Can anyone help me with this distortion ?

[PeterSt]

Hi all,

Below picture I posted before in the "DAC" topic some weeks ago, but in the mean time it is following me to my death and is quite intrigueing at the same time. I don't have the hope that anyone knows the answers, but maybe you know someone who is highly skilled on these matters, and does.

What you see below is the distortion on the mains from a 3 phase (380V) pump and its frequency converter. This is pulse width modulation, and the revs of the pump can be more or less by means of changing the pulse width as taken from the original sine on the mains. This means squares are left, and those are fed to the pump's motor.
What the picture will show is the feedback on the mains of that square impeeded process.

So far I could find nothing to remove this from the audio system (the picture is a measurement by microphone 1 m from the speaker), and separate mains groups don't solve it either.
There is already a kind of promise from me to you that "you" will have similar. Not on the same frequencies, but you will. It is my guess that every switching PSU will show this, one more than the other though. And, it depends on whether those spikes are above the noise level. In my case it is, and a freezer does the same (at two other frequencies) though with lower amplitute.

So, do I ask you to solve this ? well, not really (it will be solved by power regeneration I'm sure). But what I want to understand is this :

Each device which is added to the mains or indirectly connected to the audio system, will let rise the amplitude.
What this comes down to, is that plain white noise added will let rise the amplitude.

The frequency will stay exactly the same always. Note that this is a very steady tone, and it will be a square wave by itself (look at the side lobes).

If you didn't grab the strangeness of this, maybe these examples start to make you wonder :

To eliminate all unnecessary disturbances I removed the microphone from in my case the Fireface800 which is the capturing device but kept reading the input. So, while before I was captuting the sound through air, now it is just from the electronics. Air molucule noise is avoided now, and only the electronics' noise is left. Here too the beautiful tone is visible. Now :

When I connect an interlink anywhere in the audio system, the amplitude rises. Say, with 3dB. When I connect just a connector to that interlink, the amplitude rises again. 1 dB or so. When I connect a poor connector (like a cheap Y), the amplitude rises again. Can be 5dB.
When I connect the NOS1 DAC without cabinet (this is how the whole subject started) the amplitude rises. 20dB. Yep. And the DAC wasn't even connected to the mains.

Still here ?

I proved that the gerenal noise level did not rise with the same amount as that one tone (with all its side lobes). For example, the 20dB the tone rises means only 3dB for the general noise level. Also, you can well say that adding the connector incurring for 1dB rise of the tone, would be unnoticed at the general noise level (unless that could be expressed in a "totalled" number of course).

So, I think I found the most perfect means of measuring your cables and connectors. Send them all to me, and I will test them for relative noise. Hahaha.

No ... I think what I proved is that something is happening standing wave like, but now in the electrons in the cabling. Well, somewhere.
If you'd ever seen how aliasing may create similar in the audio spectrum ... I mean where two aliases coincidentally meet, their amplitudes are just added up, and show as mathematically added. Here a kind of happens the same it seems. But I can't reason out how or why.

The pump's square feedback on the mains looks like a fact;
There is no way I can think of that adding connectors and cables and devices will have some kind of additional noise disturbance at 17.5KHz. No way.
But what * *can* think of, is that somehow white random noise interacts with a square which is there to begin with. This, then, should only be so when the white noise are more or less squares by itself. And, they may, thinking of all various small spikes -by themselve already squary or not- adding up adjacently to more real suqares, and the area of the tone just nicely adds up for the amplitude. I say "nicely" because that really is so. Again, no way that the frequency of the tone changes or gets unstable (which would mean the frequency where it happens starts to change more or less). If enough noise is added, the whole screen is occupied by this monster, and from left to right it is all over with side lobes of the suqare's harmonics.

To be clear on things : after 2 weeks or so the NOS1 DAC laying around without cabinet I got crazy of something. I heard a high pitched tone, and mainly when I was in bed. I think this happens when you are under the influence of such a thing (which was dancing at -55dB at that time) while you actually can't hear it (if I hear 12K I'd be happy). I only want to say : first I heard something, and then I started looking for it my means of a microphone. So it is really there, and not only in the equipment.

Any contestants having answers ?

Peter

[pedal]

This reminds me about the old Spielberg movie Poltergeist. I think your house is built on an ancient grave yard. The are probably 17500 spirits residing in the soil. 

[leifchristensen]

isn´t  this the freq you blamed on the Buffalo?
best
Leif

[PeterSt]

isn´t  this the freq you blamed on the Buffalo?
best
Leif

Huh ? no ... This is what I wrote about it : (from The Verdict ...)

I must tell you that there another small background to really want to try to get the Buffalo "work";
The NOS1 exposed noise which made me mad. Some high frequency stuff I couldn't bear. HF stuff that goes with you to bed, and with which you get up in the morning. It got me crazy.
Most of last week I have been trying to get it away, after I first could measure (by microphone) it really was there. A 17,5Khz tone which was 20dB louder than the music ...
This story by itself is many pages longer than the few lines I write about it here, but what it comes down to in short is that while the Buffalo would not allow this tone (which really is a steady tone) to be "amplified" hence changed at attenuating via the TVC I have, the NOS1 would. But this works exactly the other way around as well : the TVC can attenuate it with the NOS1, and in combination of things, it could get me the tone just above the noise level, as it could in the first place with the Buffalo.
More importantly, early this morning I found what causes this tone : my 380V pump of the central heating ... Also, other tones (at 15KHz and 19KHz) are there from the freezer in the basement. And although I have completely separated audio power groups with own earth, nothing can get it away. Not even a sine re-generating battery power plant !
But in the end I could get it down into the noise level, so all is fine for the moment. One thing : I have to use the TVC for it, which I really want not. New freezer and a new pump I guess ...

So, really the opposite. The Buffalo didn't add noise as such, while the NOS1 did (and which was caused because not being in a cabinet). In either case the added noise by itself is not the real problem (the noise being at -95dB) but it adds up to that stupid tone.

Btw for your information : the Buffalo -with it's standard PSU- exposes a "radio noise" which wanders through the spectrum. Probably inaudible through less sensitive speakers, but very visible through a spectrum analyser. I hope you know what I mean with radio noise.

[GerardA]

Hi Peter,

I guess you know this already but I can not read this between your lines.
You have a seperate earth for the audio but is this completely separate with it's own earth rod?
To me it looks like you have a problem with your gear working as an antenna, and picking up noise from the mains.
This is exactly what happens to a low output-mc cartridge and by measuring this with the german audiotester software
I saw something similar. A good earth lowers the 50 Hz tone with about 15 dB, but the surrounding noisefloor goes doen a couple of dB's too.
A long time ago I decided to use batterypower for my tube-preamp to be independent of the mains. This helped a lot.
From before I remembered that using the mains earth only gave extra nasties. So if you share your mains earth only a little with your wifes appliances than you will notice it in your musical pleasures.
The extra connectors you plug in make it look like they work like an antenna.
This problem I had with my mc-preamp after I did some resoldering a long time ago. Bad connections are good antenna's.
Even a high resistance vdHull cable with carbon wire can give problems this way.
So that's why my solution to all problems I already posted before with a picture how not to do it.

If this is old news then still I like to know your opinion about it!

[PeterSt]

Hi Gerard (A),

Yep, separate earth rod as well. But I don't think this is coming through ground, and that's why this doesn't help. It's just on the phase(s). Besides that, I have no ground connected devices (but the PC, and I had it out of the wall during testing).

I have a mains sine re-gerenator (1800W) and it comes through there too. However, I don't trust this 50Kg beast when it is chargeing the battery.

Bad connections are good antenna's.

I don't know why, but I sure believe that. Typicle thing for vinyl addicts to discover btw.

So that's why my solution to all problems I already posted before with a picture how not to do it.

Yea, that was a nice one. A nice guy too. He was finished yesterday. Didn't help. (http://www.phasure.com/index.php?topic=642.msg4849#msg4849)

Thanks,
Peter

[astacus21]

i had a problem on my system but i don't know if it is related with this one. I've noticed that without playing music,when i turn my headphone amp to the max volume i could easily listen to some high frequency tone through the headphones. I had this noise only, when my dac was connected to the amp. If i had the dac disconnected, from the amp this noise disappeared. So i thought that it was a dac problem.

Some days before i found on an rj45 Ethernet cable a removable small filter (ferrite bead http://en.wikipedia.org/wiki/Ferrite_bead )
It is supposed that it reduces the common noise from the ethernet cables. So i attach this filter to the power cord of the dac and this noise is hardly audible now. I dont know if it influences the audio signal, but now i have a much quieter background. I suppose that this problem has to do with emi / rfi.

[PeterSt]

Yes, I use those. They should be on any powerchord from high frequency stuff like a DAC. If not, the DAC injects RFI etc. stuff back into the mains. It is for that direction (into the mains), not the other way around.
Unless you have it on the mains already. Then it might help I guess.

Hmm ... I thought of mounting such a core on the pump's cable (which is not a one hour job in this case), but left it alone because there's filters in the pump for this which should work just the same (but don't, so why would such a core help ... was my thinking).
But obviously I could try one piece of equipment, probably the Fireface itself not connected to anything plus the measuring PC ...

Hey, thanks.
Peter

[GerardA]

Then maybe a redesign of the power supply, better rejection and filtering of the mains should help.
If we build or buy it we cannot ask all the neighbours to buy a new fridge...

On the internet there is another interesting DAC wich uses batterysupply. Some companys offer an ac-supply specially for this.
Like Paul Hynes in the UK. Maybe he knows some tricks for you?

[SeVeReD]

Then maybe a redesign of the power supply, better rejection and filtering of the mains should help.
If we build or buy it we cannot ask all the neighbours to buy a new fridge...

On the internet there is another interesting DAC wich uses batterysupply. Some companys offer an ac-supply specially for this.
Like Paul Hynes in the UK. Maybe he knows some tricks for you?

I like the idea of using a battery supply feeding the dac.
 I'm still fighting electrical stuff at my place; even though I've got the system running off a subpanel ran with decent sized copper, audio on one leg/computer & video on the other leg, all front end coming out of a ps audio powerplant, and have even shut off all other circuits but the amps in the house, I still get low level noise coming through the speakers... some days worse than others.

How about 12V car- or motorcylce battery?

[PeterSt]

Well, I got myself a Kemp Powerstation 75, and it dropped the tone by 30dB. You can say it is virtually gone.

There is something else which IMO is more imortant :

First off, what I am going to say next is from an engineering point of view - or just from the specs if you want, just common knowledge. However, I will try to explain and virtually visualise what actually happens with - and without such a device.

Ok, this is a mains sine re-generator. It is running on-line (hence no batteries involved) and dynamical surge needed from a poweramp is just passed on to the actual mains. On that matter I found no limiting in dynamics (so far -> must test longer but I think it's okay);

Without the device, in my situation anyway, the noise level coming from the mains is at -95dB. Note this figure is relative to microphone gain and microphone distance.
This -95dB is average, and millions of peaks per second are in the -85/-90dB range. Imagine a screen with a horizontal line with all dancing spikes ...
Now, at playing music there's this general horizontal line of "music spikes". Imagine a track which is actually full of sound and never silence, but there's the general beat (supported by drums etc.), and this line is dancing up and down, and the "ups" are on the beats. Generally you could say that on the beats the line goes 30dB up compared to in between the beats. So, those beats determine the peak SPL (Sound Pressure Level), and btw they will also determine the maximum digital level used.

They key to all is that in between the beats the music is playing right in the noise. Thus, at the -95dB level from my example (and microphone distance etc.).

I have seen this for years, and did not pay much attention to it, thinking it *must* be normal, and the noise shown is just as is and is always there. Besides that, it *must* be harmless, otherwise I would notice distortion or whatever.
Also, from the highly sensitive (horn) speaker you can hear a kind of sweet (white) noise, and thus this is what I saw on the screen.

Not so ...

With the Powerstation feeding the power amps, the noise has completely disappeared from the screen. I don't know how much it is down now, but it must be 15dB at least, because before the peaks were at -85dB and the screen (I use) doesn't go further than -100dB. Moving the mike into the speaker and setting mike gain to much still did not show the noise, so it could be way more down or not there at all (one day I may be measuring this more officially).

When I now look at the music playing, it is this "line" of music only. It is not surrounded by noise anymore. It is in its black own ...

Ehh, I said "Not so ...";
The sweet noise from the speaker is still there, and it is there as it was before (meaning : I can't hear a difference). As it only now appears *this* noise is not the noise I measure. Not before the Powerstation and not after.

Before people get confused : this is measuring by microphone right at the end of the chain, hence with everything involved which influences. This is nothing electrical, but the electrical noise has its influence to the result, and I'd say, the most highly. This rather strange way of measuring (which I started doing by kind of accident) unveiled IMO properties unknown. Properties which so far are still not clear to me exactly;

Something which shows as white noise, looks like not being white noise at all. I think I (kind of) proved that by that noise not being audible as such, but being able to capture by microphone anyway. Btw, note that this is not electrical noise in microphone or cable etc. because pointing away the microphone removes the noise from the screen.
It very well may be that what I appreciate as noise (or what the analysing software makes of it) is not noise with a density of, say, 10000 frequencies it occurs at per octave, but 10 or 20 or so only. At  least that is what the screen shows, but then the screen (the software) FFT's the signal and might not show what is actually there. It is thoughout the audible spectrum all over, and everywhere at the same level. This by itself would imply normal white noise, but I just don't believe it is.

Back to the beginning of this topic, it is as strange (ok, to me) that adding noise by connecting devices and connectors and all, cumulates in that 17.5KHz tone I have by accident (because of the pump). So again, add 3 dB or so of that "noise", and the tone goes up by 10dB easliy.


For those who run into this some additional background :

At measuring the Phasure NOS1 DAC it came out that this DAC is totally immune to whatever is hanging out on the mains. A little more specific : it is totally immune to that tone. It can't be found in any measurement. About the general noise I don't know, and this takes new measurements with and without the Powerstation.

Because the Powerstation hardly can bear the DAC (which just draws enough on its own to almost overload the Powerstation 75 (Watt), only the main amps are connected to the Powerstation (they use 33W x 4 at fully needed power) and at the levels I play this is okay;
The relevant information is that this proves that the tone comes into the system by means of the power amps.
Keep in mind, the tone is still there, but 30dB lower now. I did not work out whether the residu is from it leaking through the Powerstation, or whether it comes in via a backdoor like the DAC being connected to normal mains. This cannot be tested by normal means, because the tone is just visble now, and will disappear from the screen when devices are desconnected (less general noise = amplitude on the tone).


All 'n all a device like the Kemp Powerstation 75 sure does its work at measuring things, and I am sure this is all audible too (didn't play much attention to that yet). However, right now I am more interested in what is actually going on, and if I'm possibly on to something that might improve audio playback. The convenient thing is that I now can compare which theoretically makes it more easy to find out what it actually is I am measuring through microphone. It sure is nothing like HF noise, unless I'm looking at aliases of it. But I don't think I see aliases, because I would see patterns then.