are laptop USB inputs just (low-level) noisy? (for recording audio)

[krabapple]

I noticed, just fooling around, that recording 'nothing' from an external ADC with USB out (i.e., nothing connected to the ADC input), into USB in of either of my two laptops running Audacity, produces recordings with low-level noise -- not normally audible, but audible if I record or play back at extreme levels.

Curiously, the laptop recording noise profile varies by ADC and by sample rate (44 vs 48kHz). But in a frequency plot it is always a bunch of very low frequency junk (ranging from 1-30Hz), plus just one or two narrow band spikes anywhere from 450 Hz to 15kHz. Nothing rises above -75dB even when recording level is set at maximum. In spectral view at max recorded level, it's one or two horizontal lines (frequency varies by ADC and SR) and greater intensity at <30Hz, on a background of wideband hash . When made audible it sounds like....a tone or buzz on a background of white noise.

Running the same ADC into my desktop workstation's USB in, the resulting file really is silent, even when the recording and playback levels are set to maximum.

I tried using three different outboard ADCs (all fairly cheap), different USB input ports....same result (laptops make low-level noise, desktop doesn't). Moving the laptops to different locations didn't make a difference. The laptops run on their battery power, not plugged in. The ADCs are powered via their USB connection. So I don't think it's ground loop issues?

If I disconnect the ADC and record the laptops' 'internal' sound by using Audacity Speaker loopback mode and recording a silent browser page, the file is totally silent.

So I conclude it is something to do with recording via USB ports*: they are 'noisier' on the laptops than on the desktop.

Again, this isn't noise I will hear at any reasonable recording level and playback levels. But it's there and that bugs me. Is it normal? Am I right in thinking it's the USB ports?




*One of the laptops has a 1/8" jack that has dual headphone/microphone function, and I haven't tried using that.

 

[DVDdoug]

The digital audio data is generally immune to noise.

But USB power is often noisy and with a USB powered soundcard/interface that noise often gets-into the analog-side of the ADC, It tends to be worse with microphone inputs because of the high gain preamp amplifies the noise.

There are some USB isolators that can isolate & re-generate power, and they also help with ground loops.

And there are USB audio interfaces with their own separate power supply.

*One of the laptops has a 1/8" jack that has dual headphone/microphone function, and I haven't tried using that.

The mic preamp built-into soundcards and laptops is often low quality (noisy), it's mono, and it doesn't work with "good" stage/studio mics which require a balanced XLR connection. (The combo jack also needs a TTRS plug with an extra connection for the microphone.)

 

[krabapple]

The digital audio data is generally immune to noise.

But USB power is often noisy and with a USB powered soundcard/interface that noise often gets-into the analog-side of the ADC, It tends to be worse with microphone inputs because of the high gain preamp amplifies the noise.

There are some USB isolators that can isolate & re-generate power, and they also help with ground loops.

And there are USB audio interfaces with their own separate power supply.

Thanks! So the problem is not USB digital audio data transmission per se, but that the interface is powered via the same USB connection?

But: why is the same powered USB_out --> USB_in configuration silent using the desktop, but injects noise in both laptops? Is the laptop-provided power noisier?
How do 'USB isolators that can isolate & re-generate power' work?

The mic preamp built-into soundcards and laptops is often low quality (noisy), it's mono, and it doesn't work with "good" stage/studio mics which require a balanced XLR connection. (The combo jack also needs a TTRS plug with an extra connection for the microphone.)

Yeah, I didn't have high hopes for that.

 

[somebodyelse]

I noticed, just fooling around, that recording 'nothing' from an external ADC with USB out (i.e., nothing connected to the ADC input), into USB in of either of my two laptops running Audacity, produces recordings with low-level noise -- not normally audible, but audible if I record or play back at extreme levels.

Floating inputs can pick up noise, especially if they have high impedance. If you want to see the ADC's self noise you should short the input, or use a resistor representative of the source.

 

[majingotan]

Have you tried just isolating the 5V power part of the USB cable by using an external source instead of using USB power from laptop? It should help fix the noise issues you're getting from the ADC

 

[SHB]

Wait a sec: every piece of audio gear produces noise regardless of the power source.

 

[DVDdoug]

Here are a couple of USB isolators.

Thanks! So the problem is not USB digital audio data transmission per se, but that the interface is powered via the same USB connection?

Right.

But: why is the same powered USB_out --> USB_in configuration silent using the desktop, but injects noise in both laptops? Is the laptop-provided power noisier?

Probably. There are switching power supplies/voltage regulators inside the computer, and some are noisier than others. Some audio interfaces are also better-filtered than others. there is also lots of other electrical noise inside the computer from all of the high-speed data "action". Again, this is no problem for digital, but analog audio and our ears are very sensitive to noise. (Although -75dB is not bad.)

How do 'USB isolators that can isolate & re-generate power' work?

It's a DC-to-DC converter. It's also a switching device, DC is converted to AC, then back to DC, but it can be better filtered. The isolating type (which the Wikipedia article doesn't seem to show) has a transformer which isolates the input from output, just like any other transformer or power supply... There is no direct electrical connection between the input and output. The electrical energy is converted to magnetic energy, than back.

Wait a sec: every piece of audio gear produces noise regardless of the power source.

True, but some devices are noisier than others. And under normal conditions it only affects analog. Mic preamps and phono preamps tend to be worse because they are high gain and any noise generated internally, of coming into the input gets amplified.

Digital noise is rarely bad-enough to corrupt the digital data, and when it happens it's usually terrible.

 

[krabapple]

Some people here seem to be assuming the noise is from the ADC.I need to emphasize: when the ADC(s) is connected (via USB), and its resting (i.e., nothing connected to its input) output recorded in Audacity, I get:

(a) when connected to my desktop Win 11 computer: essentially 'no' noise
(b) when connected to either of my two Win 11 laptops: low level noise

The noise referred to in (a) is visible in spectral view, and can be made audible by cranking output. It is invisible and inaudible in (b)

I tried using this cheap 'USB data and power isolator' .

Spectral views:

laptop (ignore the different lengths of the recordings, that's just me slow on the stop button):

desktop:

(in waveform view, all of these are just a horizontal line, as expected)

 

[somebodyelse]

(i.e., nothing connected to its input)

Does the noise with the laptops disappear if you short the input rather than leaving it floating?

 

[danadam]

(in waveform view, all of these are just a horizontal line, as expected)

Can you zoom-in?

 

[krabapple]

Does the noise with the laptops disappear if you short the input rather than leaving it floating?

How do I "short the input"?

 

[krabapple]

Can you zoom-in?

of the waveform? Or the spectrum? What are you wanting to see?

 

[danadam]

of the waveform? Or the spectrum? What are you wanting to see?

Yes, the waveform. The spectrogram shows that it isn't just a random noise, so I was wondering if maybe the waveform shape can give some clue.

 

[antcollinet]

How do I "short the input"?

Whatever the input connector is - use a a shorting plug with the signal pin(s) connected to the ground pin. EG if the input is RCA use an RCA plug with tip shorted to ring.

 

[somebodyelse]

And for balanced inputs it's more about connecting hot to cold than connecting those to ground. So for XLR, join pins 2 & 3. For TRS join tip and ring.

This is to make sure there's really zero input rather than a floating input that can pick up noise.

 

[krabapple]

Whatever the input connector is - use a a shorting plug with the signal pin(s) connected to the ground pin. EG if the input is RCA use an RCA plug with tip shorted to ring.

?

I can try this, but I'm not getting the point. What is the hypothesis that a shorting plug is testing?
With nothing connected to ADC input, noise is recorded in one of the test conditions (laptops) and absent in the other (desktop).
Nothing changed but the computer the ADC is connected to.
If it's some kind of noise from the ADC, why does it manifest (aka 'pick up noise') in one condition but not the other? That's the interesting question to me. (The noise itself is so low level that it's not much of a practical concern otherwise. )

Here's one of the rather humble ADCs I used that is showing this behavior (Amir tested this model, but frustratingly, he only tested it as a DAC).
Input is RCA L/R. Output is USB-A --> laptop (or desktop)

I won't have time to mess with this again until next week.

 

[antcollinet]

With nothing connected to ADC input, noise is recorded in one of the test conditions (laptops) and absent in the other (desktop).
Nothing changed but the computer the ADC is connected to.

If it is resulting from noise on the USB ground - then that is likely also on the audio ground. If there are even tiny amounts of stray impedance - eg stray capacitance or stray inductance between the unconnected input - and any other part of the circuit that doesn't have that same noise - then it will result in a different noise signal on the analogue input.

Conversely if it is USB power that has the noise - then that can be present anywhere in the circuit - again - stray impedance can couple that onto the analogue input if it is left floating.

This doesn't happen when an input signal (or short circuit) is present - because the input signal will have a very much lower impedance than the stray coupling.


EDITED to add : The difference between laptop and desktop may well be that the laptop doesn't have an actual ground to take the noise away.

 

[antcollinet]

And for balanced inputs it's more about connecting hot to cold than connecting those to ground. So for XLR, join pins 2 & 3. For TRS join tip and ring.

It's still better to connect them to ground. Balanced inputs do not have perfect rejection of common mode noise due to imperfect impedance balancing. While it is quite likely to be enough to prevent the noise getting through, it does depend on the level of the noise being coupled. We only need around 120nV getting through to flip the LSB. The actual levels of noise on the input are already MUCH higher than this - even when there is nothing "extra" that you are looking for.

 

[Sokel]

Ground 101 and it's easy to test.
Just run a thick wire from any laptop's shield connection to a solid ground around your house.

But, USB in general is not exactly the perfect PS, far from it.
You just got lucky with your desktop (and its ground/general grounding scheme), others may not be.

 

[krabapple]

It's still better to connect them to ground. Balanced inputs do not have perfect rejection of common mode noise due to imperfect impedance balancing. While it is quite likely to be enough to prevent the noise getting through, it does depend on the level of the noise being coupled. We only need around 120nV getting through to flip the LSB. The actual levels of noise on the input are already MUCH higher than this - even when there is nothing "extra" that you are looking for.

The input is RCA,. The device has RCA in/out , headphone 1/8" out, and USB out.

Either of its outs can be run into its in, of course. If that helps totest anything.

If it is resulting from noise on the USB ground - then that is likely also on the audio ground. If there are even tiny amounts of stray impedance - eg stray capacitance or stray inductance between the unconnected input - and any other part of the circuit that doesn't have that same noise - then it will result in a different noise signal on the analogue input.

Conversely if it is USB power that has the noise - then that can be present anywhere in the circuit - again - stray impedance can couple that onto the analogue input if it is left floating.

This doesn't happen when an input signal (or short circuit) is present - because the input signal will have a very much lower impedance than the stray coupling.

IIRC it is there with an input signal (my guitar), too. I will double check.

I still don't grok what 'shorting its RCA input' would mean unless it involves one of these

EDITED to add : The difference between laptop and desktop may well be that the laptop doesn't have an actual ground to take the noise away.

a-ha. I was going to say, how is any of that relevant if the desktop USB noise performance is 1000x better than any laptop USB?

Ground 101 and it's easy to test.
Just run a thick wire from any laptop's shield connection to a solid ground around your house.

You people...

what would 'run a thick wire from any laptop's shield connection to a solid ground around your house' look like? What is my laptop's 'shield connection'?

But, USB in general is not exactly the perfect PS, far from it.
You just got lucky with your desktop (and its ground/general grounding scheme), others may not be.

You and antcollinet converge here

I could use the laptop AC power source, which has a ground plug, instead of battery power, is that worth a try?