Hydrophone Pre-Amp Comparison

[waymond91]

Hello All!
Ive been working on a hydrophone design, working off a reference design by a man named Jules.

The piezo cylinder we are using can be found here: https://www.steminc.com/PZT/en/piezo-ceramic-cylinder-26x22x13mm-43-khz

The design/build is posted here: https://www.instructables.com/Hydrophone-Drop-Rigs/

However here is the schematic:




Jules is using a OPA1642 as buffer circuit and is passing the signal straight into a high-end field recorder, however I can verify that connecting the aux cable to a laptop or phone aux in still works just fine.

The design has a few challenges, in the past hydrophones have been lost if they are "knocked" (agaisnt, say, the side of a boat) and an overvoltage seems to find its way into the OPA1642 inputs.

Also, there have been issues grounding the circuit in the water to remove "hum" from the signal (presumably, this 50/60hz noise induced from power lines).

I thought if I re-worked the design with an instrumentation amplifier I could really elimnate any common mode noise, perhaps amplify the signal even better, and provide some input over voltage protection:



However, my design really works quite terribly. The input and output signal all seems within spec, however there is a terrible hiss/noise on the signal, and I had to use nearly the INA129s max amplification to hear anything.

Any ideas as to why this did not work as planned?
I thought this was going to be rather slick

[DaJMasta]

While I don't have a lot of insight to offer, how are you testing them?  I've done some hydrophone recording in a home aquarium and it being an electrically isolated box means ground loops can be an issue if any equipment in the signal chain is grounded.  There's also no filtering of any kind - maybe take a look at the spectral plot of what you're recording - if a lot of the hiss is high frequency, you may be able to cut it out with some low pass filtering in your buffer amp.

Otherwise, while I don't know how suitable it really is, what about simple TVS diodes (or similar) for protection from piezo spikes?  If you select a breakdown voltage that is still safe for the amp, it should be able to suppress those big transients and the capacitance of the diode shouldn't be big enough to mess with your sensitivity bandwidth.  An option used in audio test equipment is also a gas discharge tube - they breakdown at a certain voltage and are just an open circuit otherwise.

[Roehrenonkel]

Hi waymond91,
 
you had two channels single-ended in the first design.
How did you now connect one hydrophone (symetric output?)?

Good luck

[waymond91]

Thanks for the feedback!
Yeah, the original design was two channel single ended. But I only ever deploy one hydrophone at a time. So the instrumentation amp design is a differential input from one piezo cylinder.

I kinda thought switching to a differential input was really going to make this thing sing, but it didnt.

As for measurements - I have only used my scope a bit but I havent captured anything.
For the most part I have just been plugging it into my computer and phone and recording audio.
During this process, the none of the devices are plugged into mains power as I have a separate power supply that runs off a battery and my computer was unplugged.

I will try to post some more "official" measurements later this week.

I love the TVS diode idea if we can make it work!

[moffy]

The piezo ring has a capacitance of 6.6 nf which into 2 x 47k gives a high pass filter response of around 255Hz, is that suitable for what you want? Also since the output is capacitive if you use shielded cable the cable capacitance will absorb part of the output signal acting as an attenuator. Unshielded twisted pair is possibly a better choice.

[Marco]

If you want to use a single IC, use a JFET instrumentation amplifier like the AD8220 or INA121. INA129 has too much current noise.

Don't use those coupling capacitors on the input, replace the 47k with 1 Meg, put a couple k series resistance on the inamp inputs for protection (probably nothing more needed).

[waymond91]

OK, we definitely dont want to be blocking out 250Hz and below.
What is the harm in having capacitors in the input?

I want to avoid adding resistors in series with the instrumentation amplifier if possilbe - Im under the impression that these really high-impedance in-amps have their input resistances carefully trimmed and I dont want mess with that if its not required...

I updated my schematic with the 1M reference resistors, removed the input capacitors, and swapped from an INA129 to a INA121.

What do you guys think? What are the trade-offs of going single sided vs fully differential here?



The datasheet for the ina121/ina129 says that the device has over-voltage protection to +/- 40V, but Im not finding detailed information about it:

[moffy]

Just one small point, the Rg pins are fairly sensitive and a good noise pickup point for the intrumentation amp. If switching values if you can keep the resistors as close to the pins as possible and use say on board jumpers. The capacitors are kind of unnecessary as the sensor looks like a capacitor, they will compromise CMMR very slightly, but if there is a chance of a DC voltage being injected somehow they make sense.

[waymond91]

This is very helpful, yeah the gain resistors are kind of far from the IC, here is a picture:



For my senior project in university I had to design an EMG amplifier and I am still tramautized by the DC drift we had at those inputs XD

Its really not clear on how likely a dc voltage might be injected, but I guess I can leave the footprints in just in case

[Marco]

What is the harm in having capacitors in the input?

They are not needed and I find it inelegant to leave nodes in your circuit floating, leakage current will probably get the DC level of the hydrophone somewhere safe for the capacitors but it makes diagnosing failures more difficult than it needs to be.

[EEVblog]

I worked in the underwater seismic industry. We always used charge amplifiers for piezo ceramic hydrophones.
Sure you want to use a voltage amplifier?

[moffy]

This is very helpful, yeah the gain resistors are kind of far from the IC, here is a picture:

If you use a (2 row by 3) header pin, you could get the spacing a bit tighter but it doesn't look bad.

[sharow]

My wild guess. The cause of the noise is that
1) the output should be referred to reference.
2) TLE2425 is noisy. 1200nV/sqrt(Hz), or 0.1mVrms(10 kHzBW)

" hiss/noise" you hearing is came from TLE2425. and CMR dosen't work because output is referred to GND.

[moffy]

My wild guess. The cause of the noise is that
1) the output should be referred to reference.
2) TLE2425 is noisy. 1200nV/sqrt(Hz), or 0.1mVrms(10 kHzBW)

" hiss/noise" you hearing is came from TLE2425. and CMR dosen't work because output is referred to GND.

Good point, you would be better off using a TL431 as a 2.5V reference, it's about 120nV/sqrt(Hz) or a resistive divider with a cap and a moderately low noise opamp buffer.