Ripple due to SD-Card-Writes in hydrophone amplifier

Hello,

The background of this question:
as a part of my diploma thesis i did analyse an existing hydrophone amplifier (see attached file). It came up that this amplifier was improper designed. The amplifier gain is highly frequency depended due to the capacitive nature of an piezoelectric hydrophon or in other words the way to low input impedance...As a result i did design a chargeamp and iam happy with it

But as a "Bonus" i want to deliver an other explanation for an other obvious problem with the "old" "faulty" amplifier. I want to point out why there was need to design a new amplifier, beside the prime reason of an greatly wrong freq-response.

I want to explain why writes to an SD-CARD shows an ripple at the output signal of the amplifier chain (see View attachment FaultyHydroAmp.pdf). Iam not sure what could be the reason...

I think that writing to the SD-CARD will force the supply voltage to a very little ripple, due to filling the charge pumps inside the sd and improper designed power supplies ...
Does the reference Source (see View attachment FaultyHydroAmp.pdf) transform this supposed ripple to its output (and the virtual ground of the first two opamps)? But the gain is only 1...!? The CMMR should be high enough...

Can someone please point me in the right direction what could be the reason for the ripple in the output signal of this amplifier due to sd-card writing?

Thank you very much.

The CMRR/PSRR is usually high only at low frequency. If the amplifiers are powered by the same supply line as the SD card, the supply ripple will feed-through the opamps - that's why good decoupling/filtering is needed.

Thank you for your answer. My main Problem is, that i want to know the "entry" point of the ripple so that i could calculate its gain and take then the CMRR/PSRR arguments. E.g. if i you forget about the hydrophone (and its equivalent circuit) its a simple voltage amplifier with three stages 85*85*12.5 => 100 dB gain (but is the ripple at this point in the 1st stage?). OR is the entry point a change in offsetvoltage, which gain is only 1...

I think this is the main reason but fail to put the things together...

You are able to measure the supply rejection versus frequency with a test generator and also precisely detect where it's broken. You don't need to guess about it.

Yes, but i thought anybody with more experience could say : "hey it's obvious because the ripple enters x and has gain y"...But it's no problem, i would have been just a bonus. thank you all.

In most cases the ripple enters from the first stage.
Given the gain of 100dB you need only 3uV of equivalent noise at the input --> very likely it's a supply rejection problem.

yes, thats near my initial thoughts. I think its enough to tell that at the point before the 1.2k R at the neg Input a 3uV difference to virtual ground will end ob in ripple/noise as seen. This high gain and the poor power supply will lead the the sd-problem.

As obvious point, a system with a high gain like the present one has many possible interference pathes. Single supply is always critical, continuous power supply rails without extra filtering for sensitive stages, too. It may be even a ground loop problem. That's why you need to measure it.

In addition, I don't exactly understand the sensor interface. The sensor schematic shows a ground referenced sensor, the preamp suggests a floating sensor.

Sorry for not beeing precise. The Sensor "Gnd" is connected to the output of the reference source (1.25V) which creates the bias voltage for this single suppy system. Both Gnd are not the same, sorry.

That has been already my guess. But you would want to have a grounded case and cable screen in addition. Also the purpose of the 1k2 series resistor in the sensor ground line isn't plausible to me. It reduces the first stage's gain and makes the sensors "cold" terminal succeptible to interferences.

The advantage of the virtual ground circuit is, that reference voltage "noise" is attenuated related to the sensor voltage by about 3500 according to the schematic. So PSRR of the first amplifier stage will be the most likely problem.

yes 1k2 is another fault in this design, which disappeared in the new charge amp. thank you.