Grounding Issue or Something Else?

[msft]

Hi all,


I've been working on designing/building an op-amp based headphone amplifier. The power stage is driven by the LT3080/LT3015 linear regulators, and the output stage is driven by the LM4562 (for gain) and LT1206 (for current) op-amps.


The ground scheme is as follows: all of the grounds related to the power section are tied to a small "power GND" bus, and all the grounds related to the signal section are tied to a small "signal GND" bus. There is 1 jumper that runs between these two ground buses to tie them together.



I've built a prototype of the system on perfboard, and the audio itself sounds quite good. However there are a few issues with noise that I'm facing:

1. Occasionally I can hear short bursts of clicking sounds through the headphones. I find that this effect is more frequent when I plug into the power outlet that my computer/modem runs on. Strangely, I also find that if I hold the amp further away from the wall then this almost completely stops, which makes me think it's some kind of EMI issue (though it could also be an issue with my AC/DC converter cabling).
2. If I run the amp with my phone as the source, I can sometimes hear radio signals being picked up. I currently don't have a metal enclosure, but I'm thinking adding one and grounding the chassis could fix this?
3. With certain 3.5mm cables driven from my computer, I find that I can get different amounts of buzzing on my output. The shortest cable I have buzzes quite a bit, while my longer cables don't buzz at all.

I was wondering if anyone has any insight into what my issue might be? Especially concerning observation 1, I'm pretty lost on what to try next. I've tried increasing the size of my decoupling capacitors before the regulators, which (subjectively) seems to have helped a bit, but the clicking remains. Most of the noise in the system doesn't seem to vary much when I increase the volume of my source signal. Photos of the board are here: Imgur: The magic of the Internet


Any help would be greatly appreciated!

Thanks

 

[rayma]

I would go ahead and install this in a metal enclosure, before any more troubleshooting.
That may well take care of of most of the problems. Are there 0.1uF decoupling capacitors
for the op amp, mounted close by? Try the circuit in a different area of your home that is
away from other electronics, on a battery operated laptop or device, if possible.

 

[msft]

I would go ahead and install this in a metal enclosure, before any more troubleshooting.
That may well take care of of most of the problems.

I agree this would be good to try next, I'll order an enclosure ASAP.

Are there 0.1uF decoupling capacitors
for the op amp, mounted close by?

I have similarly sized decoupling caps by the output of the regulators, but nothing near the power rails of the op-amps themselves. What would be the purpose of adding more near the op-amps? Filtering out stray noise picked up by the traces/jumpers?

Try the circuit in a different area of your home that is
away from other electronics, on a battery operated laptop or device, if possible.

When I run the amp from sockets away from other electronics, I don't hear near as much of the clicking that I mentioned. It still happens ever so often, but with less amplitude compared to by my computer. This makes me think it's a power issue, but I'm unsure if it's related to a "noisy" ground or ripples on the inputs themselves.

 

[rayma]

Most op amps require close decoupling, to reduce the inductance in the path.
Tack on one cap at each op amp supply pin to ground.

Also try a 10uF cap at each regulator output, if you only have 0.1uF caps there now.

 

[msft]

I would go ahead and install this in a metal enclosure, before any more troubleshooting.
That may well take care of of most of the problems.

So I did go ahead and install it in a metal box, but the issues still remain. I tried multiple chassis grounding schemes. Neither connecting it to earth ground nor connecting it to signal ground made any different (earth ground was actually worse). Floating chassis didn't help either.



I have some new observations though:


1. If I touch the pads of the output 3.5mm jack, the persistent buzzing is significantly reduced. In this case I am laying my finger across the L, R, and GND channels. Touching the input jack has no effect.



2. If I touch the pads of the LM4562 gain op-amp, clicking and radio interference seems to be reduced as well. This makes me think that the decoupling caps on the power rails that you mentioned might help. I will go ahead and solder them on.

 

[rayma]

Ok, use 0.1uF 50V ceramic types directly at the power pins of the op amp, keeping the leads as short as possible.
If that does not work, post several close photos.

 

[msft]

Ok, use 0.1uF 50V ceramic types directly at the power pins of the op amp, keeping the leads as short as possible.
If that does not work, post several close photos.

Wow, those tiny caps made a big difference. As far as humming/buzzing and radio interference goes, the amp is exceptionally quiet now. Thank you for the suggestion!



However, the sporadic clicking/popping still remains, and I think I've found the culprit. I noticed that it only occurred when I held the amp near my computer/modem/router. I tried unplugging the router, and it stopped! But that leaves me a bit confused on what to do. I already have low pass filters on the signal input, cutting off at around 1.5 MHz, which should be well below what's required to filter wi-fi out. I also have ferrite chokes on the power supply input cables.



The metal chassis does seem to help a bit with this, at least slightly attenuating the amplitude of the pops, but they are still audible. I guess it's not too big of an issue, considering I can just keep the amp few feet away from the router and the interference pretty much goes away. It would still be nice to find a solution though! I'll keep investigating.

 

[sgrossklass]

The LM4562 may be at least part of the remaining problem:
Sensitivity of opamps to air coupled EM fields, especially of the LM4562/LME497X0 family | Audio Science Review (ASR) Forum
(pma is also a member here, IIRC.)

It goes without saying that keeping RF in the GHz range out is not an entirely trivial task. Lead lengths have to be kept to a minimum for capacitors to be effective. Contact points between case parts should be bare metal on bare metal. I would recommend a few pF to case directly at each input and output (L, R and GND). There should not be any slits in the case.

 

[rayma]

Try a different router, just in case that one is worse than average.

There is radiated EM interference, and there is conducted EM interference.
Often you can have both. To start we can check for conducted EMI.

Try shorting plugs in the amplifier input RCAs to see if that's where it is arriving.
If so, try much shorter input cables to the amplifier.

However, sometimes conducted EMI can be through the back door (amp output).

 

[msft]

The LM4562 may be at least part of the remaining problem:
Sensitivity of opamps to air coupled EM fields, especially of the LM4562/LME497X0 family | Audio Science Review (ASR) Forum
(pma is also a member here, IIRC.)

Thank you very much for this article. After reading through it and the replies, I decided to try changing my gain stage op-amp. I've had a good experience with Analog Devices components, so I went with their LT1361. I switched it in today and the amp is dead quiet now. No more popping from the router or any weird RF hissing. Strange that the LM4562 performs so poorly in this regard.



I've also designed a PCB and got it printed (my first time doing so). Not the ideal layout I'm sure, but it works nonetheless. Picture here: https://i.imgur.com/dJwv0tI.jpg


Cheers!

 

[Rick PA Stadel]

Also, keep in mind the 1,5MHz input LPF is both higher than it needs to be for audio (there are still plenty of high-powered AM transmitters operating in the US ), and is not a *brick wall* filter. Attenuation a few octaves above the 'cutoff frequency' is probably only 25 or 30dBV.

Congrats on finding a remedy!

Cheers