Can you "fix" a product with earth leakage current?

[paulca]

I was going insane recently because everything I touched with a scope seemed to be floating somewhere around 8-12VAC 50Hz.

Ground hums, large amounts of wideband noise the works.

It should have been more obvious since a lot/most of my breadboards have involved audio recently....

After poking around things systematically I happened to find the problem.

The Numark desktop speakers are leaking 275uA 93VAC.  The voltage I am seeing floating on everything is because everything in that vicinity is audio gear which is protecting itself by floating.  It obviously becomes an issue while I am using USB devices which are Earth referenced, as is my scope.

Tonight I had a circuit on a breadboard which was earth referenced in 2 places.  The scope and the Tenma (after I looped its DC - to it's GND ref ), however the output audio was modulated by a 50Hz hum and muffled.... unless I added capacitance to the ground loop and stuck my thumb on the top of a cap!  It was like I was missing a ground completely.  But it wasn't just fully grounded, but earthed as well!  I was going nuts.  "That shouldn't happen!".

Turns out the speakers have been dumping 90VAC at 275uA constantly into the grounds and most likely zooming around in loops creating all my noise.  Most things when measured only show 10VAC or so to Earth. 

In theory, 275uA won't bother the house trip switch, could I not just jury-rig an earth reference for the SMPS in the speakers?  Would ignorantly connecting the audio output GND to earth be at all wise?

[TheMG]

The leakage would be primarily due to the Y capacitor(s) in the SMPS.

The purpose of the Y cap is to reduce common-mode radio interference generated by the SMPS, so while the operation of the SMPS would not be affected, you should not remove or change the value of this capacitor or you risk violating EMC regulations.

Yes, you can ground reference the output of a floating SMPS without ill effect, but that may actually make things worst from an audio ground loop standpoint. Audio ground loops and how to avoid them can be quite a complicated topic.

[paulca]

Rock and a hardplace.

The ground loop tonight was obvious yet.  However what was also apparent is that it doesn't like to be pulled low at the audio ground point.  As soon as I connected the scope to the breadboard the audio got mangled as the scope was pulling the audio GND down to earth.  Without the ground connected the scope showed my signals modulated on top of a 50Hz sine wave (and a mess).

There has, come to think, always been a low level hum on the headphones when the speakers are powered on, often I disconnect them or turn them off, I put it down to a ground loop, but it's more likely coupling of that 90VAC leakage being grounded to the PC audio connection.

So if I earth reference the PSU there will be a loop formed between the PC ground and the speakers ground and potentially the leakage and other noise will just "freewheel" around the loop creating more EMF than I would like into the bargain of adding hum to the audio.

Are audio transformers a thing?  Would that even have an effect?  Would it not couple through the 90VAC as well?

I'm thinking maybe 10 years out of these is long enough.  Which brings me to a second problem.  I don't want to spend £500 on a small amp and speakers and if I go down the Amazon no-brand route, I run a high probability of encountering another leaky SMPS.  I suppose, at least with amps you can look at the back and see if they have a ground lifter / lug.

[Gyro]

...
Turns out the speakers have been dumping 90VAC at 275uA constantly into the grounds and most likely zooming around in loops creating all my noise.  Most things when measured only show 10VAC or so to Earth.
...

Just a small correction. The leakage current is developing 90VAC into the 10Meg input of your DMM in voltage mode. It is only passing 275uA into the virtual short circuit of you DMM current range. Not both at the same time.

... but yes, it can be a pain.

[themadhippy]

Are audio transformers a thing?

yes

Would that even have an effect?

maybe

Would it not couple through the 90VAC as well?

no as the 90v side will have a lower impedance to earth so wont bother jumping through the transformer.
 Try your local car audio dealer for a hum eliminator or ask your local pa/music shop for a passive DI box

[paulca]

I'm in the process of digital-izing the audio network.  It occurred to me that an optical input would fix that issue.  I already got 2xUSB isolators which use transformers and optocouplers.  Something similar for "Line level" is maybe just a matter of course.

The Numarks are 10 years old.  If I did got to amazon for a cheap "DAC" based desktop amp and/or powered speakers ... if it has an optical input it doesn't matter that it has leakage current (or is just live at 240V).  ...  If it's only connection to the audio system is via optical SPDIF.

In the meantime I might take the suggestion and try and find a non-snake-oil DI box (my brother will know that's his trade).  Depends on price. 

An option also exists to just rip the powersupply out of the Numarks and microwave it.  Then put a non-leaky or less leaky SMPS in there... or a linear supply.  They don't draw much.

[Ian.M]

Maybe replace the primary to secondary Y capacitor in the SMPSU with two in series of double the value, and earth the mid-point via a RF choke.   That will divert most of the 50Hz leakage to earth while still allowing HF leakage from the switching node end of the primary via the interwinding capacitance to return whence it came, avoiding EMI problems.

[themadhippy]

try and find a non-snake-oil DI box

snake oil is not as prevalent in the pro audio world,just dont  buy  a beringer D120 .

[paulca]

Maybe replace the primary to secondary Y capacitor in the SMPSU with two in series of double the value, and earth the mid-point via a RF choke.   That will divert most of the 50Hz leakage to earth while still allowing HF leakage from the switching node end of the primary via the interwinding capacitance to return whence it came, avoiding EMI problems.

Yes I seen it suggested in several places that is the way to create a low leakage SMPSU.  They didn't mention the RF choke part, just using 2 Y class caps.  Allow it sounded like they were just adding the second cap to earth, not putting them in series and tapping the centre.

The other consideration is these speakers are 10 years+ old now.  I know Y class should fail open (??) but maybe it's time to get rid anyway.  No point recapping or spending money on a 10yo speaker set which cost £100 to begin with.

[Ian.M]

Neglecting labour costs, you could probably do the mod and replace any dubious caps in the PSU and still have change from £20.  You certainly wont be able to get anything half-way decent to replace it so cheap.

[Siwastaja]

Earth leakage current of a few hundred µA is completely normal and cannot be "fixed"; and need not to be fixed.

Whatever reacts badly to this leakage, needs to be fixed instead.

For example, audio needs to be either differential ("balanced" because audio folks prefer different terminology for the same thing), or pseudo-differential so that instruments use two-prong (double insulated) mains frequency transformer power supplies and the audio "ground" does not form loops. Latter strategy is successfully used in many musical instruments, but higher end studio gear on the other hand uses differential audio instead.

Laptops are notoriously problematic as sound sources; their audio outputs are only designed to drive headphones. There really is no simple solution.

[paulca]

The thing that is annoying my is that I'd like to use the speakers while I'm testing on the breadboard.  Everytime I connect the 2 grounds between the speakers and the USB stuff crashes and it just feels wrong.

Some devices don't take kindly to it.  The PCM5102 DAC output is badly distorted if I have both the scope and the speakers connected.

Nothing is confirmed to have died from this, although one STMF411 did die while running, just after I connected a USB UART to it.  Could have been other things too.

Given the amount of stuff I have hooked up together the worry is there is enough capacitance in the floating 100VAC to do damage at the point of connecting an earth referenced device to the speakers floating leakage.  I know there is sod all "charge" and sod all current and even the tiniest of resistances will pull it down, but I'm worried about the transients.  How much current will flow whlie I connect the two grounds.  How much capacitance and inducance can build up and what damage can it do?

Besides.... it fuels a hum in ground loops badly.  As I think I said I used to disconnect these for ground hum reasons.

[paulca]

Neglecting labour costs, you could probably do the mod and replace any dubious caps in the PSU and still have change from £20.  You certainly wont be able to get anything half-way decent to replace it so cheap.

You are probably right.  It might be a better option to just disconnect the PSU entirely and re-use the speakers as passives.  I can buy a reasonable stereo 40W a side amp with an optical input for £50.  I get a better amp and still get another 10 years out of the speakers... and no ground shenanigans.

[oz2cpu]

here is one my solutions for this

http://webx.dk/oz2cpu/230V-trafo-isolation-low-capacity/230V-trafo.htm

and here is my final solution
https://youtu.be/4rNubfeICOE

[NiHaoMike]

Another solution is to replace the bridge rectifier with a single diode, so that the mains side would be at neutral and not much to feed through. It would require doubling the primary side capacitor and is most practical for low power devices.

[Gyro]

Pulling any significant unbalanced DC current from the mains like that can de-sensitize Type AC RCDs and RCBOs. Not recommended.

[robert.rozee]

here is one [of] my solutions for this
[...]
and here is my final solution
https://youtu.be/4rNubfeICOE

one word: aluminium (or aluminum for those who can not spell).

(you could possibly also ground your single turn, but that would involve more words)


cheers,
rob   :-)