Buzz problem detective wanted...

[molecular]

Hi everyone, haven't been here in a while, but I am having trouble with some new noises in the studio since I introduced one of Focusrite's new X2P Pre/Interface Dante boxes...

I'm still investigating some things, but thought I'd see if anyone here had any advice. I don't *think* it's a Dante issue, but here's the summary...

My X2P generates a horrible buzz which can be heard on the recorded signal, and also in the headphones on the unit. It has a preamp/playback mix knob, and the noise is there in the local mix, so I'm assuming that it is being generated in the pre, before ADC.

The buzz is present when all the following things are true:

1. The flourescent working lights in the live room are on
2. The pre is connected to a router on it's Cat6 cable
3. The mic cable connecting the mic to the X2P is close to, or lying on, the floor.

As of today:

1. if I switch the lights off, the buzz disappears.
2. if I disconnect the Cat6 cable, the buzz disappears. (all audio shuts down when you do this, but you get about 1 second of the "local" signal in the HP mix with no buzz before the audio shuts down).
3. if I lift the mic cable off the floor, the buzz disappears.

So I can use the unit, but within certain limitations which I'd like to get to the bottom of.

Worth saying that I have other Dante equipped pres in the same room which don't have these problems, and other analogue pres which I've never had the issue with.

I replicated the problem in the same room using a single ethernet router which was not connected to the wider network in the building.

As far as I'm aware, the floor is concrete with an industrial lino/marmoleum on it as the base surface.

Any ideas where I should start looking? Just to clarify, at the moment the buzz is only present when the striplights are on, an ethernet cable is present AND the mic cable is on or near the floor. Which seems like a strange selection of conditions to me, but I need to be able to work with all three of these things being true :/

Am also in discussion with focusrite, but maybe I'm just forgetting or overlooking something small or even big and obvious?

Thanks!

Hector

[Hugh Robjohns]

I'd start by swapping the mic cable and the mic for known good alternatives as a first step. It would also be worthwhile checking to see if pin 1 in the XLR is linked to the shell at the male (preamp) end, and if so, cut that link (or find a cable that doesn't have the link) to see if that makes a difference.

However, the fact that you have other Dante preamps in the same space that don't exhibit the same problem suggests this is an issue with that specific Focusrite unit. My guess would be either an internal grounding problem in that specific unit or an RFI susceptibility issue (these two possibilities could actually be one and the same, of course!)

For that reason, once you've done those basic mic/cable replacement checks, I think it will need Focusrite's direct involvement. Could they let you borrow a replacement model to check whether it's a fault with that specific unit, or a design issue?

H

[molecular]

Hi Hugh, thanks for the quick reply.

The buzz is present on a few different cables and all mics with and without phantom. The intermittent crackle I'm less sure about...

I'm waiting on a PoE injector as I haven't tested powering it from that yet, but I know the buzz has been present in other buildings... I'm going to see if I can trigger it using something other than striplights.

I have already sent one back and this is now X2P #2 which is giving me the same issue, so it does seem like they are just sensitive to the by design though :/ which is frustrating as they are otherwise my new favourite thing!

Any idea why the 20cm or so off the floor is causing the havoc, when the offending lights are a good 4m up in the apex of the roof? Is that just the effect of the steel mesh in the concrete floor?

[ef37a]

If a silly question, tell me to 'od off but. Is the Ethernet cable shielded and if so is the shield continuous all the way, "chassis to chassis"?

I guess the 'floor' issue is a standing RF wave max field strength point.

Just fort. Modern lights should have their casings earthed, do they?
Dave.

[molecular]

Excellent questions, and at present I don't know the answer to either. I'm pretty sure the problem is there when I've isolated it to just one router and the cat6 cable that came with the unit, but don't know about the shielding.

I'll ask about the lights... I am not sure anyone's been up to look at them since the mid 90s...

I was having the same problem in a completely different building but I can't remember if it had striplights or not...I may have been next to a wifi router or something.

I remembered a further oddness: The pres on the X2P are digitally controlled and the buzz is noticeably worse at specific gain levels. E.g. it jumps between 52db and 53db, then at 54db is not quite so bad...!?

I've explained that but to focusrite tech support, hence I got the replacement unit. But they haven't told me about anyone else having similar problems and I would have thought that if they were this sensitive and there's nothing weird at my end, then they would have been getting loads of complaints...

[ef37a]

http://www.broadbandbuyer.com/products/ ... eGEALw_wcB

Unshielded plugs do not have the shiny metal bits. I cannot think they would fit them to unshielded CAT cable?

You might also try foil shielded mic cable? Not as flexible but it does give 100% screening. This also might be a case FOR grounding XLR shells to pin one?

Dave.

[molecular]

Thanks both!

The problem is definitely there when using only shielded Cat6 (with shiny bits).

I don't think any of the XLRs have pin 1 tied to the shell, but I'll see if I can sort that to test it.

I'm doing a big swap around of various mics, cables and locations and am discovering some very odd results that disappear immediately when not using the X2P, so will report back here and to focusrite :/

[molecular]

OK, well the good/bad news is that I've managed to replicate the issue at home by setting the X2P up in the proximity of a wifi booster (crackle every 1-3 secs), a macbook charging (continuous hum), and a fridge (popping sounds when the light comes on and off).

The initial response from Focusrite was that they tested the one I sent back and couldn't find anything wrong, but I'll press them to consider looking into some more real world situations. I've never been in a studio that didn't have a wifi router, a macbook AND a fridge.

I'm probably in the wrong bit of the forum now, as I was initially thinking someone might point me in the direction of a suspected dodgy wiring issue or something, but I'll let you know how I get on.

[James Perrett]

One thought - is there a ground in the system? I'm just wondering whether everything is floating which can make things more susceptible to interference. What happens if you connect an input or output to a piece of properly grounded gear?

[molecular]

Not sure actually - I would have thought that if I was connected to the building's ethernet infrastructure there would be a ground somewhere (it's a college building) but perhaps not.

I'll see what happens if I connect it up to some outboard or monitors...

[James Perrett]

Not sure actually - I would have thought that if I was connected to the building's ethernet infrastructure there would be a ground somewhere (it's a college building) but perhaps not.

If they're using unshielded twisted pairs then the ethernet won't be earthed. Even if they are using shielded cables, the path to earth will probably be fairly convoluted and could easily be broken if there is an unshielded switch or connector in the chain.

[molecular]

update...

So I've been chatting away with Focusrite support, who all credit have been very attentive, but have not got to the bottom of this and thought someone here might have some more advice, If they can face sitting through 15 mins of video content :/

I was asked to make a video demonstrating the problem I'm having and did that, and tried to include some of the weird idiosyncrasies of the noise.

https://vimeo.com/283453849

password is rednet

They suggested the same thing - that the unit wasn't earthing properly and it needs to earth through the ethernet connection. (they raised the fact that my comparison dante unit was earthed through it's PSU).

So I made a second video in which the other Dante unit isn't earthed and still works fine.

Also - removing the ethernet cable altogether from the X2P gets rid of the problem, which doesn't make sense to me, as then it definitely isn't earthed any way at all.

second video here: https://vimeo.com/284113513

password is also rednet

I'm really stumped, as the requirement to earth it through the network in order to avoid interference seems weird and is not mentioned anywhere in the manual - and rules out, for example, setting up something on the fly with a small router and a laptop.

Also, obviously a whole host of small pres and interfaces work fine with no earth anywhere.

So I don't know, it's unusable in the studio, but also at home or anywhere else I've tried it, and Focusrite say they couldn't find anything wrong with the one I sent back. So if anyone has any time to sit through the two vids (13 mins and 7 mins) and see if they have anything to add I'd be really grateful.

It's otherwise a great concept and I'm reluctant to write it off and ask for a refund.

Thanks!

H.

[Drew Stephenson]

That's weird. In a number of ways. What's the peak-to-peak time for that interference noise?

[molecular]

Sorry just noticed there was a new reply, Drew!

It is, and remains, weird. I don't have the session in front of me right now, but the interference picked up is completely different in different places at different times. The one in that video is caused by the lights, but I have also had a buzz which grew over the course of about an hour, then vanished, then slowly appeared again. Also intermittent crackles, pops, all kinds. I can pick up the fridge door light going on and off at my parents house, and their WiFi booster. On two location records I picked up awful buzzes coming from god knows what.

I am in touch with someone at Focusrite who is being very helpful and responsive, but insists no-one else has registered any similar problems globally with the X2P, and they can't replicate my results, and the design team seem stumped.

But I cannot replicated the results using ANY combination of other Dante devices, or any devices, in the same venues. I've had all manner of desks, pres, interfaces, mics, hardware and whatever else through a couple of the "offending locations" over the last few years and not had this problem with a single one of them.

So I am really confused and frustrated.

[Alec Mackay]

I'm reviving an old thread here, did you ever get to the bottom of the issue?

We have a Dante network that spans 2 floors in our studio building. We have 2 X2Ps that exhibit the same noise issues you're describing. In my case the noises originate from inexpensive LED strips and the motor control units for the building's elevators. I haven't had any intermittent crackling.

In the case of the elevator, the X2P picks up a clean 16kHz tone when the motor is running, which I have confirmed with the elevator technician is the PWM carrier frequency for the motor controller. It's not audible on mics or other gear, in fact it picks up the signal even when only a mic cable is connected (no mic). I've ruled out grounding issues on the audio cabling.

Our Dante network is unshielded CAT6, but all our other Dante devices are not affected by these external noise sources. Only the X2Ps are prone to this noise, and it gets worse closer to the noise sources.

I have tried a couple of solutions like powering from the wallwart supply, standalone POE injectors, adding power conditioners on the Dante switch supplying POE, and using fully shielded CAT6e cabling all the way to the switch, known good audio cabling and mics, and lifting cables from the floor. None of that resolves the problem. The only thing that works is moving the units further away from the elevator shafts.

I'm considering making a Faraday cage around the X2P to try reduce EMI - but the chassis should already be doing that unless its poorly grounded.

Anyone else experiencing issues?

[ajay_m]

My suspicion would be ground currents introduced via the ethernet cable. Both the LED lighting and the lift are going to put current spikes back into the neutral which is then bonded to earth at some point, usually at the switchboard. But earth is not some infinite 'sink' and consequently is never an absolute zero voltage relative to other earths, which in a building may well route back to a different switchboard and a different phase if the building is powered by 3 phase AC, not uncommon in larger buildings.

You then end up with a very low impedance source of electrical noise which sees ground-to-ground cable shields as much higher impedances; i.e, you could have a noise source with an impedance of 10 milliohms or less and that can get into everything via cables which look to it like a higher impedance. This then wiggles your 'earth' reference around on a connected audio interface which then causes spurious audio signals.

An ethernet isolator might be worth trying. This will ensure there is no galvanic path for any spurious currents via the ethernet cable.

Of course radiated EMI can't be ruled out either, in which case shielding might resolve the issue. It may even be a combination of the two.

[Wonks]

The X2Ps are powered either by class 2 power supplies, or else use Power Over Ethernet.

Class 2 power supplies let the output voltage float, so there is no direct connection from the output's 0v and ground. This means the system has to be grounded through a connection to something else that's grounded. If it isn't, it can be noisy.

POE doesn't appear to be intrinsically grounded (I know the name and general principles but minimal technical details about it) but connected equipment can be grounded if using shielded cables and the shield is tied to ground at one end. If not, again you are looking at having to provide a ground through something connected to the interface.

If you are using shielded POE cables to the X2P, are you sure that the shield has a good ground connection back at the switch room?

It may be that there is no solid ground for the X2P, which is why even normally good XLR cables aren't doing a proper job.

16kHz is a typical switching frequency for the variable output frequency converters that the lift motors are powered from. The output is also often not a pure sine wave, so they can give off a lot of harmonics as well if the output cabling isn't properly shielded. So the output power wiring from them should all be screened. This is often steel wired armoured cable, but more flexible cables with screens can be used. The weak point is often the local switched disconnector (isolator) by the motor, which is so often in a strong plastic case with no screening, instead of more expensive metal enclosures. And sometimes I've come across instances where the armour grounding isn't carried through the isolator. So it may be worth getting the installation checked (but you'll probably have to get the lift installers to do that).

So in this particular instance, I don't thing an Ethernet isolator would be a good thing, as you need the X2P to be grounded. Screened Ethernet cable may be the only means of providing that ground provided the switch (or whatever) it's connected back to has the shield connection properly grounded.

[Hugh Robjohns]

I'm considering making a Faraday cage around the X2P to try reduce EMI - but the chassis should already be doing that unless its poorly grounded.

A Faraday cage would be worth trying for sure. If nothing else, it would confirm whether the interference is directly through the case itself or via external cabling.

I suspect the case is not being grounded effectively.

It won't be via the external power supply and, poe supplies are often floating at source. If you're using unshielded cat 6 cables that won't do it either. Nor will connected mics (unless valve mics with grounded power supplies).

So that just leaves the line outputs.... have you tried connecting those to anything definitely grounded? If so, did it make a difference?

[ef37a]

"The only thing that works is moving the units further away from the elevator shafts."

That ^ tells me Alec that the problem is direct pickup of RF radiation and it would seem to me that the Focusrite units either do not have efficient RFI filter circuits or/and poor internal ground routing?

You could try some clip on ferrite absorbers but CAT cable is a problem in that you are not supposed to bend it in a tight radius, as you can for instance XLR cables. Such tight winds might not matter for Dante? I don't know how much of the CAT6 spectrum it needs...Hugh will know though.

You might also see if you can beg some CAT7 cable off a network bod? That has a full foil shield, a braided shield and a foil shield for each pair. I have about 10mtrs of the stuff and it is very stiff and a b*****d to fit RJ45s to!

Then again, RF can get into kit via many routes so try ferrites on every cable?

https://www.amazon.co.uk/Ferrite-RingOn ... NY74gdYjRr

Dave.

[Alec Mackay]

Thanks for all the replies. I've been trying some of the suggestions, using the LED strip lights as a worst case noise source. Here are my results:

• Shielded cabling all the way to the switch - not possible. It's an unshielded cat6 installation, and I don't have enough length of shielded cable it to test.

• Connecting to a standalone switch, isolated from the rest of the network - no difference

• Ferrite beads on the cabling (stolen from some VGA cables) - no effect

• Line out connected to a grounded output - slight improvement

• Ground wire directly from mains power (with live pins removed for safety) connected to XLR in line out - this was very sketchy, and made less improvement than connecting to a grounded output.

• Wiring the ground cable to a screw on the chassis - minimal difference.

• Holding onto that ground wire and placing my hands on the chassis - almost complete cancellation of the noise!

It is also interesting to note that I experience the same thing as the OP, where the signal is clean for a second before it shuts down:

2. if I disconnect the Cat6 cable, the buzz disappears. (all audio shuts down when you do this, but you get about 1 second of the "local" signal in the HP mix with no buzz before the audio shuts down).
3. if I lift the mic cable off the floor, the buzz disappears.

That tells me something, but I'm not sure what. Now I need to find a way to implement that in a practical way.

(I never use the LED strip lights anyway, so I still need to repeat my tests with the elevator noise to see if I get similar results.)

[Hugh Robjohns]

• Holding onto that ground wire and placing my hands on the chassis - almost complete cancellation of the noise!

That sounds like the case is not grounded and not connected to the xlr pin 1 ground... which would be unusual.

Do you have a multimeter?

If so, it would be work checking the resistance between the case and xlr pin1s.

[Acoustic Landscape]

Oh man!

Earthing!

Replace Ethernet cable with S/FTP
(every twisted pair is shielded with alloy foil with an overall braid shield) this is the most shielding.
Or U/FTP (every twisted pair shielded alloy foil only) a lesser but good shielding.
The cheapest/reliable S/FTP cable is Roline from R.S. feels good too.

Your good fault descriptors do allow a process of elimination and suggest that the device itself has inadequate shielding, but that can mean little when it comes to earthing loops or signal loops and proximity to SMPS which chop at high frequencies and radiate that.

Fluros shouldn't be in a studio environ, Fluorescent tubes use high frequency, high AC voltages to conduct through a partial evacuated tube to create the light, efficient in their time, but terrible quality of light and radiant emissions that will affect many devices in a studio, think strat on bridge pickup! -a great detector!

Do you have filtered mains in your studio?

Is the Studio star earthed?

In the last ten years+ the massive increase in AC mains audio noise is largely due to the use of Switched Mode Power Supplies in the network. One has to be filtering down to 10Khz these days, blocking everything above that. Big inductor coils (current compensating chokes) and center tap transformers that cancel 'differential mode' to earth are the key to make the biggest difference to drop the ( earth) noise floor significantly. ( a straight 1:1 mains transformer isolates but does not cancel out) A deep topic, with plenty wolf tickets for the snake oil merchants. But that is the key info.

Take a look inside a Furman P-2300 IT E, aside the massive tran, you will see a number of smaller but still hefty current compensation inductors. If your filter's don't have all these coils, they are highly likely to be ineffective.

Another quick check is: Is the buzzing device connected in such a way that a

1 ground loop/s is created?

Plug in to same outlets as the source/ related gear is usually the quickest elimination.

2 a signal loop is created? same as above.

Metal clad shielding at source is the way to go for many of the irritants described in this thread.
The right capacitors on switches to stop arcing helps too; spark quenchers.

Three phase Passive harmonic filters can be helpful in the correct application to reduce harmonic distortion to motors.

"Triplins harmonics" is the correct term for the high frequency distortions that can raise the apparent AC voltage in the mains supply. This is what is on the increase.

[Alec Mackay]

Thanks for the insights.

The studio does not have significant mains filtering, as we operate largely in the digital domain. I am certain it's radiated interference not a noisy power supply, since the issue is significantly worse when moving closer to the elevators.

I've tried shielded cabling, but it was inconclusive. I haven't tried with a long enough length to get from the switch close enough to the elevator to rule that out. Our installation is all unshielded CAT6.

I should clarify that we do not use fluorescent lighting, and all primary LED lighting is on a separate phase. This has never been an issue for us.

The cheap LED light strips I used in my tests were just the most noticeable and consistent noise source I had available, with quite ugly switchmode supplies (they even emit an acoustically audible high frequency noise).
I have probably skewed my results by testing with them, because of their shared electrical connection with the device in question. I will need to repeat my tests with the problematic elevator motor controller noise.

[Acoustic Landscape]

There is a good deal of crossover in my last post, in that one thing leads to another, hence 'kill it at source' approach. The mains today, from even fifteen years ago is completely different, compromised.

The first entry at the top of a list of affected devices in this main noise pollution, are unquestionably digital devices, being the most susceptible to main interference, particularly if they don't have a well designed PSU, most of them won't as the filters they are obliged to fit to their PSUs to stop injecting triplins back into mains only satisfy 'manufacturers codes'.

The lower 'standard' Mains in line filters on the market start filtering from 100kHz, which is no longer good enough.

Medics, aerospace and mil markets design for a lower f, essentially low bandpass filters for incoming AC below 10Khz. Much of this is to do with reliability. Audio is too niche to be a standalone category but is definitely in parallel with those markets requirements.

LED smps "they even emit an acoustically audible high frequency noise" Yup, thats the 'chopper' chips oscillating.

"The cheap LED light strips I used in my tests were just the most noticeable and consistent noise source I had available, with quite ugly switchmode supplies (they even emit an acoustically audible high frequency noise)."
...a good test.

more lateral thinking required!

In such inline mains filters circuit it's very important to have the first current compensating inductor (thats connected to the mains) have a capacitor from each output (load side) of that inductor that joins to a single inductor, that goes to earth.
Most of them don't have this and makes a huge difference for digital devices.

A recipe for clear reliable audio of hard won info worth £K's ! above

One caveat is that every mains supply/ building / equipment combination is unique and some of the dozen or so know general AC mains supply problems can be more elevated in some factors than others. For example: three phase supplys will be exposed to heavier switching spikes, industrial, large SMPS up line can also have huge effects. Making sure the phases 'power draw' are balanced is another healthy observation.

[molecular]

Hi Alec,

this is a blast from the past: I'm afraid I'm mainly only here to report that I never solved the problem, and have moved on from that studio since 2019.

As far as I can remember, using the X2P as a local device (so far example, simply taking a laptop into the booth and setting up in there) was the only thing that guaranteed a clean signal.

The situation did, however, improve over time and the interference became less instrusive so I *think* my conclusion was that the wider Cat6 network was part of the issue in a part of the building I couldn't do anything about, but that maybe a piece of particularly problematic kit or infrastructure had been moved elsewhere when it improved.

The building was a college building so honestly it could have been anything from telecomms to a washing machine. The only thing that I knew for sure about the space we were set up in was that it used to be used as a TV skills teaching studio, so potentially had some weird extant power infrastructure or something.

Anyway, I was never able to replicate the issue with anything other than the X2P... not even other Rednet or other AoIP devices.

Sorry not to be any help, but they really should have sorted this out by now (7 years!?)

HM