Trying to filter some unwanted noise over a DC voltage signal...

Thread Starter

Zurn

Joined Mar 4, 2019
117
Hi everyone,

I split this question into to parts: Part 1 is the problem, Part 2 is a solution that needs improvement.

Part 1
I'm trying to resolve an issue where noise from one part of my system is leaking into another part of my system via the shared DC voltage line.

The system is an oceanography device that records audio on four hydrophones (basically an underwater recording studio). Also attached is a separate instrument that samples various physical attributes in the water called a CTD. The CTD is what's producing the noise - every time it samples (twice a second) it produces a noticeable signal spike in the audio files, which can be viewed in the following spectogram:

noise-sample-spectogram.png

The two components are only attached via the DC (16V) and ground lines. I'm curious what an experienced technician (or, heaven forbid, an engineer) might try in this scenario.


Part 2
I've almost completely filtered the noise with the following RC circuit:

CTD-filter-drawing.drawio.png

However, there's an unwanted voltage drop at the instrument (it's reading 12V instead of 16V). I believe this is from impedance mismatching - any suggestions as to how to resolve this? I believe the impedance of the instrument is roughly 500 kOhms, however, I arrived at this by merely sticking a multimeter on its DC input/ground pins so please correct me if this the wrong way to go about making such a measurement.
 

Ian0

Joined Aug 7, 2020
10,035
If the drop is 4V over 20Ω then the current must be 200mA. If the instrument draws 200mA at 12V then its resistance must be 60Ω not 500kΩ.
Does the instrument still work with a 12V supply?
It looks as though the interference goes from 5kHz to 30kHz.
Have you tried some ceramic capacitors across the electrolytics? 100nF is the usual value!
 

crutschow

Joined Mar 14, 2008
34,689
You many have to add an inductor in series with the power supply output after the branch to the other electronics.
Try one with no more than a couple ohms resistance, and as large an inductance as physically reasonable for where it will be located.

To avoid possible power-on resonant overshoot from the inductor-capacitor combination, add a diode (e.g. 1N4002) in parallel with the inductor (cathode towards the positive side of the inductor).
That will limit the overshoot to less than 1V.

It's possible some of the noise is due to circulating currents in the ground connections.
If so you may have to isolate both the power and the instrument ground from the CTD power and ground using a common-mode choke.
On one project I worked on, I used such a choke to remove the IR sensor cryogenic cooler motor noise out of the video signal, when other types of filtering had failed.
 

Thread Starter

Zurn

Joined Mar 4, 2019
117
Aaaah yes...

If the drop is 4V over 20Ω then the current must be 200mA. If the instrument draws 200mA at 12V then its resistance must be 60Ω not 500kΩ.
Of course. How embarrassing.

Does the instrument still work with a 12V supply?
Yes it works as low as 8V. Unfortunately we use the instrument's voltage reading to monitor the battery elsewhere in the system, so it's important that it accurately reflects the system voltage. I realize I could just mathematically add the voltage drop back to the reading, which I may end up doing if I can't get rid of it...

Have you tried some ceramic capacitors across the electrolytics? 100nF is the usual value!
This seems to have further improved the noise filtering - thanks a lot for the tip.


You many have to add an inductor in series with the power supply output after the branch to the other electronics.
Try one with no more than a couple ohms resistance, and as large an inductance as physically reasonable for where it will be located.

To avoid possible power-on resonant overshoot from the inductor-capacitor combination, add a diode (e.g. 1N4002) in parallel with the inductor (cathode towards the positive side of the inductor).
That will limit the overshoot to less than 1V.

It's possible some of the noise is due to circulating currents in the ground connections.
If so you may have to isolate both the power and the instrument ground from the CTD power and ground using a common-mode choke.
On one project I worked on, I used such a choke to remove the IR sensor cryogenic cooler motor noise out of the video signal, when other types of filtering had failed.
I'll be sure to give these a try. I do have a couple of small inductors (47uH and 12uH) kicking around here. Though, I suppose I should ask: is the inductance to deal with the voltage drop or the noise? (I'll be sure to research this myself as well).

Greatly appreciate the replies!
 

crutschow

Joined Mar 14, 2008
34,689
is the inductance to deal with the voltage drop or the noise?
Both.
The inductor can have a lower resistance, along with a higher impedance to the noise.
Also the inductor gives a 2-pole low-pass rolloff of 12db/octave versus the resistor 1-pole rolloff of 6dB/octave.
 

MisterBill2

Joined Jan 23, 2018
18,986
The inductor filter scheme is certainly the next choice to try. I suggest, if there are any electronic hobbyists involved, seeking one like was used in AM car radios, because they were intended to stop exactly the same kind of noise, also in 12 volt circuits. Keep the capacitors, but replace the resistors with at least one inductor. The voltage drop should not be very much at all.
One amazing thing is that some of those inductors included a magnetized core, which provided a greater effective inductance for current in one direction than in the other direction..So the current direction might matter, if you find one of that variety.
 

MisterBill2

Joined Jan 23, 2018
18,986
If no other inductors are available, I suggest iron core inductors of at least 0.5 henry inductance and a resistance of less than the ten ohm resistors. In addition, add a diode across the battery connection, connected so that it is reverse biased and not conducting. That will allow the inductors to supply some current if the battery is momentarily overloaded, in a manner similar to switch-mode power supplies.
 

Ian0

Joined Aug 7, 2020
10,035
If no other inductors are available, I suggest iron core inductors of at least 0.5 henry inductance and a resistance of less than the ten ohm resistors. In addition, add a diode across the battery connection, connected so that it is reverse biased and not conducting. That will allow the inductors to supply some current if the battery is momentarily overloaded, in a manner similar to switch-mode power supplies.
An inductor that large on a gapped lamonated silicon-iron core would have a self-resonant frequency in the low kHz range . In other words, at the frequency of the interference, it will be a capacitor rather than an inductor, and will couple it in nicely.
 

MisterBill2

Joined Jan 23, 2018
18,986
NOT a "gapped" inductor, but an old-school iron core filter choke, which would not be saturated by that kind of current. And consider the capacitors following. 10 mfd and 100 mfd.
A higher inductance would be able to deliver power during a current spike as it's magnetic field dropped a bit. So it is not likely that any resonance would be a problem.
 

LadySpark

Joined Feb 7, 2024
188
Yes it works as low as 8V. Unfortunately we use the instrument's voltage reading to monitor the battery elsewhere in the system, so it's important that it accurately reflects the system voltage. I realize I could just mathematically add the voltage drop back to the reading, which I may end up doing if I can't get rid of it...
What is the 16V supply?
If its a linear regulator that has a 'remote sense' line then there is where the filtering should be happening.
 
Top