1
\$\begingroup\$

I am using a linear regulator (MIC29302WU) to power a computer mounted inside my car. There is significant noise being passed along the ground of the regulator to the main computer. I verified this by connecting an inverter to the battery and using a 12 V power supply which eliminated the noise.

Can anyone suggest an isolation circuit? I would like to consider cost and board area. Would this require a flyback design? Or can I get away with not using a transformer?

\$\endgroup\$
7
  • 1
    \$\begingroup\$ You could use a DC-DC boost regulator instead. 14 to 19V instead of a square wave AC inverter \$\endgroup\$
    – Tony Stewart EE75
    Jun 3, 2019 at 13:51
  • 1
    \$\begingroup\$ I tried a SMPS after the linear reg, but no reduction in noise. Probably because input/output shares a ground. Are you talking about isolated boost circuit? Do you have a schematic you can link to? \$\endgroup\$
    – physiii
    Jun 3, 2019 at 13:57
  • \$\begingroup\$ The combo of DC to AC ( square wave) and AC to DC charger has too much common mode noise at line freq + harmonics so it interferes with audio quality to line input on car stereo. My suggestion is avoid the line freq (square wave) and go DC-DC non-isolated 50kHz to laptop for charging battery. Possibly 50~65W. This Could be a purchase item online \$\endgroup\$
    – Tony Stewart EE75
    Jun 3, 2019 at 14:07
  • 1
    \$\begingroup\$ Okay I think there is a miscommunication. I am currently using a non-isolated DC-DC regulator on my circuit (MIC29302WU)...This is for a stereo that goes into the dash a car. However using the linear reg I get noise because ground planes are not isolated, I can stop the noise by using an inverter, to psu like I said but obviously that's not what I'm actually implementing, I was just telling you how I stopped the noise by using isolation. Now I need to figure out how to implement the isolation on my board \$\endgroup\$
    – physiii
    Jun 3, 2019 at 14:19
  • 1
    \$\begingroup\$ You could define all DC V,I ranges and AC noise characteristics better in your question with measurements and design so one can know you analyzed the EMI correctly between radiated, conducted, CM and DM noise and if speaker drivers are differential or single ended \$\endgroup\$
    – Tony Stewart EE75
    Jun 3, 2019 at 14:45

2 Answers 2

Reset to default
0
\$\begingroup\$

Get a 12V DC DC converter with isolation. Make sure the input voltage can handle 12V and the output voltage is a 12V output.

There are some that can be found at digikey

With some searching some can be found with terminals.

Before you try that, try a clamp-on ferrite on the wires to see if you can increase the inductance and stop high frequency noise.

\$\endgroup\$
4
  • \$\begingroup\$ The cheapest 3A isolated DC DC converter from your digikey link cost $15 (digikey.com/product-detail/en/tamura/BPM1234SJ/MT7391-ND/…) That seems like a costly solution, I would think there is a cheaper way to get galvanic isolation with some chip/transformer combo. \$\endgroup\$
    – physiii
    Jun 3, 2019 at 18:57
  • \$\begingroup\$ Cost is relative. It was a suggestion, you are more then welcome to do your own shopping to find what you need. There are cheaper isolated DC to DC's of chinese make. Try the ferrite first. Transformers don't work with DC, they need a chopping circuit. \$\endgroup\$
    – Voltage Spike ♦
    Jun 3, 2019 at 19:05
  • \$\begingroup\$ Will try ferrite, thank you. And yes ofcourse you need an alternating b-field to induce current in the secondary of the transformer, this is why I said chip/transformer pair where the chip controls the switching current on the primary. \$\endgroup\$
    – physiii
    Jun 3, 2019 at 22:19
  • \$\begingroup\$ You could always roll your own DC to DC, linear makes some good controllers. Just for the PCB you'd be spending 5 to 7$, the controller will be at least 2$. \$\endgroup\$
    – Voltage Spike ♦
    Jun 3, 2019 at 22:31
0
\$\begingroup\$

Assuming this is what you got...

enter image description here

The problem here is the ground loop. The voltage reference for your signal is the "0V" point at the source (in this case the computer playing the music). That is, the computer outputs an audio signal voltage referenced to its own ground.

The stereo is also connected to chassis ground, and it inputs a signal voltage referenced to its own ground.

Since these two grounds are not the same, due to current flowing in the chassis, any voltage between the chassis grounds of the source computer and the stereo is added to the signal.

In addition, the current drawn by both devices is not constant, so even if nothing else in the car was drawing current, their ground potentials would still be different due to their own supply currents flowing in ground.

There are several solutions:

  • Isolate the computer from the chassis ground with an isolated supply.

This can work, but the isolated switching supply will add noise of its own, mostly common mode.

  • Isolate the signal

You can either use an audio transformer (easy to find, usually sold as "ground loop breaker", pretty cheap) or add a differential receiver. In both cases, the goal is to read the computer's output signal, referenced to its own ground, and output a voltage that is referenced to the stereo's ground, thus removing the noise between both grounds.

The easiest solution would be to use a SPDIF TOSLINK digital audio connection, which is isolated due to being an optical fiber. But you won't find a TOSLINK input port on a car stereo. Another solution would be to transmit the audio using bluetooth.

\$\endgroup\$

Your Answer

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

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