Noise issue- single ended signal and different ground reference?

[longwaytofall]

I have a troubling noise issue that I spent a fair bit of time diagnosing today. My amplifier (3e audio unit) is installed in my car, and is fed by a dc boost module to go from 12v to 48v. The amplifier is fed by a DSP, which runs on 12v. Because the DSP has single ended RCA outputs (not balanced), I modified the amplifier to accept the single ended input. I am getting a roughly 20hz tone (I think this may be PWM noise by a component in the car). My question is, is it possible the noise is being caused because the RCA cable signal ground is grounding to the 12v dsp and the 48v amplifier?

Here are the combinations I tried that made me suspect the signal ground:

DSP>3e amp=noise
Phone with usb c to rca converter>3e amp=no noise (rules out issue with amplifier)
DSP>amplifier/speaker outside of car (with long rca)=no noise (rules out issue with DSP)
DSP>commercial car audio amp=no noise (I'm not sure if car audio amps run fully on 12v throughout? Or they manage the signal ground differently?)

I also did several other things like run the dsp from a different 12v source to eliminate ground loops, tried different RCAs/shielding etc.

Another thing that makes me think an issue with the signal ground is that when I insert the RCA into the amp while it is carrying a signal from the dsp, I can start to hear the music when the middle signal pin makes contact without any noise, but when the ground connection is made the noise starts.

Any help would be greatly appreciated, I am stumped on how to address this!

 

[wwenze]

Looks like the ground voltage at the amp is different from the ground voltage at the DSP alrite.

One way is to run the DSP on 48V side. The 48 to 12V converter will also inject ground noise but if you use a small one it should hopefully be less.

 

[longwaytofall]

Would it be possible to tie the grounds of the 48v and 12v together, even though the power supply is "isolated"? Here is a sketch of the layout:

 

[wwenze]

It is possible, however it is possible the ground is already connected, because:

when I insert the RCA into the amp while it is carrying a signal from the dsp, I can start to hear the music when the middle signal pin makes contact without any noise, but when the ground connection is made the noise starts.

So adding additional ground connection may solve the problem, but you would need to add a very big one, because clearly adding a small one via RCA connection just diverted all the ground current to the RCA. Worth trying, anyway. Check with the 12V-48V converter's manufacturer regarding safety.

48VDC is starting to get pretty high where voltage safety is concerned, by tying the car ground and the "48V ground" together you are essentially putting your human body on the same earth as well, so you touching the 48V will close the loop.

If you want to go that route, treat the 48V with the same level of safety as you would with AC mains i.e. double-insulated cables, no exposed connectors etc, 2 layers of protection.

And I searched the background of this project a bit... Balanced connections are excellent for this kind of situations. Heck, even by simply rewiring your amp from single-ended input to balanced input, essentially treating the RCA as a differential connection (Connect the DSP's RCA ground to the amp's -ve connection) Because if your source ground and receiver ground are different, you want the amp to reference to the source ground and not the receiver ground.

 

[longwaytofall]

Thanks for the reply! I temporarily grounded the 12v ground to the 48v ground, and the noise stayed exactly the same. That is interesting regarding switching back to balanced inputs and using the single ended signal, I'll give it a try!
This amp (4 channel, 2 on the 3e class D, and 2 on seperate lm3886 boards) has noise on the lm3886 board I tested earlier as well, which does not take a balanced input. If it works for the 3e board, I would still need to come up with a solution for the noisy lm3886 board.

Edit to add: something that did reduce the noise was simply taking a wire and touching either end to the outer shield portion of the RCA, basically offering additional shield ground pathway.

 

[AnalogSteph]

Because the DSP has single ended RCA outputs (not balanced), I modified the amplifier to accept the single ended input.

And therein lies the crux, I bet. How exactly did you go about doing this?

Car audio is a maze of ground loops by default, so amplifiers have to make accommodations for this, and their unbalanced inputs tend not to be a straightforward deal.

If the amplifier came with balanced inputs, the correct approach would have been making cables to make use of them. Specifically, RaneNote 110 #17 or #18 or equivalent depending on input connectors. You can fancify the whole affair even more by replicating DSP output impedance between RCA shield and the black (cold) wire. (Let's say the DSP has 100 ohms and a 10 µF capacitor in series in its output. Then you'd replicate exactly that.) I'd strongly suggest doing that and going Ctrl-Z on the whole amplifier mod.

 

[longwaytofall]

The modification involved adding another op amp to the board, per the user's manual. I have since removed the op amp and returned it to it's original balanced input, and it got rid of almost all of the noise! Working on the other amps now. For replicating the impedance and capacitance, the DSP has a normal RCA output, so I only have +/- with - also being the shield if it's a coax cable. You are suggesting I measure between the - at the dsp, and the (unconnected) shield end with the Rane #17? Or this is where I would add the resistor and capacitor? Here is a picture of what I think you mean.

 

[staticV3]

The modification involved adding another op amp to the board, per the user's manual. I have since removed the op amp and returned it to it's original balanced input, and it got rid of almost all of the noise! Working on the other amps now. For replicating the impedance and capacitance, the DSP has a normal RCA output, so I only have +/- with - also being the shield if it's a coax cable. You are suggesting I measure between the - at the dsp, and the (unconnected) shield end with the Rane #17? Or this is where I would add the resistor and capacitor? Here is a picture of what I think you mean. View attachment 369072

Here's the ideal RCA->XLR cable:


You want identical output LCR from both the DSP's RCA Signal and RCA Ground pin.

Most RCA outputs don't have that out of the box, so by retrofitting it with a custom cable you establish impedance balance and get the full CMRR that your Amp is capable of.

If your Amp uses a different jack for its differential input, the same thing applies.

 

[longwaytofall]

Ok, this is making more sense. How do I establish the value of that resistor? Here is what the DSP specs say:
Analog Outputs: 16 x RCA, 4 Volt (5 KOhm), Impedance: 100 Ohm
Thanks for the help, I appreciate it!

 

[staticV3]

Ok, this is making more sense. How do I establish the value of that resistor? Here is what the DSP specs say:
Analog Outputs: 16 x RCA, 4 Volt (5 KOhm), Impedance: 100 Ohm
Thanks for the help, I appreciate it!

A 100Ω resistor should get you up and running.

For precise LCR matching (likely not required), additional instrumentation would be necessary.

 

[longwaytofall]

Hey, question before I order the resistors, will this decrease the gain output of the DSP?

 

[staticV3]

Hey, question before I order the resistors, will this decrease the gain output of the DSP?

It will not.

 

[tmuikku]

^ post #8 example has RCA sleeve connected to cable shield, which should not. Snip the shield from rca end to make it like rane #17 mentioned earlier so that the shield is connected only at the XLR end. Try the resistor if you want, my expertise ends there and I guess effectiveness of the resistor depends on both the output and input circuits implementation. To be sure, you should assemble the cable with alligator clips or some other means to try which configuration makes least noise.

I'm quite sure though, that if cable shield was connected to both XLR pin1 and to RCA sleeve it will very likely make ground loop between the devices and almost certainly add noise. Purpose of the pin 1, or cable shield, is to carry any error current due to external electromagnetic field straight to ground. It is an extension to device chassis and should never go to audio circuits. Chassis is then grounded to safety ground, or in case of car there is likely some other common ground scheme. It is very possible the input on the amplifier is poorly implemented and pin 1 is not connected to anywhere inside the amplifier, or worse yet it goes to pin3 or audio ground or something, in which case it might be less noise if shield is attached on RCA end instead.

I highly recommend Henry W. Ott book called Noise Reduction Techniques in Electronic Systems, or bit more heavier book Electromagnetic Compatibility Engineering which I think has pretty much same important content, and you'll be able to solve all your ground and noise issues rest of yer life without doubt. Reading that it's relatively easy to imagine what the noise mechanisms are and how to avoid problems with them. At least twenty years ago hobby recording equipment I had was so bad with this stuff I believe no one in the company had read the book and I'll assume a lot of consumer electronics still goes to sale as good enough, without proper noise supression techniques. So, read it and make your systems perform better than many

 

[rcstevensonaz]

Here's the ideal RCA->XLR cable:
View attachment 369073

You want identical output LCR from both the DSP's RCA Signal and RCA Ground pin. Most RCA outputs don't have that out of the box, so by retrofitting it with a custom cable you establish impedance balance and get the full CMRR that your Amp is capable of. If your Amp uses a different jack for its differential input, the same thing applies.

I realize OP's context is car audio, which has very different grounding and RFI/EMI interference than a home audio setting.

In a home audio setting, how useful is it to add a resistor to impedance match on the RCA shield connection to the Pin 3 wire? And do we lose all of the CMRR benefit when connecting an unbalanced turntable, CD/DVD, DAC, etc. if we don't add such a resistor to match the impedance?

Also, @staticV3, do you agree with @tmuikku (Post #13) that the cable's shield should not be connect on the RCA side? As noted in a different post, I am seeing conflicting answers on whether to connect the cable's shield wire to the RCA sleeve. That is, should your diagram be modified to this:

 

[solderdude]

Archimago shows the effect of ground loops and solutions in a new article.

The E1DA Cosmos Stack: ADCiso and 2 APUs. An example of bad multichannel - Neil Young's 'Harvest' on DVD-A (2002), fixed in new Atmos mix.

A blog for audiophiles about more objective topics. Measurements of audio gear. Reasonable, realistic, no snakeoil assessment of sound, and equipment.

archimago.blogspot.com