Late night listening (aka your mains supply)

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Stirred into action by the 'How big do the capacitors have to be thread', I mentioned the phenomenon of HiFi sounding better late at night.

It's something I've noticed for years, and there are obviously several possibilities as to why.

Generally ambient noise levels are lower, so quieter sounds and more detail can be heard. There's something more fundamental than this though, that affects the system beyond simple resolution improvements.

The noise levels present on the incoming mains supply are a candidate, and with this in mind, spurred on by another correspondant I made some interesting measurements using the spectrum analyser.

The measurement system is a simple wall-wart transformer (AC out), and an attenuator on it's output to reduce the mains fundamental to suitable levels. A quick check with a signal generator to ensure it's bandwidth is suitable and off I went.

Initial plots were eye-opening, you may think your mains supply is a nice 50 or 60Hz sine wave (even mine looks good on a 'scope) but there was an extended sea of odd-order harmonics extending right across the audio band.

Could these be part of the reason for the improvements late at night?

I left the PC recording the audio data from the mains overnight, and started to examine a few gigs of data the next evening.

The following FFT plots show what I found, of particular note is that whilst THD does drop later at night / early morning, the harmonic extension reduces considerably, levels at where the ear is most sensitive all but disappear in the early hours of the morning.

With my system, despite almost everything being fed from regulated PSU's with low noise levels and ultra high line rejection there still appears to be a mechanism by which this noise affects sonic performance.

I've also noticed at times during the day, when the system sounds better I can usually correlate this with an improvement in the mains noise levels.

What can be done about this is something I need to apply a lot more thought to, since the easy solution of applying line filters to the mains inlets of the HiFi is a sonically degrading solution that kills the music.

Thought it might be of some interest - thoughts and ideas welcome!


  • mains at 0100_s.gif
    mains at 0100_s.gif
    45 KB · Views: 1,010
Generally ambient noise levels are lower, so quieter sounds and more detail can be heard. There's something more fundamental than this though, that affects the system beyond simple resolution improvements.

Totaly agree with these points, and power pollution too. I suspect another source of pollution is seismic. Low level vibration affects EVERY component in your system. Could be from traffic or industry depending on where you live, which is reduced late in the night. If you haven't tried isolating components from vibration you might get a surprise, even if your components are not in the listening room.

Music Machine
late night ... or early morning ... electrically & envirnomentally?

In commercial cases or even homes, computers , printers, copiers, etc introduces a lot of harmonics to the power lines and that could be one of the reason you see so much noise on your scope. There are lots of ways to remedy that but generally it gets expensives the "cleaner" you want the power line to be.

For me, the best time (or to get the best "high') to do critical listening is first thing in the very early morning ... right after I get up and during my first cup of coffee. I find that in the early morning, my mind is "fresh" from the previous night sleep, my ears has not been bombared with noise pollution and typically the environoment is quietier. As you can tell, it's not an easy thing to do since everybody in the house is sleeping ... not to mention the neighbors!
Thanks for the spectra ALW. I'd seen similar done before when I was apprenticing to an engineer and we were trying to trace a noise problem in a system, so we put the specan across the mains and looked, and saw something similar to what you posted. We were in an industrial area with some businesses that ran 24/7, but there was still a reduction at night. There was also a difference between individual phases on the mains with A phase being the worst and C the best. I would expect this to be even more pronounced in a residential area where a majority of homes are wired across the A phase, especially in areas with overhead wires, simply because it's easier to reach the A phase.

The solution to the problem at the time was to use a huge surplus 5kVA isolation transformer and a grid of earth stakes. The bandwidth of the iso trans was only a few hundred Hz, so EVERYTHING above that was attenuated drastically.

Nowadays, I won't build gear with torroidal transformers for the exact same reason: too much bandwidth lets too much junk into the system which is a real hassle to try to filter out. I like first order solutions. Even a huge mains torroid I got recently with an electrostatic strap isn't very effictive at isolation.

BTW, if any Aussies in the BrisVegas area know where I can score a large surplus iso trans I'd be grateful for the tip.


I've done similar measurements in South Africa and came up with very similar results. Didn't show that clear a correlation between day and night though. My feelings towards mains filltering is very similar to yours, or maybe i've just been unlucky. As with everything else you get a huge spread in opinion about this issue, mostly i suspect due to individual sound preferences. It's similar to the preference for sound absorbtion/dispersion in a room. If you prefer high absorbtion you're likely to prefer mains filtering, paper-in-oil caps in amps, lots of chokes in valve power supplies, damping in arms. All of the above calmes the sound but of course at a price.
May i suggest that we start a new group project dedicated to power regeneration? My modest attempts have shown that 0.5%thd sounds a lot better than 5% albeit on the small scale of my experiments (<50W).

Feeding audiophile VooDoo!?

I too have noticed the phenomena but this measurement would have been a lot more meaningful to me if it was done on the dc supply rails of the amp and then audio in and out checked for any coupled mains noise. How smooth are the dc rails at different times of the day? If the power supply's properly designed and none of the noise is coupled into the amp anywhere else, then it's just biology and environment. You're winding down at the end of the day, it's dark outside, any background noise is lower, your mind is quieter, there's no other distractions, you can focus better, etc.
Feeding audiophile VooDoo!?


These effects are orders of magnitude subtler than what is measurable, same as everything else that matters sound-wise. We can only surmise what transpires in the transformer, rectifiers and subsequent circuits when the mains is distorted, poluted or of variable impedance. It's very naive to think that 'properly' designed regulators filter all mains effects. Well, they probably do, only noone knows how to design them :) Real life regulators allow a critical listener to hear the presence of a fuse, cabling and a whole lot of stuff in a pre-regulated PS.

Born sceptics

The points raised about measuring effects at the output or the preamp power rails are valid and sensible, but flawed in that the assumption made is that the effect of the mains supply noise is obviously measurable.

My low level supplies are all super-reg based, and the inherently massive line rejection and wide bandwidth of these devices means that simple steady-state noise measurements are meaningless. None of the mains supply noise gets through to the power rails, yet I can make changes to the raw supply (different caps, transformers or filters etc.) and the sonic effects to the unit being powered are obvious.

Adding for example a line filter ruins the sound, yet the mains noise drops when using one, and the output of the regulator looks unchanged!

I'm not saying it's unmeasurable, but more that I don't know what to measure.

As an engineer I find this stuff fascinating, I always want to understand and be able to measure the effect I'm obviously hearing, yet have often been unable to do so.

Measuring power amps rails is pointless, they all look much noisier than the incoming mains, owing to the high levels of harmonic generation through narrow conduction angles.

It's defintely not just biology or environment, though I accept these have equally obvious effects on perceived sonic performance.

I like the malt suggestion, personally.

RE: Test suggestions

I also agree with Sydney and Bill.

ALW has published the measurements of the "input" of audio gear but we need to know also how it affects the audio signal. (BTW: the measurements are somewhat reduced - many EMI sources "work" also in MHz areas and such frequencies have also some influence on the circuits which could result in audible effects/distortion).

I would suggest to measure the inluence of the supply rail changes on some signal e.g. at least sine wave. Would it be possible to do the same test by measuring the changes in THD of 1kHz sine wave (or to measure IMD)?

I agree with you that making component changes to a power supply can make a huge difference to sonics for numerous reasons that all have their roots in electrical theory. You also said:-

"Measuring power amps rails is pointless, they all look much noisier than the incoming mains, owing to the high levels of harmonic generation through narrow conduction angles".

I cannot believe you have an amplifier with dc supply noisier than the mains supply, if so you should probably be building a new amp :) or at least a new power supply?

Also could you tell us a bit more about the narrow conduction angles and harmonic generation?


lateral thinking

If you can hear the difference when you add a mains filter or change other input components but you cannot measure the difference at the output of the supply then it might be that your measurements are not sensitive enough or you are not measuring the right thing or you are not measuring in the right place.
If you exhaust the first two and have no joy then perhaps you might consider whether the audio signal is being affected somewhere else in the system other than the psu outputs? If this is plausible then you might consider a controlled experiment whereby you have two psus operating at the same time, then swap the pre-amp from one to the other. Perhaps have a dummy load on the unused one. Then see if the affect still follows the psu.
Art Of Noise...

Andy, first off, thanks for going to the trouble of logging and posting the noise graphs.

"Adding for example a line filter ruins the sound"
I have a rather large sealed power filter out of a junked mini-computer system (20cm x 15cm x 7cm) mounted on a board with a mains plug lead and a dual GPO.
Every system that I have hung from this portable filter is system quieter, including cheapish shelf systems through good hifi through several kW PA systems.
By this I mean that the system is intriniscally quieter, and more relaxed and this seems to lead to less dynamics caused IMD, which although low level add a mask of broadband noise to the final sound.
"As an engineer I find this stuff fascinating, I always want to understand and be able to measure the effect I'm obviously hearing, yet have often been unable to do so."
I reckon that the effects are a lot to do with leakage coupling to grounds.
This leakage current being mostly capacitive coupled is bound to vary according to the spectrum and level of the mains borne noise.
If this leaked noise current is getting tangled with audio grounds between equipment items, valid audio will be amplified.
The validity of the mains earthed equipment grounds is important.
I find using stacked piggy-back plugs provides a single good system earth reference point, and this helps, but the order of these plugs can influence too.

I agree that on first listen a filter can seem to deaden a system, but on extended listening I find this to be due to reduction of 'false' sounds, and reduction of colourations.
This implies power feed and DC supply resonances, and earth leakage current caused noise i think.
I find that line level audio isolation transformers provide a useful benefit also.
I find that a good test is to listen close-up to the loudspeakers, and note any changes in the 'white' noise, no signal base system noise character.
This can help you to home in on a good arrangement.

Brett, I have a range of small medium and huge, old and more recent isolation transformers that I have scrounged from various sources, and mostly for less than $10.00.
Flea markets, industrial machinery and demolition yards, and importantly metal recycling yards are the sources of choice.
I picked up a Sola 3kW constant voltage/isolation transformer box (2 man lift) for $25.00 from a metal recycler.
They are also the bestest value place to for power wire and telephone cable, and there is usually a range available for experimenting with real cheap.

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.