Guitar and audio interface noise

[Francesco Chieffo]

UPDATE:

I was able to carry out further experiments in the last few days, finally being able to connect one of the free outputs of the audio interface to a small amp. Much of the noise actually went away proving that the first problem was lack of grounding.

Unfortunately, however, that's not all, in fact in that mass of noise there was also a rather acute whine, which according to the spectrum analyzer seems to be precisely located on 3k. This very sharp and annoying sound persists even when connecting the amp to the Audient, or for example using an external amp modeler connected to the audio interface.

I'm back to ask for your help since this story is driving me crazy. What is it about in this case? How can I fix?

[Francesco Chieffo]

Anyone? Should I start another thread?

[Wonks]

That's very specific noise, which indicates to me that there's some electronics working at that frequency and causing interference. It might be within the equipment you are using, or it might be something external.

For instance, 3kHz is a typical switching frequency for a frequency inverter (used to control variable speed motors). But it could be a multitude of things which you'd need to know a lot of detailed information about to try and narrow it down without physically turning devices on and off.

So you really need to try moving things around and turning items off one at a time to see if anything makes a difference.

[merlyn]

Is the whine present when the guitar is unplugged? If not, that would suggest the pickups are, eh, picking up electromagnetics, maybe from the computer screen?

If the whine is present with no inputs connected to the interface, that could be something wrong with interface.

Can you work out where in the signal chain the whine is being introduced?

[ajay_m]

There is such a thing as a "power break in usb box" . It's a usb isolator. All the signals are optically coupled and the downstream PSU is isolated. It can only source around 400mA though but you can modify the output cable to splice in a separate PSU to resolve this. Only catch is that the cheap units are only usb1.1 speed but this will handle 4 or even 6 audio channels just fine. They will fix most noise problems in my experience without introducing any significant latency and they do not of course compromise the audio signal

[ajay_m]

Sorry duplicate post

[Martin Walker]

I once went to the trouble of opening up a USB cable with a craft knife and removing some 10mm of the screen. I had a laptop that buzzed with an AI when on mains power but not on battery. Open circuiting the screen did not fix it.

As I understand it, a USB cable only has its screen connected at the source end, and not at the device end, so removing some of the screen wouldn't defeat a ground loop.



To do that you would probably have to cut its black GND wire (or preferably add a small resistor in series with it, say 50 or 100 ohms, which is a great cure for ground loops caused by audio cables).

CAVEAT: I've not tried this myself, and with devices drawing larger amounts of current the series resistor would also drop the 5V supply.

[Francesco Chieffo]

I believe it's not the instrument since I own several ones and the problem occurs with all of them. For clarity, the only noiseless scenario is guitar into the audio interface (with an amp sim plugin) and headphones connected.

The whine already occurs just connecting the Macbook (I just bought a new one, but it is also there with the previous one I owned) to its power supply. Of course the noise is imperceptible if you don't activate amp modeling plugins, that obviously introduce saturation. Usually, however, instead of plugins I use the Universal Audio Dream '65 pedal as a modeler (the unity has been tested recently).

Could it be a dirty power problem? I've been living in this house for a few months and last year (another house, another city), I didn't have this problem.

[ef37a]

I once went to the trouble of opening up a USB cable with a craft knife and removing some 10mm of the screen. I had a laptop that buzzed with an AI when on mains power but not on battery. Open circuiting the screen did not fix it.

As I understand it, a USB cable only has its screen connected at the source end, and not at the device end, so removing some of the screen wouldn't defeat a ground loop.



To do that you would probably have to cut its black GND wire (or preferably add a small resistor in series with it, say 50 or 100 ohms, which is a great cure for ground loops caused by audio cables).

CAVEAT: I've not tried this myself, and with devices drawing larger amounts of current the series resistor would also drop the 5V supply.

Did not consider that Martin! I had just assumed a USB cable screen was continuous like an audio cable but as you say, they are not, at least the A B cable I pulled from the rack at random is not.

Op's 3kHz whine smacks to me of a low grade 16 bit converter? These were common in early 16 bit USB mixers. Even my A&H Z10 generates a low level whine at around that frequency. They soon changed to a very quiet 24 bit converter.

Dave.

[Francesco Chieffo]

Op's 3kHz whine smacks to me of a low grade 16 bit converter? These were common in early 16 bit USB mixers. Even my A&H Z10 generates a low level whine at around that frequency. They soon changed to a very quiet 24 bit converter.

Dave.

I use an Audient iD14 mkII. Shouldn't its converters be working at 24 bit?

[ef37a]

Op's 3kHz whine smacks to me of a low grade 16 bit converter? These were common in early 16 bit USB mixers. Even my A&H Z10 generates a low level whine at around that frequency. They soon changed to a very quiet 24 bit converter.

Dave.

I use an Audient iD14 mkII. Shouldn't its converters be working at 24 bit?

Yes indeed it should and I very much doubt it is responsible for the noise but it should be easy to eliminate it.

Note, it is not the 16bit working that causes the noise, most AIs can do 24 and 16 bits and the latter free from artefacts, it is some design shortcoming in these low end converters causing the whine.

Dave.

[ajay_m]

I have also noticed that the most obvious noise in these scenarios is around 3KHz though it can vary. As far as I can see, what's going on is that within both the PC and the audio interface, we have a lot of digital circuitry. Each transistor (strictly, a MOSFET in most cases) acts as a small switch, with very fast rise and fall times and cumulatively they impose a very erratic load on the main PSU in each unit (or via bus power in some cases). Obviously the designers know this and there are significant amounts of capacitors sprinkled around, as anyone who has done electronic design knows, you normally decouple each chip with a small capacitor across power and ground.

But while this ensures that transients are smoothed out, as seen by the chips themselves, those transient currents still flow through the capacitors and they make their way to the ground plane. The transient currents can be very high, in the order of amps, and so depending on the design of the PCB, its possible for portions of the ground plane to experience a voltage drop as a result, relative to other portions of the ground plane as these currents flow around the circuit.

Now the impedance of the ground plane is very low but the currents are very high and it only takes a millivolt or so of drop and you have a problem now.

Depending on the design of the PC and/or audio interface, you then end up in a scenario, as I said, where the ground reference at the audio input or output jacks is not precisely zero volts, but is in fact jiggling around by a millivolt or so. This represents an incredibly low impedance source which can easily provide a LOT of current. (I would guess maybe 0.01 ohm impedance or even less).

Both the PC and the audio interface can then have ground planes which are independently 'jiggling around' by a tiny amount of voltage. This is probably why some combinations of PC and interface are particularly bad.

On the other side, the 'hot' signal wire represents a very high impedance signal path and carries negligible current. So now if you were to put a sensitive meter between the two units across the hot side wire, you'd see 0 volts, because so little current flows through this wire it looks like a short circuit.

But do the same on the ground side and you'll see a tiny signal. Because this signal is sourced by an incredibly low impedance high current source, any normal piece of wire looks like quite a high resistance and consequently the two ground points aren't, in fact, at the same potential. Since the audio interface amplifies the difference between the signals, this appears as noise.

When you use balanced I/O this helps somewhat because now, in theory, the difference between the two hot lines is amplified and any 'common mode' signal is rejected.

But the idea of balanced I/O is to handle a common mode signal on the hot side - typically, 50Hz mains hum. In this case the common mode signal is being introduced via the single ground wire.

Unfortunately this digital noise has an incredibly wide bandwidth - it will go out to several megahertz and possibly beyond. Most opamps used in audio devices are, understandably, specced to handle signals in the audio range and their common mode rejection ratio figures - which we rely on when using balanced I/O - degrade rapidly outside the audio range. As well as this the common mode noise is being injected into the ground plane on the audio interface so it can radiate out from there into signal paths that don't have any rejection.

This then allows the spurious signal to get back in via pathways that have no intrinsic common mode rejection, and as a result of intermodulation effects it can reappear in the audio spectrum. (and as well as this I think, to do with the underlying software, there are routines that run with scheduling frequencies in the audio range, which imposes a 'modulation' on these current spikes that then appears as an audio 'whine'.)
Also LEDs (including display backlights) seem to be driven by PWM signals that end up in the audio range. I've often found on equipment where the LEDs can be turned off, that this affects the level of the resulting whine.

Unfortunately the interaction between PC and audio interface somewhat randomly then ends up with noise issues. Bus-powered units seem to be more vulnerable but there's no real pattern unfortunately.

It does look as though USB2 optical isolators are starting to become available, albeit they aren't cheap compared to the USB1.1 solutions (which are only about £15). This is a magic bullet solution though and is pretty much guaranteed to resolve the problem. If you only have 4 or 6 audio channels, then USB 1.1 is perfectly adequate. You may, though, as I said, have to splice a PSU in on the output side (by cutting the red wire and then splicing into the black wire without breaking it.

Now your ground path is completely broken, so any ground plane voltage transients can't propagate across the isolator because the signals are optically coupled and neither the 5V or ground wires have a circuit between the devices. Instead, downstream power is now supplied by a second PSU which has no electrical connection to the PC, as the PSU is double insulated.

Why don't the manufacturers make sure that the ground points for USB ports and audio jacks are always at 0V?. Well, it's very tricky. There's no point at which the ground plane is ever really at 0V, because those transient currents also have to flow back up into the main PSU (which for a laptop is always external) and so the resistance of the ground wire means that relative to true ground (which is buried inside the PSU effectively), the ground plane will always jiggle up and down a bit.

Digital circuitry doesn't care about this because the threshold voltages for the logic gates are much higher than the groundplane noise. But analogue circuitry cares very much, which is why we get problems.

[merlyn]

I believe it's not the instrument since I own several ones and the problem occurs with all of them.

I'm not suggesting there is anything wrong with your guitars. Pickups by their design can pick up electromagnetic waves.

For clarity, the only noiseless scenario is guitar into the audio interface (with an amp sim plugin) and headphones connected.

The interface is OK then, if there is a whineless scenario. The iD14 is 24 bit, and it's not the convertors.

The whine already occurs just connecting the Macbook (I just bought a new one, but it is also there with the previous one I owned) to its power supply.

There are a few combinations and permutations of which gear is connected. Let's say you have :

Guitar > Interface > Headphones (Mac on battery) -- no whine
and
Guitar > Interface > Headphones (Mac on AC) -- whine

Now if you pull the guitar lead out is there still a whine? If so it's not the pickups picking up electromagnetics, and it may be worth trying a USB isolator, as suggested by ajay_m.

Could it be a dirty power problem?

Probably not, despite what the power conditioner industry might tell you. The Mac works on DC. The power supply converts AC from the wall into DC, so the DC should already be cleaned up and filtered if the Mac works.

[James Perrett]

Back in the early days of computer audio manufacturers of internal audio cards used to go to great lengths to reduce computer noise to a minimum. This would include paying extra attention to grounding and power supply filtering.

It seems that many modern manufacturers of external interfaces possibly aren't paying the same attention to noise reduction. Perhaps they think that having the convertors in a box away from the computer is enough, without any other measures being needed.

It is interesting that most dedicated digital audio connection systems like SPDIF, ADAT, AES and even networked systems use isolated connections to prevent grounding issues. My personal choice is to keep the computer and audio parts separate by using ADAT or SPDIF links between the convertors and the audio interface.

[Francesco Chieffo]

Unfortunately I have a not very big desk here and I have to keep everything not too far apart. As for the USB isolators, could you point me to one? I tried looking for optical ones and they seem quite expensive. I haven't actually found any USB3 ones, the MacBook Air I have now only has these ports.

At this point I have to wonder why last year in the house where I lived I had none of these problems.

[ajay_m]

Just Google usb isolator Amazon. The first two hits I got were around £17 or £18 . They use an admu3160 chip. Note that for bus powered audio interfaces the downstream power available is less than 400mA which may not be enough for some interfaces. In this scenario I would splice in a 5v PSU like a phone charger to provide power. These should work fine with a usb3 port but they are usb1.1 12Mbps devices so they should handle 4 to 6 channels at 44.1khz ok (at 24 bits this is approx 1Mbps per channel not allowing for protocol overhead)

[merlyn]

I have also noticed that the most obvious noise in these scenarios is around 3KHz though it can vary ...

I think your explanation makes sense and hangs together. What I wonder is how you would explain, in this case, why there is no whine when working on battery power?

[ajay_m]

On battery power CPU clock speed is usually reduced and the idle clock speed is often set very low as well. Plus backlight intensity is usually reduced. This could significantly affect the amount of interference, quite a bit could be coming from the voltage regulation subsystem which handles 10s of amps at very low voltages. Also with a PSU connected you have a fairly good antenna to radiate interference out for possible pickup via leads, guitar pickups etc.

[Francesco Chieffo]

Just Google usb isolator Amazon. The first two hits I got were around £17 or £18 . They use an admu3160 chip. Note that for bus powered audio interfaces the downstream power available is less than 400mA which may not be enough for some interfaces. In this scenario I would splice in a 5v PSU like a phone charger to provide power. These should work fine with a usb3 port but they are usb1.1 12Mbps devices so they should handle 4 to 6 channels at 44.1khz ok (at 24 bits this is approx 1Mbps per channel not allowing for protocol overhead)

Thanks a lot, I will definitively try this.

[merlyn]

If you want the 480Mbps bandwidth of USB 2, there is the Topping HS01. It also has a USB C input for a phone charger, which the Audient iD14 will probably need.