Our great mains

[PeterSt]

Hi all,

Throughout times I have been posting about "noise" problems I experience once in a while. This is another such post ...

All is about "it can be happening to you just the same", but you wouldn't know it. I mean, I'm in the position of having an analyser at hand, so I will know instantly. Or at least I can check things a normal human being wouldn't be able to. He would just perceive lousy sound.

So, I have been away for 3 weeks (back now), and when I got home of course the first thing I did was switching on the amps and stuff. Heat it all up, and at last listening to decent sound again. But it wasn't there;
Since my amps need some 24 hours to get normal after being off for a day (let alone 3 weeks), I thought it would be that after listening the first night. But it wasn't.

At working on some "DAC ideas" the next day, I noticed great distortion from measurements, but since this was about this "new idea" I wasted a day on getting that right, while in the end the normal situation measured as bad. Hmm ... what happened ?

Well, to be brief on that, at this moment I still don't know. All I know is that some huge noise is coming from the mains (or through the air, I can't decide for that at this moment), and using protective earth (PE) makes it the worst (what about noise being there up to -40dB !!).

2 days after being back I once again noticed one of my devices being off, while I didn't recall to have shut it off. This time I looked more close, and saw it was still switched on, though not working (this is my FireFace800). Aha, so that must be something to keep in mind. But again, what happened ? Did that device die by accident, or has something happened during my being away to it was killed (lightning perhaps -> must ask the neighbours), or did that device die in a dangerous fashion and something else is destroyed without me knowing yet ?

In any case, something must have happend while being away, and now I'm up to the task of finding out what is the problem.
I have talked before - in between the lines - how much nasty stuff is going on anyway within our mains, but I never talked about it in concrete because I first need to know *all* the ins and outs about it. But you can bet there's much more going on than all of us know. I mean, I asked around amongst some other engineers I know, and nobody has a clue about what I all experience, once some base "problems" can be mimiced, and worked out further to extensions you can get crazy about. For now I give you one example, only because it is happening to me right at this moment (and this wasn't so before) :

I plug my analyser in mains without PE (otherwise it will measure noise natively because it receives noise itself);

I measure the analogue outputs of the NOS1, and it is way too noisy (something like -80dB while it should be at -113dB or so in my situation). The NOS1 is NOT connected to the means at all. Notice that this is dangerous, because in this situation all is out of (ground) control, so not connecting it to the mains may incur for MORE noise (think about that one).

I plug the NOS1 into mains again without PE; There's the On/Off switch which has double poles. So, hot and neutral are both switched off.
In this situation there's not even connected anything towards the DAC itself. So, behind the mains inlet are the transformers, and behind the transformers are the PSU's. But, the PSU's are not connected to *anything* (all is cut). And remember, since the mains switch is off too, and the PE is not connected, there shouldn't be connected anything, right ? So, the cable is in, and the only thing which connects a little, is the PE from the cable towards the all outlet. There it stops, because there's no PE in that outlet.
Well, only putting in the cable makes the noise go way up (to really unacceptable levels).

Thus, there's a PE wire in the cable (cable is 2 meters or so long), that PE wire is connected to the case, and the case connects to 100% nothing in my situation (all ground inside is floating). The interlinks out of the DAC are connected to the analyser, so what happens from there (the analyser) is unknown. In any event, plugging in that mains cable to the DAC will let rise the noise significantly.
In my view this can only be something picked up from the air. It should be a combination of the PE wire in the mains cable acting as an antenna on one side, and the Faraday construction of the case doing something on the other side (the latter in a kind of upside down behaviour ?).
And if I, in the same situation, put the cable in a PE connected outlet, the noise goes up to -40dB.

What I expect at this time, is that something very nasty is going on in my mains to begin with (some device shortcutting to PE ?), and that this by itself radiates through air as hell. The FFT of both situations (mains cable connected with and without PE) looks exactly the same, but the former at a 40dB or so higher level (now the noise being connected directly).

And about problems people may have ...
This is so bad, that turning an output level knob of an active crossover I have sitting in my chain somewhere, makes stop working the soundcard in the DAC ... (when all is normally working and connected, though without PE in this case).

For now the message is : there is no reason such a thing won't be happening to you. But how will you know without something like an analyser. You won't. You will have bad sound though.
Also, one day things may be all right, but a next day all will be wrong. That too will go unnoticed;
Or like in my situation, something happens while you're not there. Sound is suddenly bad, but you will be without clue forever, because you don't pocess discrete measurements from before. Maybe 7 years later when you replace your TV for a new one, suddenly all is right again.

I think I said it before ... part of the base equipment of any self-respecting audiophool should at least be a scope ...

Peter

[Nick]

Peter hi,

Very interesting, I certainly have system "bad days" for sound....

One question, can you use your analyzer to do an FFT on the mains 60hz supply ? Perhaps setup a resistor potential divider to pick up the single level you need for the analyzer.  It would be very interesting to see a post of what the spectrum looks like. Bet its not a nice sharp spike at 60hz.

I'm in the car now but I'm going to have a careful read through your notes later.

Nick,

Ps change number 4 on the NOS1 is running now, will pm you with the details

[PeterSt]

Hi Nick - somehow I expected you to respond here ... 

I'm not sure whether you're talking about the 60Hz "possible" output of my 230-110V transformer ...
"Possible" because I expect it to output just 50Hz (same like the input).

Actually I never did an FFT of the 230-50Hz ... maybe because I don't want to know and sleep better.
But somehow I expect the sine not to be a sine at all, and merely have it mixed with the other phase (attenuated/leaked through / shifted a little perhaps).

Peter


PS: Tweaks 1.5, 2 and 6 have been applied, but no measurement results yet because the picked up noise has the overhead at this moment.

[Nick]

Hi Nick - somehow I expected you to respond here ... 

I'm not sure whether you're talking about the 60Hz "possible" output of my 230-110V transformer ...
"Possible" because I expect it to output just 50Hz (same like the input)

Peter


PS: Tweaks 1.5, 2 and 6 have been applied, but no measurement results yet because the picked up noise has the overhead at this moment.

60 Hz doh! 

Tweeks 1.5 and 6 sound interesting as I have only 1 to 4. hope all is sounding good even if they cannot be measured 

Nick.

[Gerard]

For now the message is : there is no reason such a thing won't be happening to you. But how will you know without something like an analyser. You won't. You will have bad sound though.

Hi Peter back in holland!

That is not what i expect to hear from you after your holiday  especially after Nick his findings. Is it you who is static loaded 

How was your holiday? Some inside info before you are totaly in the NOS1 stuff You survived a tornado 

 

[manisandher]

Hi Peter, thanks for sharing.

Since completing my dedicated mains line, I've become convinced that the mains is the single most important element contributing towards the sound. However, as you said, it's complicated and simply buying a mains filter, isolator and/or regenerator may not be the answer. Like you, my experience is that there seems to be a substantial air-borne component also.

I think I said it before ... part of the base equipment of any self-respecting audiophool should at least be a scope ...

Agree 100%. I'd like to buy one - do you have any recommendations?

Mani.

[PeterSt]

Well, today I spent on finding the culprit, and I found ... nothing.

The below is a nice 50Hz picked up through air;

Analyser is not connected to PE, analysing PC is on batteries, DAC is switched off but PE-connected mains cable is in (see earlier description about the latter). Interlinks are XLR connected to the DAC, but in-DAC there's no connection at this time (to make this as clean as possible so nothing else can pick up (but since the interlinks are "open" at the DAC side, this is of course not the best idea for "no noise", but it really doesn't matter because of this overwhelming noise)). SE connected is even much worse.
The Audio PC is not connected and Off.
This 2mV compares to the normally 6 uV.

Without PE connection there's no problem at all.

I wish I had measured my mains before, for reference. Now I can only tell that this morning it was 217V (seems way low to me !!), and that the same voltage is in between phase and PE. Between neutral and PE there's a varying 200mV and 650mV or so (it is one or the other !).
The FFT (mind you, all picked up through air) shows a very regular pattern of peaks at certain (higher level) frequencies, that dropping, and right after that another higher frequency peak briefly arising and dropping back again. This seems to be happening at a simlar interval I see the beforementioned 200mV and 650mV alternating (each ~15 seconds).
Also there's a steady DC of 474mV between neutral and PE, 20-24mV between neutral and phase, and 26-30mV between phase and PE (both vary at the same time with a distance of 6mV amongst them).

I have no normal means anymore to get rid of the normal DAC-output DC offset values (which always have been 4-7mV), and *if* I succeed, it may go way up by itself later (but "by itself" seems to be switching off the TV or something).

[Nick]

Peter hi,

I plug the NOS1 into mains again without PE; There's the On/Off switch which has double poles. So, hot and neutral are both switched off.
In this situation there's not even connected anything towards the DAC itself. So, behind the mains inlet are the transformers, and behind the transformers are the PSU's. But, the PSU's are not connected to *anything* (all is cut). And remember, since the mains switch is off too, and the PE is not connected, there shouldn't be connected anything, right ? So, the cable is in, and the only thing which connects a little, is the PE from the cable towards the all outlet. There it stops, because there's no PE in that outlet.
Well, only putting in the cable makes the noise go way up (to really unacceptable levels).

Thus, there's a PE wire in the cable (cable is 2 meters or so long), that PE wire is connected to the case, and the case connects to 100% nothing in my situation (all ground inside is floating).

One idea but mainly questions.

Reading you first post the only connection from inside the DAC to the outside is the PE which goes from the DAC power socket into the mains cable but is not connected to the household PE at the wall socket.

Could the 2 meters of PE wire within the mains lead running parallel with the live and neutral leads be coupling the noise. Reading a few papers recently (as you know) there may be capacitive noise coupling between the live / neutral and PE wires within the mains cable. There could be more coupling here to mains noise than say from air borne RF. Could this be injecting the noise ?

• Does the noise still happen with no mains cable connected to the DAC ?
• What happens to the DAC noise if you connect a cable in the same way but with no PE inside the cable (ie just a two core live and neutral cable)?
• If the 3 core mains cable length is varied (eg the PE to live parallel run is varied in length) does this inject more / less measured noise ?
  

217v does sound like a lot of sag its more than 10%, could there be something on your distribution spur (external to your house) that is dragging down the supply voltage and injecting supply noise ?

Nick.

ps Is the plot your mains wave form or an air bourn noise trace I am not clear ?

[PeterSt]

Hi there Nick,
 

Reading you first post the only connection from inside the DAC to the outside is the PE which goes from the DAC power socket into the mains cable but is not connected to the household PE at the wall socket.

In all cases the PE wire is in the cable, and connected to the DAC case (I never tried detaching it from the DAC case because I'm fairly sure no problem would be there then). So, the problem emerges when the PE wire is connected to the mains PE in the wall outlet.
 

Reading a few papers recently (as you know) there may be capacitive noise coupling between the live / neutral and PE wires within the mains cable. There could be more coupling here to mains noise than say from air borne RF. Could this be injecting the noise ?

Theoretically, yes. So, the phase and neutral in the cable (formed in a triangle fashion in my cace) could project something on the PE wire (remember, the neutral and phase are not connected in the DAC (double pole switch is off) but both are connected to the wall outlet).

Does the noise still happen with no mains cable connected to the DAC ?

No.
(the answer really is a little "yes", but too long winded for here, and not relevant; anyway what then happens is just normal).

What happens to the DAC noise if you connect a cable in the same way but with no PE inside the cable (ie just a two core live and neutral cable)?

I did not try this, but I can't imagine anything happening. Remember, this is to be compared with the not PE connected cable and switch Off. Then nothing bad happens either. So, you may want to compare it with switch On and see whether something which is OK (because it is in that case) then suddenly is not ?

If the 3 core mains cable length is varied (eg the PE to live parallel run is varied in length) does this inject more / less measured noise ?

Didn't try that either, but I can imagine so. On the other hand, this will be similar to being able to measure DC with all off (neutral and phase cut by the switch) and cable connected to PE (which actually is another story from longer ago). At first I always thought this was capacitance in the (PE) cable or something (and the length indeed may vary the capacitance), but today I think this is about a ground being able to create (DC) voltage from a Faraday cage which latter acts like a (magnetic) collector or something. Put your hand in it (touch nothing) and the voltage goes away.

217v does sound like a lot of sag its more than 10%, could there be something on your distribution spur (external to your house) that is dragging down the supply voltage and injecting supply noise ?

All I know is that I have been measuring in the past and for sure saw more than 230V, if not in the 240's range ...
Whether there's something dragging on "my" mains is to investigate, but it is not what I expect (btw, I live in the country side). What I merely expect is out of phase stuff. So, very simplified (because the calculation goes different) : when we'd have 3 phases (and I do have three coming into my house) two of them creating the 240, but they are not properly alined, then all individual phases are still 120 (hence no exceptional drag) but my received voltage is less (but the two 180 degrees out of phase and 0 will remain).
In the mean time this will create "noise" as such, because no sine will result from it.
How this relates to the PE is something else, and depending on how the mains is setup in my situation (outside of the house and yet to find out), PE *can* be a reference for the phases, so it can be influenced as well. But this is not necessarily true at all, because it can work without PE reference as a floating system. But but but ... there you have it again ... if the floating system is not in balance (degrees deviation) the presented PE is not 0V opposed to ... my earth pin which I use for audio, but not for the PC.

You see ? I must stop (or shouldn't start) because there are tons to investigate first.

ps Is the plot your mains wave form or an air bourn noise trace I am not clear ?

As per the reasoning of my previous post, it must be through air. Or in either case it is not a direct connection to the mains (then you would have seen that 217V, while this is 2mV).
Let's also keep in mind that what we call "noise" here, is totally inaudible through my system as noise as such (it's completely quiet). But, the system sounds like sh*t, because everything rides on it. For the DAC all is fine when not using PE (remember, at this time it's not even used and only connected to the case !), but for the amps it still will. So, tonight I will connect those without PE, but it's a little "dangerous" to just try it without measuring *those*, with again no reference for me, because I never measured them.

Peter

[Per]

Hi Peter, thanks for sharing.

I think I said it before ... part of the base equipment of any self-respecting audiophool should at least be a scope ...

Agree 100%. I'd like to buy one - do you have any recommendations?

Mani.

Hi Mani and others, If you consider an AC regenerator / monitor, maybe this would pass the bill as it is an improved version of the older powerplants AND - from what I understand from Paul McGowans newsletters - a true scope built in... Don't know anything about the price, though...

http://www.psaudio.com/ps/products/detail/perfectwave-power-plant-5?cat=power

Per

[manisandher]

Hi Per, I think I'd prefer the flexibility of a separate scope. Also IIRC, the new PS Audio P5 and P10 units are quite expensive, >$4000. I've never bought a scope before, but I suspect you could get quite a nice one for this sort of price.

Of course, a scope wouldn't clean the mains up for you... But having had a PS Audio P300 (still have) and P600 (sold a few years ago) and also a PurePower 2000 (still have), I'm not convinced that a pure sinewave is necessary to get the best SQ. Indeed, on their older products, PS Audio offered 'multisine' functions which deliberately flattened the top of the sine wave, in order to improve the SQ.

I haven't given Peter's first post in this thread enough time to analyse and fully understand, but it seems to me that the earthing/grounding system is key. My understanding is that no filters, isolators or regenerators are permitted to manipulate the earthing/grounding system, for safety reasons. My hypothesis is that it's not just earth/ground loops that we have to worry about, but also 'noise' getting transmitted from one component to another via the earth/ground connection. This situation is exacerbated by using a PE which is shared with the rest of your house... and street! I was first alerted to this by the work of Arthur S. Kelm at 'Ground One' (http://www.ground1.com) who seems to be pretty much the expert in this field. If you read his site, you'll notice that I've adopted much of his approach and am very happy with the result. But I think air-borne 'noise' still needs to be addressed, ehich he doesn't seem to cover.

I'm still interested in hearing recommendations for a good scope...

Mani.

Mani.

[Per]

I think you are absolutely right, Mani. Ground is very important. Some deliberately make ground wires thicker than signal / hot wires - both in a powercord and inside cabinets.

Regarding seperate scopes I have heard praise for the Tektronix scopes over at the Naktalk (Nakamichi tapedeck) forum. If I am right you won't need one in the multichannel / gigahertz range - a two channel megahertz scope will be enough for audio - even to monitor for instance a S/PDIF squarewave signal in the low megahertz range.

I bought a used 2-channel 20 mhz HP scope back in 2006 for 1000 DKK (around 120 pounds) but never got to (learn how to) use it due to a worsening in my muscle disease.

Wish you a good hunt for a cheap used scope or, maybe a handheld one or a PCI card that uses an older PC as screen and recorder.

Best Regards,

Per

PS: Please take VERY much care measuring the mains - at 110 / 230 volts ac it takes only mA to kill you if goes left arm - right leg...

[PeterSt]

Mani,

I didn't reply to your scope question yet, because it is not so easy for me to judge this. All I know is this :

My second hand analogue scope is 20MHz, and it is just sufficient for even 40MHz range measurements, if only the 10x "magnifier" is used and you'd know in advance what the results should be. I'm talking audio clock levels here, but which you normally wouldn't need.
So, 20MHz is sufficient for sure ?

No, sadly for what we're talking about here, it is not.

With 20MHz, unveiling normal noise levels already is quite difficult. Normal for NOS1 output stuff would be in the us range, while this is something like mv range. With my 20MHz you may see a somewhat thicker signal line, but nothing like "moving" and certainly not for analysing the wave shape.

If we'd take the picture in this post : http://www.phasure.com/index.php?topic=1559.msg15810#msg15810
as an example of something you'd want to know, you'll see that this needs a resolution on the X-axis of around 0.2ms and something like 10us or better on the Y-axis. So actually that are your requirements ...

Peter

[Scroobius]

Hi Peter

Not sure I can help much on this but it seems difficult to believe this could be a problem caused by anything within your home. Mains droop down to 217V would be of big concern here in the UK and I am sure it must be in NL. I would be on the phone to the electrical supply company right away. I have come across peculiar ongoing supply problems in the past when I worked for an IT hosting company and when we took it up with the electricity supply company they found problems with partial transformer failure in the local sub station. In my experience Electrical companies take reported supply problems very seriously.

Just out of interest is your house earth bonded to the incoming supply cable earth or (if you are remote with an old supply system) do you have an earth "spike" sunk into the ground (this is sometimes the case in the UK with older properties). Also do you any friendly neighbours where you could make some similar measurements in their house?.

Hope you get it sorted soon as days without NOS1 are not as good as days with NOS1 as I have come to learn.

All the best

Paul C

[PeterSt]

Ok, a couple of days further I seem to have tackled my problem. The story is too long (and maybe too unsure at this moment) to lay it all out, and FYI I just reported what happened in 3 or for pages to an electrical engineer helping me with this.
But what came from it, is a little easy to apply trick to find out whether your SQ can be changed for the better. Or at least this is something which *can* happen to everybody, if not that it *will* happen, depending on your particular situation. I only use an anlyser to check for te truth of this little trick; what you need is a volt meter (230V) ...

Let me first simply "state" that having more mains groups around the house is prone to prolems.
This, while we thought doing good with that. So, very generally, and a bit speculative for now, is that more of these mains groups - meaning each with their own Protective Earth (PE) pin or otherwise with different PE's, may or *will* create a potential difference between those two PE's (potential for our dutch readers in this case means "pontential to ground" (and not potential as in "possibly" ).
This does not seem to be an inherent thing (meaning : without anything connected to your means the difference will just be 0V - so no problem), but it is created by the devices you connect. Example :

Put the PC in one mains group, and put the audio devices in the other, thinking that the PC will spit its noise into its own group only. Aren't we all thinking just *that* ?? I think we do.
But for simplicity, let's think that whatever it spits out, exhibits as the voltage difference between the two groups. And *that* is noise, because the noise now can't go anywhere (please leave this latter be for now, and let it come across as something logical to you).

So, before I attempt too many explanations for phenomena I really can't at this moment, let's go and measure our differences in potential between *all* the groups related to audio. So, look at really *all* the devices you use, even if it's a liniar PSU for your USB connected device. Think about how all interconnects (via-via-via), and of course explore the different mains groups you use for it. Remember, groups with different PE (pin, etc.).
Hook up your multimeter (ready for 230V) to the earth pins of both groups (just find two outlets, each of the other group), and look what you've got. Remember, have *all* devices and interlinks connected as should !!

If you see more than 0V, you can improve.
If you see more than 0V, an analyser will show something like you saw in the earlier picture in this topic ...

If you have the problem, immediately stop trying to solve it by whatever commonly known means, because you probably will make it worse (and you can't see that without an analyser), nevermind you end up with 0V in the end.
Instead, move all your noise spitting sh*t to this one group of which you think it is made for it anyway. Remember, 100% everything.

It also works the other way around :
You see 0V, but this may well be because the problem has been solved the wrong way. Example : there is a USB connection between the PC and your audio chain, and the "solution" to it now will travel over this connection. Mind you, I am *not* saying this is really bad, but it isn't solved the way it should. This noise should travel over PE, and now it doesn't.
So what does this mean ?

If you use different (PE) groups but see 0V, work the other way around. Start DISconnecting stuff like USB and actually everything like interlinks, and wait for the voltage to go up. If it does at disconnecting something, you may think about what the connection actually will do for you, knowing that it actually carries away noise. Mind you, this is not the proper expression, but let's all agree that PE is for this, and you just prooved to yourself that this something else is used.
So ... if you see the voltage go up, again think like I explained before, and better move all to this same group. And yes, the first thing prone is the PC ...


Again, the situation needs too much writing to explain in full, and is super complex at the same time. The past week I have been trying to find people who could explain what is happening, and although still someone may be able to, at this moment I brought some "elements" together from quite some different stories and ideas, knowing that at this moment no single person came up with the real idea. I only want to say, these are my own findings / conclusions / trials of indivudual elements, and it may all appear total BS in the end. But no matter what, you can have the problem without being able to know it, and you really will find exactly nobody coming up with an analyser knowing what to do. So, at least I can in my own house, and this came from it. And mind you, I will give the example of my situtation right now, and keep in mind that picture ! :

Yesterday, in a kind of random situation - as much plugged out as possible, sutting down all the groups etc., I measured 67V over the two PE's of my groups involved.
Knowing this, this morning I started experimenting a bit, and this time all was connected again, and derived from the whole subject of this topic I solved my problem (already yesterday) I measured 0.00V. I didn't know that yet, and had the idea the 67V would still be there.
So, problem solved, and 0.00V now. Right.
Thus I started "phase two" as described above, and for your real life example - at removing an USB connection which via-via is linked to the audio stuff in the other group, I now saw ... 99V. And for those who know the subject, I also immediately got crazy of my PWM regulated pump for the heating system, which square waved exhibits its peaks at -40dB in a bad case.
No audible noise from the speakers (except for that 17.8KHz tone), and if I hadn't had that pump, I wouldn't notice it at all, except for most probably bad sound. But of course the analyser would show *that* picture again ...

At moving the PC to the audio group, nothing happens. 0.00V no matter that USB (or anything else like the DVI cable from the NOS1) connected or not. 0.00V stays forever, and the analyser will show normal 7uV or so noise levels.

Without further explanation, I know that this is not pure PC related. I mean, I had my subwoofers in again another PE audio group, and there too the voltage would occur.

And thus it is my conclusion that however it goes and travels, no different devices are allowed to interconnect once connected to different PE mains groups. Each group of such devices will create their own potential, project in onto PE, and the two (or more) PE's will create a new potential opposed to eachother, and that now can't go anywhere.

It is good to undertstand that our houses are full of noise. Yes, we knew that allright. But what it really is about is how to move out that noise out of our systems. This is not about cancelling the noise, but move it out over PE (tough to explain for me ). The ever example is my PWM pump, which really won't produce less of its square wave noise onto my mains if I punt some little gnd cable somewhere in order to let go away that noise. It helps, put the pump keeps on spitting;
When I have those differences in potential over the PE's, that noise won't go to earth, but it will go into my system instead. It will start to travel over the neutral or phase (whatever), as will any induced noise. Also think about the voltage which normally will be there between phase and PE, and you can see how it can sneek into your system, if PE doesn't move it to earth via the shortest path. Now, it will go from the one PE to the other (read : via your devices) because that path is shorter (distance between your devices is a few meters, while distance to earth may be 20 meters). Something like that.

Disconnecting all your devices from PE most probably will create hum, because now *really* all is travelling over the normal wires (neutral and phase), directly. All is "floating", but each device will create its own potential and all will be in unbalance, PLUS switching Off / On at some time may create new ground paths opposed to before (hey, I did nothing, but now it all sounds sh*t !).


Please try this out (only if you have different PE groups), and I urge you to let know the results in this topic; As you may have learned from some in between the lines texts from me throughout time, I'm really heading for working this all out at last, while nobody really seems to know. This is one step of that, although one month back I didn't now about it at all. So, there is much more out of this little teasing subject, but in the mean time this may be the most important one (oh, did I ask you to look at that picture ? ).

I guess this also puts the "isolated from the PC" in another perspective;
I don't know what others (manufacturers) are doing with this and how commercial this stuff all is, but

a. as per my example you just will create the problem (but use different PE mains groups);
b. I really can't see by measurement that a PC spits noise over its gnd connection.

Ad b.
But it really should when just that connection (think USB for your conveniency) solves the problem (described above, and when the voltage goes to zero at connecting it), and :
Ad a.
Without that connection (now meaning : connected by galvanically isolated means), you *will* have the problem (but won't see the analyser picture).

Well, again long ...
Let me know your findings please !
Peter

[listening]

That's a very interesting discussion! The sound engineers in the recording studios must be faced with the same problems (depending on mains?). How do they asure the correct environment?

Georg

[listening]

That's a very interesting discussion! The sound engineers in the recording studios must be faced with the same problems (depending on mains?). How do they asure the correct environment?

As I have seen in Articles the mains supply is very good planned. At first they are using the three mains phases for different electronic devices: The first for audio-devices, the second for video and the third for "ordinary" devices. Then the voltage is stabilized by a mechanic or magnetic stabilizer and at least an active filter ist used sometimes supplemented by a passive filter additionally.

Georg

[Nick]

Peters findings have helped here.

I was defiantly suffering from “bad day sound syndrome”. Sound could be very different (relative to what I know could be delivered), to the point that I would try to listen and then just turn off and come back later to see if things had improved. The last couple of weeks things have been stuck in a rut with poor sound most of the time...very frustrating  .

Almost all of the system has been powered until now from one double wall socket, with only one of the active subwoofers being powered from a wall socket across the room. At the wall outlet both sockets were in use with a couple of 4 gang extension leads in series to get reach and sockets to the right locations. The PC, Monitor, DAC, Amplifier, second active sub and a linear power were all plugged in sort of randomly depending on what was being done at the time. This is not quite what Peter is describing I think, if I understand correctly in Peters house there may be use of PE in separate ring mains / wall sockets / supply phases ? Anyway there were some relative earth voltages between components in the range 0 to 15mv.

Out came the old 4 socket extensions and everything was plugged into a single 8 socket extension using only one of the wall socket points. The active sub on the other side of the room still uses a separate socket but this does not seem to drive additional PE voltages. The result is that the relative earth potentials have dropped to between 0 and 0.5mv, and the sound seems back to normal    I have not done many system power cycles so it may be that bad day sound comes back but there was certainly an easily detectable change when the 8 socket extension was connected.

Thanks Peter for posting your thoughts and findings it has made quite a difference, I hope it proves to be a long term solution to bad day sound.

Nick.

[Scroobius]

Hi Peter

Very interesting. My PC is on the main house ring (with the fridge the TV etc) and all the audio devices are on a separate dedicated spur back to the main distribution board. The idea as you suggest was that the "PC could spit its earth rubbish" onto the house ring. So I got my DVM out and with the PC running and the audio devices also running I measured the voltage between the house ring PE and the dedicated HIFI spur PE. In my case the voltage measured was 0 volts. I have short PE runs back to the dis box and recently much of the wiring was changed in the main house ring main so maybe mine is not a typical example.

But I think that you have a very good point clearly (now you have pointed it out) connecting devices to different mains groups (rings) does not appear to comply with the requirement of star earth requirements where everything must connect to a single point. In fact a normal house ring main where the PE connects to everything and then two connections at the distribution board does not seem to be a good idea in respect of star earth requirements. So a dedicated hifi spur (one PE connection ie no ring) looks an even better idea now after your post. I have moved my PC over to the HIFI spur now anyway just to be on the "safe side".

All the best

Paul C

[GerardA]

When everybody takes there DVM's out, then another small trick is to test the best way to put your plugs in.
Remove all connections from other equipment, measure voltage between safety-earth and mass of one pice of equipemt, turn the wallplug around and measure again. (AC and DC)
The lowest measured value is the best, do this for all pieces of your equipent and put a small piece of red tape on the correct side of the connectors and the wallplugs.

PS, But why Peter is your voltage so low? That does not sound like an earthing problem...

[manisandher]

Peter's findings are exactly the reason I went for a 'star earthing' system  (see http://www.phasure.com/index.php?topic=1445.0) - all components (including the PC) have 0V between them. HOWEVER, there are a couple of things that I feel need to be done over and above this for the best SQ:

1. The earthing system should have an ultra-low impedance to ground. This may save your life some day. But apart from this, any noise generated by components (think switching power supplies, etc) will have an easy way to ground, and therefore be less 'tempted' to travel across interconnects to get dissipated.

2. The earth used for the music system should be isolated from that used by the rest of the house. Now, this poses a real problem, as the Neutral is bonded to the earth at the inlet to the building (in the UK at least). The only solution I know of is to use an isolation transformer.

Mani.

[BertD]

The lowest measured value is the best, do this for all pieces of your equipent and put a small piece of red tape on the correct side of the connectors and the wallplugs.

Thanks for the reminder (this IS important!) but how is this done when the signal ground is connected to the mains ground (earth)? Use a socked without earth? How about safety then?

I am asking this because there is a lot of equipment on the market where the signal ground is connected to earth (usually the housing of the equipment which is then connected to the mains earth). Manufacturers do this to bring their own noise down, the most easy way to reduce hum...

The trick above works on equipment that is double isolated (i.e. power cords without earth connection) but how about the other equipment like the NOS1? The NOS1 is connected through earth through the mains cable but also through the DVI-cable passing the PC and then connected to earth again...but also the signal ground is connected to earth?

Peter, why is that? To my idea the signal ground should never be directly connected to the mains ground (earth) in the first place? Mains earth is for safety only...and it is a free ticket for noise present on earth to infiltrate the signal ground, isn't that the problem you have besides the Voltage changes (combined with DC perhaps too)?

Bert

[PeterSt]

Bert,

Actually you pin pointed the most important "phenomenon" of it all : how to get independent from the other equipment, at the same time not producing noise yourself. At least that is what I always had in mind with the ground the design of the NOS1. Also, I had some great examples, and you may recall ... your own amplifiers. Remember ? (mind you please, this is not at all about someone's ground design, and especially not yours for this "example" ... and instead it is about the necessary required independance from the other equipment). So. I'm referring to your "within the blink of an eye" cutting the signal to PE wire in ONE of the amps, just to avoid some perceived ground loop, at testing a signal with just one of the amps, the other one being open and out of reach for testing.
And so indeed your both amps (4 actually) are still the example for at least me and the NOS1 to cover for all, just because they create the problems. ... Which is good, because anyone can create equipment like that.
And for you fun and background : in my momentary setup (actually all being well again at the NOS1 side) I only yesterday evening had to move your amps from PE connected to not PE connected, because now I have a sizzle through the speakers otherwise ...
(btw, awaitening the final solution for this new "mains setup")

So ...
So the problem is exactly about the necessity for a signal ground reference which theoretically can come from a connected chain where all is creating a floating ground which is equal in each device ... but which is rarely the case. Btw, the NOS1 does this, but if e.g. my active crossover does not (you know what I'm referring to) current will be flowing from the one equipment case to the other, and the transport means will be the shield of the interconnect ...
Not good.

Having said this, the NOS1 is virtually (!) floating, unless I missed something somewhere (see below for further explanation).
Unless I overlooked something or made an error in the mean time (during real production). And indeed I have the example of that :
Those without the little volt meters inside (like yours), have an error in the connection of the display, which is not isolated from the case, that creating a strange (and unwanted) direct connection via the PC over USB when the display is connected. However, since the display is not in use yet, I let it be so far since the solution is super simple (connect the display (3 points) with nylon washers).

But there is more, and it is not all *that* easy ...

The version without the volt meters have general ground connected to the case. This is not explicitly signal ground, but the ground planes from the PCBs (not PSUs). And, since signal ground obviously come from the PCBs (ground planes) at some stage, signal ground *is* connected to the case, while the case is connected to PE. And so ... so you are right afterall, but not litterally spoken and certainly not if it were about how ground paths should go where, and whether signal could pickup current flowing through the case. And not to forget, I measure, and all measures is just fine.

The NOS1's with litte voltmeters are 100% floating which -I think- works out for the better if only people have their further devices in the chain in order, which could be something to anticipate upon, and in the end is available as an official upgrade, although I never talked about it publically and I rather (let) apply that upgrade for people with noise etc. problems (but reports about that are zero so far).

Again :

Manufacturers do this to bring their own noise down, the most easy way to reduce hum...

I think this is may be true, but at least for me it didn't work like that, and here too it is coincidentally you who indirectly pointed out how useless this is. Well, in theory. This is about my ever back email that I did my stinking best to let all inherently work with good measurement results, but as soon as all was connected to the main amps, the noise was all over because the main amps had signal ground connected to PE, BUT the main amps were not PE connected, so it lacked the proper reference (and ground potential difference). So, if you think back about this, now you know why I indirectly referenced to PE myself in the first batch of DACs - because otherwise people with amps like mine would have a problem.
(the story is a bit longer, but I'm sure you get the grasp of it)

The NOS1 is connected through earth through the mains cable but also through the DVI-cable passing the PC and then connected to earth again

IMO this is the most normal within itself;
Not thinking about the signal ground connected to PE in this case (which actually is unrelated in this context, and which for the later NOS1's indeed is the situation), this connects the shield of the DVI cabe from actually the PC case to the DAC case. If this hadn't been done like this, the ground wire within the DVI cable would have been used instead for this, so the noise from the PC would be carried over the "DVI signal".
There's one little annoyance with this (well, for me), and that is that in-DAC this shield signal is connected to the PCI ground plane and from there it's the indirect-to-signal-ground situation again (which I rather had avoided, but which is impossible). Anyway, here too the later NOS1's have complete isolation from the DAC case, so that too is not connected to PE now, so really nothing is connected to the case, except PE itself.
Btw, since this is my momentary situation, this setup perfectly allows for not have the case PE connected (but the PC is), which exactly was my "solution" to my sudden noise problems (but amps (everything) has to be disconnected from PE now), and at this completely floating situation at least works (by now worked) for me as the solution. But mind you, this already stops working when the NOS1 output is made floating by itself, like it would be the situation with output transformers (which again isolate). So, here it would take a direct connection from the primary common to the secundary common (which is "created" ground), or otherwise I could not find a solution without PE (and my amps).
See ? it is not all *that* simple ...


The whole situation seems a lot more complicated to me than we usually think about it, which mainly is because of the dependancy from other devices out of our control. But also because of the reactance of interconnects with other devices up to the sheer possibilty of measuring 5V USB between (PE connected) case ground and signal ground, with 0% connection of any ground anywhere (no DVI cable, no interconnects), USB ground cut. In this situation the ground flows right through the air, although it will be inductance impeeded (think transformers), so things even happen which we can't physically see. This by itself is about a case easily being able to act like an antenna, only because PE is floating so to speak, while in the mean time (magnetic) inductance is put on the case by means of the content. Ehm, better connect that content then ... which means connect signal ground to PE (directly or indirectly).

Earlier I found some paper which explains all rather well for the insider, but just this morning I found a derival of it (from the same universaty in Sweden) which I guess everybody should read. Just try from top to bottom as it is only 10 pages, and without real understanding you will at least get the notice of what just *is* going on, and that there's actually no good general solution (meaning : one manufacturer creating a device which will work for all without any problems). By pure coincidence it uses a PWM regulated pump as noise source - which is exactly my own situation, which in the far end is great because that too must be covered for, might you have such an enormous noisy device in your environment; If all is connected well, nothing is seen from it (with general noise of -135dB). If things are not organized so well, it can easily show up up to -60dB (-40dB in rare cases), and this -60dB in my situation was exactly that antenne act of my DAC case, really connected to PE only, neutral and phase cut (I too use the 100% floating setup).

Have fun with this article,

Peter

[PeterSt]

In a separate post, because I think it is the most important of it all, and we can't see it :

The left of the below picture shows a radiating "device". It can be our audio case (with content). How it can radiate is more or less described in the link I provided at the end of my previous post.
In order to understand how devistating this can be for noise or other unwanted signal transfer, look at the right side of the picture. Well, that is just how a transformer operates. So, the right picture actually comprises of a left half and a right half, the left half being equal to the left picture, and when the right half is missing it radiates in our room to whatever can accept it and transform it to energy again (just like a transformer does it internally). In other words :

If you happen to have created that radiating device with no "receiver" opposed to it, it will radiate further into the room. Your amps may start to be the receiver of it. Also, if there *is* a receiver nearby, the energy gets absorbed by that receiver, and there will be no problem further down in the room.

In my view the above can workout two folded :
The radiated signal can be picked up within the device in the case itself and will create noise problems there
or
The radiated signal is not picked up because no proper receiver is there, and now may be picked up further in the room - there creating noise problems.

Of course both described problems will workout very differently.

It may not be so easy to see how the left part of the picture can emerge within itself, but for simplicity let's say that each current flowing somewhere can, and it needs a bit of an "appropriate shape" of the wire which carries the current in order to let it radiate in the coincidental wrong direction (also depending on how "dipole" the antenna shape is, which is related to the directivity and thus spread and short distance, vs. directive and long distance).

On this matter, a small hint to how important balanced interlinks are :
The shield of non-balanced interlinks (SE/RCA) *will* pickup this radiated noise which comes from anywhere in your room. If this were some white noise as such only, it wouldn't be that much of a problem. However, the DC "base" of the signal changes as well, and starts to wobble. In my case this is from plus to minus offset over 100uV, the noise itself being some 30uV (peak-peak). The noise itself (rather white) IMO can not be audible, but the constantly (and random as it seems to me) DC offset ... I wonder.
No such thing happens with a balanced connection, which not only cancels out white noise to some degree, but which will also cancel out the ever changing DC offset to almost nothing.

Peter


PS: And I again promise you ... this topic is still the coincidental begining of a much larger subject, but I can't get my finger behind that yet, since nothing really is on the internet about it. But I guess at some stage I will understand it and report ...

[PeterSt]

Ok, for those who really want to know a little more, and some 4 hours of spare time to read it all through :

http://epubl.ltu.se/1402-1757/2006/64/LTU-LIC-0664-SE.pdf

I (roughly) noted a couple of pages for those who don't want to read it all, but want to grasp the merits of it more or less anyway :

Page 34 , 57, 58, 92, 107 (add 4 pages to each if you want to use the search for page numbers in Adbobe Reader. So, the pages I mentioned (34 etc.) are the numbers printed on the pages themselves).
Although you won't see it much on the pages mentioned, do notice that this paper is audio based.
Do not miss the merit of the "Modulated Switch Frequency" little subject on page 107, which (when you read the whole paper) tells you that devices which radiate heavily at certain frequencies, get tested OK because they apply spreading of that one frequency, the original peak now being low enough (and which peak can be at 200KHz, see PS below), now returning noise throughout the spectrum.




PS: Keep in mind that no regulations/norms/standards etc. exist for radiation under 150KHz. This is believed to be harmless. Sadly this allows devices like my PWM regulated pump (which is only an example) to radiate at 17.8Khz. But also my fridge does (at 15.5KHz or so), and who doesn't have that. But even more important is that this is square waved distortion, which means that only the peaks show up at mentioned freuencies; in the end the whole (audio) spectrum is influenced, since square waves comprise of an infinite number of sines of all frequencies ...

[Scroobius]

Well this is very interesting reading following what happened to me last night.

In my post above I said that my PC was connected to my house ring main and my audio components (mains amps & NOS1) were connected to a dedicated "HIFI Spur". I also said that I could not measure any voltage between the house ring main earth and the dedicated HIFI spur earth (so I thought it was *OK*). But regardless of that last night I connected the PC to the dedicated hi-fi spur along with the main SET amps and NOS1. So then I sat back to listen to music and sat there open mouthed. There was a very big improvement in sound quality I mean very very big everything imaging dynamics detail fluidity (something I did not think possible). To begin with I thought it was just a "good sound" evening or that I was kidding myself but no this was a too big an improvement. So just to check my sanity I asked my old HiFi mate (fellow geek!) David who is well familiar with the sound of my system to check it out and likewise he confirmed the improvement.

So what has happened? to begin with I thought that maybe it was something to do with the sound card and PCIe card being on different earths. But that does not appear to be the answer if my NOS1 is completely floating as Peter states above. So perhaps it can only be that the RFI being generated by the PC is being disposed of much better now. The HiFi spur earth is very good and has a short (10m) run back to the distribution box earth point. But this would not seem to explain such an improved SQ. Any ideas anyone?  by the way my main SET amps have the std tube arrangement of a PE to the cases and a signal earth separated from the PE by a 100 ohm resistor.

Boy does this system sound good now!! the dynamics are enough to wake the cat next door!!

PC

[PeterSt]

Hi Paul,

Well, interesting !
But to be honest, while you seem to be the only one really reporting about two PE rings, you also report no voltage difference between the two, and this puzzles me a bit. If not about your not perceiving a V difference, then about my situation just having that. So what is wrong ?

Let's say your system. Haha.


Ok, I had an enormously long reply at first, but for now I'll stick to the notion that when the signal ground from all your devices is connected to PE (or at least the two which will virtually interconnect the PE rings), the PE rings will be interconnected as well and no voltage will be seen over them.
What this means for your further situation and whether now using two PE rings or one makes a difference - was my reply about, but it's too much guessing from my side. So I stick to this nice short text for now ... (but have saved the long one)

Peter

[Scroobius]

Hi Peter

I think you are probably right.  One thing all this highlights is just how important earthing is. Well I kind of knew it was important before but now I * REALLY * know how important.

Great that you brought this important subject up Peter.

All the best

PC

[Scroobius]

Hi Peter,

Perhaps you could clarify this point. As I understand it ONLY the case of NOS1  is connected to PE and everything inside NOS1 is NOT connected to the case. So as far as I can understand from what you have said the sound card inside NOS1 is connected to the PC earth via the DVI cable and the line output amp card earth is connected to the SET power amps signal earth via the RCA interconnects. Also presumably inside NOS1 all the electronic module earth planes are connected together.

So in my system as it was before with the PC connected to the house ring main and the other audio components (SET amps and NOS1) connected to separate dedicated hifi spur I would expect that the earths would be connected but there would be a 100 ohm resistor (the SET amps have signal earth and PE separated by 100ohm resistor) between them (maybe 50 ohm as they are in parallel). Not a good earth arrangement I agree but maybe I will check again to see if there is some small voltage between the earths (or maybe I will just listen to music  )

All the best

PC

[Nick]

Paul,

A decidedly low tech contribution to a complex subject  .

A few years ago when I lived in a different house that had PE bonding to the copper water system, I noticed that bonding points were a little corroded so I went round and cleaned and tightened up all of the live and neutral ring main connections and cleaned and tightened the earth bonding points. At the end of the exercise the was a definite audible sound improvement. I guess your dedicated audio ring main would be relatively new and would not be a candidate for this, but its interesting that such a simple measure helped so much.

Come to think of it, I'v not done this in my current house, maybe a job for the week end ....

Nick.

[PeterSt]

As I understand it ONLY the case of NOS1  is connected to PE and everything inside NOS1 is NOT connected to the case. So as far as I can understand from what you have said the sound card inside NOS1 is connected to the PC earth via the DVI cable and the line output amp card earth is connected to the SET power amps signal earth via the RCA interconnects.

Maybe I don't understand what the real question is, but the situation is this (for the DACs with meters) :

Nothing is explicitly connected to the case, except for one thing :
DVI cable shield (via a capacitor, which should work with the same effect as your 10R resistor -> slowing down current paths to external, so they stay inside).
However, within the PCI arrangement, this shield "ground" connects to the digital signal ground which in the end is (viavia) of course connected to all internal floating ground.

This all means that when you'd measure the connection between signal ground (e.g. on the RCA cable) to PE, you will see the 100% connection (0 Ohm). Still all is arranged so that nothing is spit from the DAC, because it just doesn't spit (this is fair to say).
The DVI shield connection is necessary because otherwise PC noise will travel over the data line (more theory than noticed practice).

Peter