horrible "hum" with lot of harmonics in audio recording

[frank10]

I tried to make some audio recordings with 2 mics battery powered to XLR (they outputs a minijack TRS that connects to an adapter XLR with pin 1-3 shorted).
Well, I get some different behaviour:
a) main problem is with device powered by transformer (12Vdc):
  a1) in one building I get HORRIBLE and strong hum+lot of harmonics
  a2) in another building I get little hum on the mains frequency of 50-100Hz
  a3) in the 2 last building tests, I get no hum at all (it was 2 commercial centers, maybe lot of filtering on mains?)
b) all these hum problems disappear when I disconnect from the transformer, with device battery powered the signal is perfect (it's perfect even when using balanced mics with phantom power, device also powered by transformer). But they remains if I leave the transformer's cable near (some cm) the input of the device: it needs to be far away to be ok!

Then I tried to disconnect the mic and leave only one short cable connected to XLR adapter: the same hum result! So, it's not the mic battery powered but just one cable connected, making an "antenna"?


Just to see the spectrogram:

worst, awful hum:




little hum, even when power cable near the device:




perfect signal (commercial center, even if there are some higher frequencies  with mains):



So, my questions are:
1) why these problems only when connected to mains through the transformer or when near to the device only on some buildings? This does not seem a ground loop at all (I'm connecting this only device to mains, no other connected to it).
It seems the external cable becames an antenna and with battery not... so it should become a long antenna through mains cables? (It is also when not connected, only at short distance.)

2) If the problem is the mains on a specific building, should I see the bad frequency signals making a FFT on the mains, or similar test?
Why there is no hum on commercial center, how they filters their mains?

3) The more important one: the easy solution is recording with batteries, but anyway, how could I solve the problem on the transformer connected through mains, if I had to power it for many hours/days?

[capt bullshot]

From your description I'd guess improper or not connected shielding, maybe bad adapter (try another one)

[frank10]

I tried also making an XLR adapter+cable myself, soldering it. So I'd exclude this.
Furthermore, if bad adapter, it should behave always bad, not only on some buildings, no?

I absolutely don't know if it's related, but I posted also this (on the same unit):
https://www.eevblog.com/forum/beginners/strange-dmm-measurement/

[capt bullshot]

Yes, I'd say it's related.
It seems the ground of your recorder box is charged with some leakage current from the power supply. This isn't a bad thing in general. quite usual with switching power supplies, less with simple line frequency transformer units.
You could try to ground your unit (its case or a dedicated grounding pin) to the safety ground wire of the wall power outlet and see if this fixes  the problem. But anyway, this problem (the hum, not the leakage current) should never occur with properly designed equipment ...

[frank10]

I have tried to connect USB cable while transformed powered and the hum diminishes a lot... but it's always there!
Even when touching the metal case it improves the problem, but doesn't solve it.
I'll try the direct to GND tip.

I'll try to change also the power supply, maybe could solve?

Anyway, for the rest, I'm very happy with the quality recording and features of this unit, it's really great for the price, it's the new Zoom F4 portable recorder.
But I agree with you, this shouldn't happen!
I wonder if it's a problem of my unit only, but it seems not easy to discover asking someone else or their engineers 

[frank10]

I already tested its transformer and has this ripple of about 13mV:

apart the noise spikes @100Hz

It seems not bad ripple+noise comparing to other switching I have...:
https://www.eevblog.com/forum/beginners/building-bnc-cables-to-measure-ripple/msg1128826/#msg1128826

[capt bullshot]

Changing the power supply could improve the hum, try one of these old "iron and copper" (non smps) ones.
The scope picture doesnt show line frequence ripple, did you test it without load? What you see there could be the result of the low power standby regulation, the waveform may look completely different if the supply runs with its normal load.

[Andy Watson]

Furthermore, if bad adapter, it should behave always bad, not only on some buildings, no?

Have you actually bleeped-out the continuity between the ground on your recorder and the shielding/ground on your microphone?

[frank10]

@capt bullshot
Yes, it was a test without load, I'm at beginning in testing with oscilloscope, so I'll already planned to do next tests with load.
I measured the frequency spikes with cursor and it was 100Hz.


@Andy
Yes there is continuity between grounds.

[capt bullshot]

I'm not familiar with this scope, it appears to be set to 20ms/div so there should be exactly two spikes per div if it was 100Hz. Additionally, the waveform at the output of a simple rectifier wouldn't have such large spikes. If you look closer, the spikes appear to be somewhat uneven distributed, what you see there is not the 100Hz from a line frequency transformer powered rectifier.

[frank10]

Yes, I relooked at that: it is not a fixed frequency, it oscillates horizontally... I made a Freeze and measured that, it was 100Hz, but... not fixed and more, not even distributed.

[frank10]

Thank you, capt bullshot!
I tried a linear power and the hum problem improved a lot!
When I connected the USB or the direct GND cable, it disappeared completely!
So, those solved the problem. It's needed to be careful not to put mic cable near some mains cable, because it'll capture something; with these precautions, all is well. I'm happy.


For the ripple tests with load, I'll try this schema:

https://www.eevblog.com/forum/blog/eevblog-594-how-to-measure-power-supply-ripple-noise/msg409891/#msg409891

adding a third parallel cap of 1000uF to better simulate the load (as I was suggested). Is it correct?

The problem with that config is I don't have differential probes, but they said it can be done also with normal probes: so, I used 2 normal probes @1x, one on the V+, the other on V-, without GND connected, but I got strange high values, like 67V... How should I try a (pseudo) differential test?
If I can't make this I'll test with a single probe tip+short GND.
 
What load should be used? (I have switching power supply from 12V2A, 30V5A, 12V10A, plus linear 12V0.5A) Something like 0,4A max could be well? (I don't have more W power resistors...)

[capt bullshot]

It's needed to be careful not to put mic cable near some mains cable, because it'll capture something; with these precautions, all is well. I'm happy.

In my opinion this is not desirable for this kind of equipment. There must be still something wrong, like the shield of the mic cable isn't connected to the recorders shield. Has the recorder a GND lift switch?

For the ripple tests with load, I'll try this schema:
...

Seems overkill to me. If your supply is floating, just set the scope to AC coupling and use one channel. Try the probe at x10 and x1 setting. Ensure a real short connection from the supply output GND to the probes GND (don't use the pigtail wire supplied with the probe).

If your supply isn't floating (output connected to safety GND), you can still use the same setup, either plug the scope and the supply into the same wall socket and ensure a real low impedance GND between them, or use a isolation transformer (breaking the safety GND loop) on one of them and ensure low impedance GND.

This setup works in most cases for me.

[frank10]

It's needed to be careful not to put mic cable near some mains cable, because it'll capture something; with these precautions, all is well. I'm happy.

In my opinion this is not desirable for this kind of equipment. There must be still something wrong, like the shield of the mic cable isn't connected to the recorders shield. Has the recorder a GND lift switch?

I also discovered that with linear transformer, GND connected in normal operation is perfect, but if I put mic cable near a mains cable, the little resulting hum is greater than without the GND connected! So, it depends on situation: linear is always best, GND connected it depends (normally yes).
I tested 2 mics powered by batteries, only one of them has this problem with the cable near mains. So it's possible it has some cable issue or construction material issue.
In conclusion, it seems there are 2 problems, one with the recording unit due to leakage case with power switching (constant) and the other (unfrequent) due to specific mic...

Anyway, apart from linear transformer that solved it, should the problem be solved with some external filter, using the standard power switching? Or is really a problem due to the switching tecnology, so it can't be bypassed?

[capt bullshot]

Anyway, apart from linear transformer that solved it, should the problem be solved with some external filter, using the standard power switching? Or is really a problem due to the switching tecnology, so it can't be bypassed?

Remote diagnosis isn't easy, I your particular case I guess the hum results from the leakage current of the switching supply. I do not expect the switching frequency and its remainders to be the cause, the internal power supply of the recorder should be able to deal with that (ripple etc.). Filtering the input or output of the switching supply won't help.
Due to the EMC compliance the switching power supply must have some capacitance (typically some nF and special rated Y capacitor) between primary and secondary. This causes a power line frequency (and harmonics) current to leak from the AC line into the recorder. As the recorder is isolated, it represents a quite high impedance against earth, resulting in a voltage (measurable between PS output and earth). This is quite normal and nothing to worry about.
Now some other failure kicks in: as you wrote, the mic lines are suspectible to line frequency hum if you place them near power lines (I do consider this as a serious failure). Placing some AC voltage (by the leakage current) on the recorders GND, this results in the mic lines beeing suspectible against hum by placing them near to earth (which is basically everywhere). So the recorder hums if you use the original switching supply.
The typical linear supply doesn't have capacitors between primary and secondary, so the leakage current is lower by some magnitudes. So your stray voltage on the recorder GND to earth is lower by magnitudes and the mic lines are much less sensitive to earth. That is what you see with the linear supply.
Grounding the recorder (by a explicit wire or any other way) also reduces the stray voltage on the recorders ground greatly. So your mic lines are less sensitive to earth.
The basic problem isn't solved but mitigated (by grounding the recorder or reducing the leakage current). You wont be able to filter (keep it away from the recorder) the line frequency leakage. You could try to insert a very low leakage isolated DC/DC converter between the power supply and the recorder to isolate the recorder from the leakage.

I don't know any details about the XLR inputs of the recorder, an XLR input should be some kind of symmetrical input and should have a solid ground for shielding the cable. Can you figure out, which of the XLR pins is connected to the recorders circuit GND (maybe accessible at some of the outher connectors, maybe metal case). Ensure that this pin is connected to the cable shield. Try connecting this pin to each of the signal pins, try leaving the signal floating. Try connecting the mic cable shield to the housing or GND of the power supply instead of a pin of the XLR connector (do not connect the shield to a signal in this case, as this might damage the input).

[frank10]

Thank you for the great explanation.
Some doubts:

Remote diagnosis isn't easy, I your particular case I guess the hum results from the leakage current of the switching supply. I do not expect the switching frequency and its remainders to be the cause, the internal power supply of the recorder should be able to deal with that (ripple etc.). Filtering the input or output of the switching supply won't help.

But, this doesn't explain why in some buildings the hum is quite imperceptible or nothing at all. So it must be something in the mains of some buildings that causes this problem to arise. Otherwise it should be there always. Hence my question on filtering. Maybe instead of the output of the transformer, filtering the input mains...?

I thought maybe with some inverter that has some filtering/stabilization on mains: so I connected to one.
Instead the hum is worse when connecting to mains from a (good) interactive Inverter UPS (that should only stabilize (and filtering?) the mains when not on battery). Instead of cleaning the mains signal, it cause more hum!?

Now some other failure kicks in: as you wrote, the mic lines are suspectible to line frequency hum if you place them near power lines (I do consider this as a serious failure).

Not completely true: the mic lines are susceptible IF there is one particular mic that has this problem in its cable. Otherwise all mics should behave the same and give interference if put near power mains: that's not the case, I tested one mic that hasn't this prob. So, I'm thinking this other problem is not a recorder prob, but mic one, probably of its cable.
[/quote]

[capt bullshot]

EMC / hum and buildings: the world has seen many interesting things happening in this context ...
Just a guess:
- there might be more or less harmonics on the power lines, causing more or less harmonic leakage current
- (forgot this one) especially in larger buildings there may exist circulating / stray ground currents, causing magnetically coupled hum
- one building is build of electrically more isolating material (like wood and drywall), another one is more conductive (like concrete). This causes different leakage to earth from the recorders floating ground.
- the leakage path inside the switching power supply may not be symmetrical, different result if you plug it in the other way (live and neutral swapped)

An UPS usually does not isolate its input from the output, so the path for the leakage current is still there. Try an isolation transformer.

Some mic hum, other don't:
still sounds like a shielding / grounding problem. More investigation required, you name it: the mic or the cable can be faulty or not suitable for this recorder.

[frank10]

I tried an isolation trasformer like this:
http://www.ebay.it/itm/262471171920?_trksid=p2060353.m2749.l2649&var=561276497169&ssPageName=STRK%3AMEBIDX%3AIT
there is less hum, but still it's there.
If I connect the USB it goes away. So the isolation transformer alone doesn't solve this issue.

[bson]

I tried a linear power and the hum problem improved a lot!
When I connected the USB or the direct GND cable, it disappeared completely!

You can also connect the negative battery terminal to an earth ground if you have one on hand.  It sounds like the whole mic apparatus is floating, connected to a high-impedance amplifier that either presents a poor ground (on the XLR ground pin) or has poor CMNR - or is floating itself and the problem manifests further down the chain.

[Loboscope]

Apart from the problem with the power supply discussed above, I will refer to the first information you gave in the first post:

I tried to make some audio recordings with 2 mics battery powered to XLR (they outputs a minijack TRS that connects to an adapter XLR with pin 1-3 shorted).

What microphones do you use? Following the description that they are battery powered and have mini-jack TRS, they should be more cheap microphones, because more professional microphones will be phantom-powered and could always be connected via XLR-symmetric cable and connectors. The ZOOM F4 is a more professional recorder and it delivers 48-V-phantom power and symmetrical XLR-connectors. By shorting pin 1-3 of a symmeterical XLR-connection you will loose any advantage of the symmetrical connection, which will result in a very bad common-mode-rejection and a bad noise-level.

If one of the microphones will be a bad cheap scrappie one and the cable used will also be a cheap and bad one, you will have any chance to capture any hum and noise buzzing through the air. It don´t make me wonder, that you will pick up noise from a power chord when this cable lies immediately alongside.

Do you have a possibility to test the equipment with a rather good, almost semi-professional microphone connected via symmetrical XLR-cord and powered directly from the P48 of the F4?

[frank10]

Yes loboscope, they are of course cheap mics.
I have other mics XLR balanced powered by 48V Phantom and they have no problem at all, as I stated in the point b) of my first post.

Of course this is a professional unit and normally it's used with professional mics, as I usually do.
But, sometimes it can happen you must use the mics you have in that moment or in emergency use some cheap mics, maybe not yours.
So, that was a test with some cheap (but nonetheless good quality recording) mics (one is the Aputure a.lav).

As noted by capt bullshot, a professional unit should not show that behaviour (of switching power).

For the mic cable noise, yes I think it's due to cheap mic... useful to know you must stay away from contact with power cable...
Yes, it's always better to have a balanced connection from mic to recording unit, but it's not always possible.