Help with headphone amp grounding issues.

[dschwartz]

Hi all the masters here!
I'm having kind of a frustrating issue with my NCP2809 headphone amp boards.

The NCP2809 is a headphone amp in a chip, 125mW, 5Vcc with virtual ground for capless outputs.
I used the datasheet circuit with a stereo volume pot in front of it.

https://www.onsemi.com/pdf/datasheet/ncp2809-d.pdf

the issue i have is that if i'm using an isolated power supply (guitar, HP amp, Headphones), there's no actual ground getting into the circuit, so it hums like crazy and becomes an antenna. The only solution i found is to connect it to another AC grounded appliance, then the hum stops (as it becomes properly grounded)

have anyone had a similar issue before? how can i make the enclosure to shield from EMI without AC grounding it?

cheers!!!!

[anotherjim]

You'd connect the shield to the supply 0v.
Where is the 5v coming from? The 1uF cap shown on the chip power is probably a recommended "local" bypass close to the chip. Your supply may need better filtering, especially if it's USB power.

[dschwartz]

That's what bothers me. The shield IS connected to gnd, and the 5v are regulated via 7805 with hefty 22uF each side.
If i use my DIY LM7809 regulator power supply (where gnd connects to mains gnd) there's no issue. But with a strymon OJAI or similar supplies with isolated grounds, there's no faraday effect..the whole box becomes an antenna and buzzes if i touch it.

[MikeA]

The Strymon Ojai designers probably assumed you would use it for pedals that get an earth ground from your amp, via the shield lead of the pedal's output cable.  No need for their power supply to provide an earth ground to your pedal.  They probably didn't plan to power an isolated device like your headphone amp.

[anotherjim]

The capless datasheet schematic shows inputs grounded to supply ground. What happens if the inputs (and screening) reference the internal output splitter ground?
Strangely, your link to the datasheet pdf worked yesterday but it now only goes to the technical document search page for me.
I wonder if this one will stick...
https://www.onsemi.com/download/data-sheet/pdf/ncp2809-d.pdf
...yet it appears to be identical but does go straight to the pdf. Weird.

[Rob Strand]

That's what bothers me. The shield IS connected to gnd, and the 5v are regulated via 7805 with hefty 22uF each side.
If i use my DIY LM7809 regulator power supply (where gnd connects to mains gnd) there's no issue. But with a strymon OJAI or similar supplies with isolated grounds, there's no faraday effect..the whole box becomes an antenna and buzzes if i touch it.

What are you calling the shield?

Connecting 0V to mains earth tends to improve most situations.   When you can't do that or don't what to do that then you need to use shielding.

If your circuit is placed in a metal enclosure and you connect the 0V/ground of your circuit to the metal enclosure then the enclosure acts as a shield.

If your circuit is placed in a plastic enclosure then you have to add shielding.   The simplest form of a shield is a larger ground plane on the PCB.  Better is some metal panels surrounding the PCB, with all metal connected to circuit 0V/ground.

A good test/experiment to see if a shielding will help a plastic enclosure is to wrap aluminium foil around the enclosure and connect the foil to circuit 0V/ground.  If that doesn't work you might find the problem is elsewhere.

If you get noise problems when powered from battery then it's likely the enclosure shielding isn't good enough.

What's your power source exactly?    There's plenty of examples where people are getting noise from the power supply itself.   One issue is some wall-wart/plug-packs oscillate unless you have a minimum load - a good value is 220 ohm to 100ohm across the plug-pack as a dummy load.   Another issue is the isolated power supply injects noise between mains ground and its output.   That's hard to fix.   Connecting the circuit to mains earth obviously fixes it however if you don't want to do that, or can't do that, then using another power supply is a better alternative.

Beyond that you would have to investigate other problems like the circuit oscillating (yes grounding can stop this sometimes) and RF issues with the circuit itself.

[dschwartz]

By shield i meant the enclosure, which is 2mm thick extruded aluminum. It is definetly grounded via the pots cases (i use 9mm metal shaft)

I power it with a strymon zuma, which is exceptionally silent..weirdly, i don't remember having this issue with a cioks supply. But anyways the issue dissapears if i connect one of the other outputs to anything connected to mains.

My guess is that the zuma voltage is not referenced to mains and both 0v and 9v are floating..hence, the ground is susceptible to EMI

[Rob Strand]

By shield i meant the enclosure, which is 2mm thick extruded aluminum. It is definetly grounded via the pots cases (i use 9mm metal shaft)

If the pot cases are grounded to chassis but there is no connection between the circuit's 0V and the chassis you can get problems.    By that I mean you need to measure 0 ohm between circuit 0V and chassis.   In fact without the connection to 0V the grounded pot chassis become very efficient points to couple noise into the circuit.  Noise getting in by touching the chassis is a sign this is what is going on. 
(A cap between circuit 0V and chassis also work but I'll leave that for another discussion.)

If your 0V is connected to the chassis then that points to a whole different problem.   Very likely to be caused by the PSU being switch-mode and with isolated outputs.

I power it with a strymon zuma, which is exceptionally silent..weirdly, i don't remember having this issue with a cioks supply. But anyways the issue dissapears if i connect one of the other outputs to anything connected to mains.

My guess is that the zuma voltage is not referenced to mains and both 0v and 9v are floating..hence, the ground is susceptible to EMI

There's been a few weird noise issues reported on the forum with these later generation isolated supplies.

When you have a switch mode PSU with an isolated output it's very easy to develop noise between the outputs and mains earth.     Switch-mode PSU's normally use (special) caps to provide an RF path from the PSU output back to mains in order to pass EMC.  That can be enough to stop noise through the ground path but not always.   

With the *multiple* *isolated* PSU outputs on these pedal power supplies I have a feeling this RF path is not provided on each output.   The cap might be present on only one output or they re-jig things to have no such cap at all.    Whatever is being done obviously causes problem on some set-ups - especially in floating set-ups like yours.

I have not back-engineered one of the PSU's to see what's going on.  I'm only commenting based on common issues which arise with switch-modes.

I suspect if you wired a 47n to 100nF cap from the PSU output 0V to the PSU box chassis your noise problems would be gone.  (That's assuming your 0V is connected to the chassis, as per first paragraph.)

[PRR]

You usually can't play electric guitar inside a room with electric power unless your amplifier is referenced to the average electrical field in the room. Some form of connection to the electric supply. Otherwise BUZZ.

ONE direct connection to safety ground/earth (nearly all post-1960s US amplifiers) is usually best.

A less-ideal result comes from a 0.05uFd cap to the less-live side of the power outlet (what you find in pre-1965 US amplifiers).

Two connections in one amplifier chain is liable to buzz. So, many of these "pedal supplies" try to be totally isolated from the wall outlet, letting the main amplifier's 3rd-pin bonding be the one ground. Now you have NO link to ground (ambient field) and buzz is likely.

That Zuma has a 3-pin plug, but deliberately does not feed that through to the DC outlets. That's best when you have a real amplifier with a real wall-ground. Since you don't, try to find a neat way to bond your amp chassis to the Zuma chassis. A clip-lead will do for a test: hum should drop sharply. For gigging you want a more robust connection.