I want to make a cockpit voice recorder for small GA planes. Typically the headphones of such aircraft are stereo, e.g. left is coming from the radio (ATC), and right is the in-cabin intercom. I want to record both channels. My device will be powered from the aircraft's 12- or 24- volt power, so it will share grounds with the aircraft's systems. Being a big/unknown system, I'm worried whether the headphones' return signal is also the same GND. It's tempting to assume it is and connect it to my circuit's GND, but if it's not, then I'll be creating a short. Even if it's the same, I'll be creating a ground loop.

I'm not sure what to do about it. The best I can think of is to introduce an isolation transformer for the audio, like that:


simulate this circuit – Schematic created using CircuitLab

Is this enough to ensure the audio will come in clean and undistorted, and the ground loops/shorts avoided?

The other solution is to use an isolated DC-DC converter.

Realistically, of course, the circuit will be more complicated than this, as I won't be replacing the headphones; I'll be just tapping into their signal (a T-shaped joint). I'm not sure how the audio system in the plane will feel about driving two circuits instead of just one, but I guess it would be fine.


Of course, I know adding this kind of modifications to a crucial aircraft system is very dangerous, and we'll of course do thorough ground tests before flying with that. Part of the reason to ask the question is the fact there's no room for error, and I want to design the system properly upfront.

  • \$\begingroup\$ Since these audio signals are a vital part of the plane's equipment, I would be very careful in interconnecting to them. I would definitely use isolation transformers and an isolated DC-DC converter so that your equipment cannot interfere with the plane's ground system. Even then, I would investigate the rules on connecting to vital plane equipment. Doesn't the plane have a cockpit voice recorder? \$\endgroup\$
    – Barry
    Feb 17, 2021 at 0:17
  • \$\begingroup\$ Yeah, I'll be very careful, too. We'll do lengthy land-based trials before flying. Surprisingly small GA planes do not have cockpit voice recorders. \$\endgroup\$
    – anrieff
    Feb 17, 2021 at 8:29
  • \$\begingroup\$ Doing trials doesn't make regulations violations more legal \$\endgroup\$ Feb 19, 2021 at 0:22
  • \$\begingroup\$ If there are any violations :) which part you think is more problematic, the power-supply side, or the audio side? Because on the audio side, the headphones you connect can be classified as similarly dangerous (some of them even have batteries, for noise cancelling). So we'll look if they have some sort of certification, and seek to be certified similarly. \$\endgroup\$
    – anrieff
    Feb 19, 2021 at 0:34

3 Answers 3


If you want to stay absolutely safe, your interface between the recording device and the aircraft's electrical system must ensure that the connection of your device does not alter the normal operation of the avionic system.
The problem can be broken down into these points:

  1. Electromagnetic Compatibility
  2. Galvanic isolation
  3. Minimum impedance for headphone amplifier
  4. Power consumption from aircraft DC bus.

The last point is trivial and I assume that the power consumed by the entire recorder is really low and that the aircraft's DC bus can deliver it without any problems.

The first point (EMC), it is perhaps the most critical part also because if a device is not EMC compliant it cannot be used and sold (here a summary of the legals proceedings about non-compliant device caught on the market). But apart from that, for a safety issue, you must be sure that the noise emitted by the recorder does not bother the aircraft and that the noise emitted by the aircraft does not bother the recorder (which is the sense of EMC).

Taking into account points 1 and 2, the architecture I would use is the one in the schematic below.

  • Galvanic isolation is provided by mean of isolate DC/DC converter and audio transformers. In this way the aicraft ground and the device ground are completely separated without any ground loops. Ground separations also helps to reduce the common mode noise.

  • Electromagnetic compatibility for conducted emissions is provided by the EMI filter in combination with the choice of a DC converter with a convenient switching frequency and his appropriate input filter. You want also Electromagnetic Compatibility for and from radiated emissions, but this will be provided by appropriate shielding of the device.

  • For point 3, you can simply choose an audio transformer with an input impedance comparable to headphone common impedance, for example 16 or 32 Ohm. In this way you are sure to avoid the headphone amplifier distorsion.
    EDIT: if you want to capture the audio signal "in parallel" to the headphones, the impedance of the audio transformer must be high enough to not lower the impedance seen by the amplifier below the minimum impedance that can be driven without excessive distorsion (often 2 or 4 ohm).


simulate this circuit – Schematic created using CircuitLab

  • 1
    \$\begingroup\$ Hmm, I really like your answer, but I wonder why you (and the others) insist on double galvanic isolation (both on the power-supply side, and the audio-side). Particularly on the power-supply side, if I share grounds with the aircraft systems, I'd be able to measure DC rail voltage, which can be useful to detect flight/non-flight conditions. It's perfectly possible to make this work even with galvanic isolation (put a small MCU, measure, transmit digitally over an optocoupler), but it is a lot more work, so I'm wondering if it is well justified. \$\endgroup\$
    – anrieff
    Feb 25, 2021 at 9:41
  • \$\begingroup\$ Well, galvanic isolation is not strictly necessary, but in general it prevents the onset of possible noise related and safety related problems. So in safety critical applications it's ever a good choice. But in this case there is a more specific reason: You don't know if the hdph amplifier gnd is the same of DC bus gnd. If yes, the audio transformers will works with the same gnd reference on the primary and secondary winding, and this makes them useless and does not allow them to break gnd loops. Maybe you could take advantage of some features of the DC/DC converter to detect flight condition \$\endgroup\$
    – LDEG
    Feb 25, 2021 at 10:15

Since most headphone outputs need to drive headphones of various impedance, they often have a quite low output impedance. 4 Ohm or less is typical for output impedance on a phones jack. This should allow you to add any high input impedance audio amplifier into the circuit in parallel with you headphones without any worries about overloading the output. Having said that, I'm a bit doubtful that a cockpit intercom follows any consumer audio conventions. I would recommend having a look at the signals with an oscilloscope before designing any circuits. Isolation is going to be important, and you'll also want to scale the signal to match your analog conversion input. depending on your flavor of ADC, you might need some signal amplification in order to get the maximum dynamic range out of it. If the phones output follows typical Line-Out conventions, the signal will typically be below 1.2V P-P at it's loudest (your pain threshold). There is a guy that already makes the device you probably want. Here is a link - https://www.crazedpilot.com/audio-recording-cable-for-your-iphone/ This device has a 3.5mm TRRS output plug. I'm not sure about any isolation that they have implemented (or not) since they likely assume you will be plugged into a battery operated device that is fully isolated. Might be worth a look though.


Your circuit definitely has the potential for ground loop.

I had the exact same problem with a car Bluetooth receiver that was connected between the 12VDC power of the car and the AUX of the audio unit.

This resulted in a horrible noisy output. To fix the issue, I simply added an isolated DC/DC converter for the power supply.

In your circuit, despite the isolation transformer, both sides are still connected to the ground, thus creating a ground loop between the aux and the power input.

You have a very high chance to get noise in your headphones with a ground loop, but it can depend on how the cockpit electronics are implemented so the effect might be from unnoticeable to major.

The best solution in this situation is to isolate the power input as you can keep the audio input. There are a lot of cheap DC/DC converters that are isolated and for 2-3$ you do not need to worry about the rest, so then you only have a ground point which is the audio input.

If your cockpit power output is 12VDC, you can simple use an isolated 12V to 12V DC/DC without changing anything else. Make sure the input and output ground of the DC/DC are not connected.

You may also want to add quite a few ceramic caps and perhaps an RLC filter at the output of the DC/DC converter to reduce the risk of noise from the DC/DC converter.

  • \$\begingroup\$ "In your circuit, despite the isolation transformer, both sides are still connected to the ground" — this assumes the headphones' return ("Common") is the same GND as everywhere else, and it may not be. But your experience with the Bluetooth receiver is exactly what I hope to avoid. Do you, by any chance, have it now? Can you check if it has audio isolation transformers? \$\endgroup\$
    – anrieff
    Feb 25, 2021 at 9:49
  • \$\begingroup\$ Yes exactly, it really depends on a lot of unknown factors related to the implementation of your cockpit. The audio jack might already be isolated, the supply as well. It's a risk to take. That is why my answer proposes a cheap and simple way to ensure you won't have noise despite not knowing the cockpit implementation. \$\endgroup\$
    – Damien
    Feb 25, 2021 at 10:45
  • \$\begingroup\$ @anrieff I don't have it anymore, but basically my Bluetooth device was USB powered, I just made a small USB -> USB device (power only) with an isolate 5VDC -> 5VDC DC converter and plugged it between my 12VDC to USB plug and the Bluetooth device. I used a simple veroboard and encapsulated it on some silicon. \$\endgroup\$
    – Damien
    Feb 25, 2021 at 10:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.