Something wierd!

evonimos

Member
2019-05-12 3:11 pm
I'm not sure where it's best to post this, but since it relates to transformers I'll post it here.

So, I swapped transformers in a headphone amplifier, from classic E-shape core type to a toroidal.

Voltage ratings are same on both, while current on the secondaries is rated @1.5A for the E-type and @0.7A for the toroidal.

I did this for size considerations, since the toroidal is smaller.

Now comes the wierd part:

I decided to do recordings before and after the swap, just for the heck of it.
I didn't expect much, but to my surprise the damn thing sounds different.
Particularly the soundstage width.
It sounds a bit more wide.

Granted, the difference is small.
Yet, with a good pair of headphones and some concentration from my part, it's audible.

I'm no electronic engineer or an expert, so I'm clueless whether this is to be expected, explainable or even possible.

Oh! There's something even more wierd:

I decided to do an additional recording on the new toroidal after a couple of days of continuous use.

Listening and comparing the recordings, I noticed that the sound has changed further. Like it has 'evolved' a bit further.
Again, the soundstage width was a bit more expanded.
The change this time is to a lesser degree than after the swap, but it's still detectable by my ears.

So, what is the deal here?
Either I'm imagining things or there's a real phenomenon.
A so-called 'burn-in' on the transformer(?).
This alone happens to affect the sound(?).

Any thoughts will be appreciated and forgive me if I miss something that I should know.
 
Last edited:

evonimos

Member
2019-05-12 3:11 pm
A better power supply could lead to hearing better channel separation due to less distortion or noise common to both channels.

That makes some sense to me.

But what about the 'burn-in'(?).
Is there a plausible explanation?

I can understand burn-in in mechanical terms, but in electric terms it's certainly a totally different landscape.



By the way:

I can post links to the recorded files I use for the comparison.
They are in WAV format, 24-bit @ 48kHz.
Time edited and adjusted for same gain (within +/- 0.2dB).

The player i use is foobar2000, which has a neat plugin called 'ABX comparator'.

It is free and you can use it to switch between the files on the fly.
It also gives you the choice to maintain the same playback position when changing tracks.

This little plugin is an invaluable tool for anyone that wants to do these sort of comparisons.
The brain has a very short auditory memory and it is impossible to remember so minute differences when listening to tracks in a sequential manner.
Switching between tracks instantly provides a massive help in identifying the differences.
 
Last edited:
This sounds very suspicious. If it's audible, it should be plainly obvious in measurements. Have you thought of performing an RMAA loopback test, at least with manual playback and recording of test files (RMAA can be a bit flaky and absolutely wants reference tone levels higher than -10 dBFS)?

I would not be surprised if you saw some irregularities in channel separation, possibly indicating a shared ground issue. If so, try varying load impedance for an idea where the problem is. If it's the output side, inspect every ground solder joint from the output jack back. Maybe something came loose while you had the amp apart. While this would arguably be an embarrassingly trivial fault, I prefer those to almost any other ones.
 
Last edited:
I'm not sure where it's best to post this, but since it relates to transformers I'll post it here.
It actually belongs in the Psychoacoustics Forum (if there is none , somebody should create it, sometimes I feel 50% of posts belong there)

So, I swapped transformers in a headphone amplifier, from classic E-shape core type to a toroidal.

Voltage ratings are same on both, while current on the secondaries is rated @1.5A for the E-type and @0.7A for the toroidal.

I did this for size considerations, since the toroidal is smaller.

Now comes the wierd part:

I decided to do recordings before and after the swap, just for the heck of it.
I didn't expect much, but to my surprise the damn thing sounds different.
Particularly the soundstage width.
It sounds a bit more wide.

Granted, the difference is small.
Yet, with a good pair of headphones and some concentration from my part, it's audible.
You hit the nail in the head.
Thanks for being honest. :up:
I'm no electronic engineer or an expert, so I'm clueless whether this is to be expected, explainable or even possible.
Engineering has nothing to do with it.

Oh! There's something even more wierd:

I decided to do an additional recording on the new toroidal after a couple of days of continuous use.

Listening and comparing the recordings, I noticed that the sound has changed further. Like it has 'evolved' a bit further.
Again, the soundstage width was a bit more expanded.
The change this time is to a lesser degree than after the swap, but it's still detectable by my ears.
Ears are part of the chain, then there´s heavy processing by the brain ... which is sadly influenced by Psychology, expectation bias, etc.

So, what is the deal here?
Either I'm imagining things or there's a real phenomenon.
A so-called 'burn-in' on the transformer(?).
This alone happens to affect the sound(?).
No such thing as transformer burn in.
Except for fresh varnish smell and I am not kidding.
I wind my own transformers and they stink for a couple days when heating up, until solvents evaporate and solids cure or harden.
Any thoughts will be appreciated and forgive me if I miss something that I should know.
Thanks for offering your honest experience.
 

evonimos

Member
2019-05-12 3:11 pm
This sounds very suspicious. If it's audible, it should be plainly obvious in measurements. Have you thought of performing an RMAA loopback test, at least with manual playback and recording of test files (RMAA can be a bit flaky and absolutely wants reference tone levels higher than -10 dBFS)?

Will give that a read, since i'm not familiar with the term.
Thanks for the suggestion.


I would not be surprised if you saw some irregularities in channel separation, possibly indicating a shared ground issue. If so, try varying load impedance for an idea where the problem is. If it's the output side, inspect every ground solder joint from the output jack back. Maybe something came loose while you had the amp apart. While this would arguably be an embarrassingly trivial fault, I prefer those to almost any other ones.

I'll give it a thorough check.

However, the audible differences among the files are not that great to indicate a possible fault.

In fact, it probably best to upload the three files here so if anything, other folks can try it and chime in.
 
The brain has a very short auditory memory and it is impossible to remember so minute differences when listening to tracks in a sequential manner.

A lot of people repeat this like it's a quote from the Bible.

Yes, i strongly disagree. The ability to memorise sounds is fundamental to this hobby.

And power transformers do have their own sound. Even between similar construction types and similar power ratings. It's been known for years.
 

evonimos

Member
2019-05-12 3:11 pm
Here are the recorded tracks for anyone interested to download them a give them listen or to run any kind of useful analysis on them.

Because as they say: 'the proof is in the pudding'.

Recording using E-type core transformer in the power supply:
Rec via headphone out - E-type core - 'Corina' - Corina.wav - Google Drive

Recording using Toroidal-type transformer in the power supply, with less than 1 working hour on it:
Rec via headphone out - Toroidal without burn-in - 'Corina'.wav - Google Drive

Recording using Toroidal-type transformer in the power supply, with more than 24 working hours on it:
Rec via headphone out - Toroidal after burn-in - 'Corina'.wav - Google Drive


The recorded track is 'Albert King - Corina'.

It's a pretty good recording, with decent stereo image.
I believe it can be quite useful in highlighting differences in soundstage width/height (if any).
 
Last edited:

evonimos

Member
2019-05-12 3:11 pm
A lot of people repeat this like it's a quote from the Bible.

Yes, i strongly disagree. The ability to memorise sounds is fundamental to this hobby.

And power transformers do have their own sound. Even between similar construction types and similar power ratings. It's been known for years.

I guess there must be certain exceptions when it comes to people's abilities.
But the general consensus is probably that auditory memory is quite short.

Personally, I can assure you that I belong to the short-memory group.

In any case, I'm quite mystified by the possibility of a power transformer having any sort of sound to it..

And it gets even more mysterious when one starts contemplating effects of the so-called 'burn-in' to the sound of it.
 
In any case, I'm quite mystified by the possibility of a power transformer having any sort of sound to it..
And it gets even more mysterious when one starts contemplating effects of the so-called 'burn-in' to the sound of it.
Don't be mystified, it's all in your brain, psychoacoustics, there are people who hear differences in solder.
 
Pleasure!

But, exactly what do you mean by it?
Would I have any reasons not to be..?

Seriously mate, I don't understand the context of that remarks.

Obviously I don´t know you from Adam, but in general in the Audiophile World (and in Art / Politics / Religion / Beauty / Sports / etc.) it´s very common to offer biased opinion cloaked as Universal Truth .

That you offered your observations but admitting something might be different and was eager to hear opinions is , believe me, quite unusual, and deserving praise :)
 

evonimos

Member
2019-05-12 3:11 pm
I have the option of running a true ABX comparison in foobar2000, with the free ABX plugin.

Actually any of you can try it too, since I've uploaded the links to those 3 files.

The differences are pretty small in my opinion, hence the need for concentration by the listener's part.

So, one needs to pay attention to identify them regardless of the method of testing.

Of course, a true ABX comparison eliminates the variable of confirmation bias.
It's surely the most serious and valid method of testing.