done on the Citroen
DS21 (the big one) but at the end of the day the shock absorber, or a
simulated one, does the lion's share. Ringdown is energy and the only
way to get rid of some of it is dissipation.

Also, transducers don't like to be slammed into reverse, they have a
limited bandwidth. For stuff such as PZT-5H that's usually around 30-40%
at -6dB, not matter what the glossy marketing materials say. And it
varies a bit from one transducer element to the next. PVDF is a lot
wider but that doesn't have much efficiency to write home about.

There's the acoustic transit time, at a minimum.

In general it's probably better to subtract the ringing out in
software--but that requires really good control, i.e. keeping the
transmit pulse exactly the same each time--same phase, same length, same
transients.

Well, to an ultrasound guy keeping pulse pattern uniformity should come
as natural as balancing a wheel comes to a mechanic. If it doesn't, it's
best to get help.

Regards, Joerg