The No.1 Website for Pro Audio
Gear changing it's Sonic Footprint
Old 30th January 2020
  #1
Gear Maniac
 
jontornblom's Avatar
Gear changing it’s sonic signature?

Hi! Hoping someone in this sub forum can give me some more specific answers with an issue I noticed.

Timeline

9:00 am Master the song in question
10:30 am - 8:30 pm I master other music, taking breaks but leaving my gear on between sessions
9:00 pm Mix revision for song in question is delivered and I simply need to recall my settings and print.

The issue is that at 9pm the master sounded noticeably darker. I am able to perform exact recalls, and investigated my settings thoroughly. I was able to set the song up to sound the same, but reset the controls to my original settings and yes, the sound was noticeably darker.

Edit: and also I printed versions of each and A/B or A/B/C compared each version with the appropriate scrutiny and thoroughness

Edit #2: This is not a consistently recurring phenomenon that I can simply measure at will. I have only observed this twice in 12 years of mastering.

The next morning, I turned on my gear, let it warm up, and without touching a single thing, the darker sound from the night before was totally gone and it sounded identical to the original. For all you Occam's razor type people out there let me just emphasize that I checked my settings extremely carefully and multiple times.

I've done thousands of recalls and this only happened one other time about 6 months later with basically the exact same timeline as above:

The sound changed after my gear had been on for 11+hours.

I ruled out my tube compressor since that seemed like the most obvious culprit. What's left on that master is a totally passive device (unlikely since it's away from all the hot pieces), an inductor based EQ, and a solid state based EQ.

So I know the workaround is to only work reasonable days (8 hrs), but the nerdy part of me NEEDS to know what is going on.

My current (heh) theories are

a) the inductor alpha coefficient in my EQ is drifting due to temperature increase after many hours (it's a known behaviour of inductor cores that temperature increase can lower the saturation point for the passing audio, but why only after 11 hours? I'd assume thermal equilibrium is reached after about an hour)
b) the voltage delivered to my system seems to be higher than usual after regular business hours (I suppose when a lot of commercial/industrial loads are drawing less power from the utility.) This goes from 121 V during business hours to about 126-127 V after hours. But I don't know what a higher voltage would cause since the PSU's in my units are all fairly robust. Maybe that's the source of additional heat and a) and b) are working in tandem?

Note: All my equipment is switched for exact recall and I've recalled sessions from 6 months prior or 6 hours prior that sound indiscernible from the original master. Recall is not the issue. (I made this bold and italic to try and prevent the typical forum knee-jerk responses to questions like this)

Sorry for the long description, but any insight would be excellent and greatly appreciated!

Last edited by jontornblom; 6th February 2020 at 03:36 PM..
Old 31st January 2020
  #2
Lives for gear
Alternate knee-jerk response... what is changing is your hearing.
It fits your evidence.
After 12 hours of semi-intermittent listening, even if the music isn’t at jackhammer levels, such prolonged listening can cause your hearing to change, temporarily in this case. Stuff sounds darker... After sleeping (we hope), ten or twelve hours later, you listen again and everything sounds correct. Your ears are rested and fresh.
Possible?
Remember that you are the most changeable, malleable element in your mastering and listening chain.
Old 31st January 2020
  #3
Lives for gear
 
Radardoug's Avatar
 

i agree with Bushman. How many coffees in the 12 hours? How many beers?
Old 1st February 2020
  #4
Lives for gear
 

Agree with above. It's your hearing, or at least your perception of sound, that is changing.

Consider how much your visual perception of darkness and brightness changes between night and day.
Old 2nd February 2020
  #5
Gear Maniac
 
jontornblom's Avatar
Quote:
Originally Posted by Bushman View Post
Alternate knee-jerk response... what is changing is your hearing.
It fits your evidence.
After 12 hours of semi-intermittent listening, even if the music isn’t at jackhammer levels, such prolonged listening can cause your hearing to change, temporarily in this case. Stuff sounds darker... After sleeping (we hope), ten or twelve hours later, you listen again and everything sounds correct. Your ears are rested and fresh.
Possible?
Remember that you are the most changeable, malleable element in your mastering and listening chain.



So all three of you guys assume I wasn't A/B/C comparing the three masters the next morning? OR A/B comparing the morning master against the evening one?

And no, Radardoug, I don't drink alcohol while I work. Especially at 8am in the morning.


Last edited by jontornblom; 2nd February 2020 at 12:54 AM..
Old 2nd February 2020
  #6
Lives for gear
Quote:
Originally Posted by jontornblom View Post
So all three of you guys assume I wasn't A/B/C comparing the three masters the next morning? OR A/B comparing the morning master against the evening one?

And no, Bushman, I don't drink alcohol while I work.
In reverse order...

Please point out what I wrote that you take as implying drinking has anything to do with this. The thought did not cross my mind.

I apparently didn't understand your A/B/C comparisons. I thought you were saying that the identical mixes and recalls SOUNDED different to you at different times of the day but we’re not different when listened to side-by-side at the same time of day.
If you are saying that the night mixes and day recalls sound obviously different when both are compared in the morning, OK. Different problem. And from a distance, it isn’t clear what the problem could be. I’ve never encountered anything like it.
One question... if you make a mix to whatever medium you use as storage, and then immediately do a clear and recall, does the recall sound identical to the stored mix? You may have already addressed this.
Old 2nd February 2020
  #7
Gear Maniac
 
jontornblom's Avatar
Quote:
Originally Posted by Bushman View Post
In reverse order...

Please point out what I wrote that you take as implying drinking has anything to do with this. The thought did not cross my mind.

I apparently didn't understand your A/B/C comparisons. I thought you were saying that the identical mixes and recalls SOUNDED different to you at different times of the day but we’re not different when listened to side-by-side at the same time of day.
If you are saying that the night mixes and day recalls sound obviously different when both are compared in the morning, OK. Different problem. And from a distance, it isn’t clear what the problem could be. I’ve never encountered anything like it.
One question... if you make a mix to whatever medium you use as storage, and then immediately do a clear and recall, does the recall sound identical to the stored mix? You may have already addressed this.
Hi Bushman. Apologies! I had edited my original reply so the beer comment was directed to the guy who said it: Radardoug. But too late heh

Yeah, I do realize I didn’t say I had versions to A/B but I had already sent a master to the client, so it felt really obvious to me that when I was setting up for the revision that I would be closely comparing the sound with the recalled settings to the original mix’s master. I guess I should have realized that not everyone knows what the workflow is like when mastering in the analog domain. It’s not the mastering forum after all. So I guess that’s my bad.

Yeah, the settings were recalled and checked several times both evening and morning (although like I mentioned I hadn’t changed a single click from the evening to next morning. the most compelling part is that the settings I logged sounded different at 9pm than they did in the morning on both days.

I agree, it’s weird and I’d really like to figure it out.

Thanks for the continuing dialog on it

Edit: and yes, all my controls are switched and I’ve recalled masters from various periods of time past with exact results every time except the two I’ve mentioned.
Old 2nd February 2020
  #8
Lives for gear
I’m trying to ask, are you comparing a previous version from a digital or analog storage medium to a master “live” off of your board?
If so, it isn’t exactly like-to-like.
Old 2nd February 2020
  #9
Gear Maniac
 
jontornblom's Avatar
Quote:
Originally Posted by Bushman View Post
I’m trying to ask, are you comparing a previous version from a digital or analog storage medium to a master “live” off of your board?
If so, it isn’t exactly like-to-like.
No, it's 24 bit/96 kHz digital files that were passed through the exact same signal chain all three times and are compared to each other in a DAW using mute and solo buttons. The only difference between the versions was time of day/time powered on.

Even if that was the case, the 0.1% difference that is audible between the analog signal and the captured 24/96 audio through the Apogee 2x6 SE converters is much less significant than the 10% darker sound that ended up on the files. These are not real numbers, by the way, just an illustration of how significant the difference was.

Generally speaking, mastering engineers don't have mixing boards in their mastering room.
Old 2nd February 2020
  #10
Lives for gear
I meant the analog part of your chain, not necessarily a console.
Old 2nd February 2020
  #11
Lives for gear
 
Radardoug's Avatar
 

If you have digital files from both states, you can put them in a workstation, phase reverse one file, and then mix them, which will give you the difference between the two files. Then the residual will tell you if there is a difference.
Old 2nd February 2020
  #12
Gear Maniac
 
jontornblom's Avatar
Quote:
Originally Posted by Radardoug View Post
If you have digital files from both states, you can put them in a workstation, phase reverse one file, and then mix them, which will give you the difference between the two files. Then the residual will tell you if there is a difference.
Why would I do that when the difference is obviously audible?
Old 2nd February 2020
  #13
Lives for gear
The difference file is objective proof. A difference that “is obviously audible” to you is hearsay to all of we not-yous.
Some people see fairies in the garden or Bigfoot in the woods. But there hasn’t been objective proof I’m aware of that either exist.
Old 2nd February 2020
  #14
Lives for gear
 
Radardoug's Avatar
 

My understanding from your post is that you have a problem that you are not able to quantify exactly. This is one way of getting you closer to a solution. So why would you not do it?
Old 2nd February 2020
  #15
Lives for gear
Quote:
Originally Posted by Radardoug View Post
My understanding from your post is that you have a problem that you are not able to quantify exactly. This is one way of getting you closer to a solution. So why would you not do it?
What do we not want to know because we can’t handle it?

Cue climactic trial scene from “A Few Good Men”.
Old 2nd February 2020
  #16
Gear Maniac
 
jontornblom's Avatar
Thumbs down

Quote:
Originally Posted by Bushman View Post
The difference file is objective proof. A difference that “is obviously audible” to you is hearsay to all of we not-yous.
Some people see fairies in the garden or Bigfoot in the woods. But there hasn’t been objective proof I’m aware of that either exist.
Old 2nd February 2020
  #17
Lives for gear
Quote:
Originally Posted by jontornblom View Post
Is that you beaming up? Going back to Saturn?
Old 3rd February 2020
  #18
Gear Maniac
 
jontornblom's Avatar
Ok, I’ve put those two guys on my Ignore List.

If anyone has any insight regarding inductor temperature coefficients or any other actual attempt answer to my electronics question, your suggestions are welcome

As a starting point, it’s a well understood phenomenon of inductors that a change in temperature creates a change in inductance primarily due to thermal expansion of the coils, which creates a frequency dependent inductive reactance change as well.

My two questions are

1. If a competent designer would be able to design the inductor based EQ circuits such that these changes are above/below the audible band (depending on the filter topology).

2. If it turns out that temperature changes will affect the audible band, why the audible change in question would only happen at some times and not others, could the extra heat source be the higher mains voltage in the evening that I regularly observe.

Last edited by jontornblom; 3rd February 2020 at 12:32 AM..
Old 3rd February 2020
  #19
Lives for gear
Quote:
Originally Posted by Radardoug View Post
If you have digital files from both states, you can put them in a workstation, phase reverse one file, and then mix them, which will give you the difference between the two files. Then the residual will tell you if there is a difference.
Surely being scientific and quantifying the actual difference that is happening could provide meaningful insight into your electronics problem? Perhaps not to you, but perhaps to others here from whom you are seeking solutions?
Old 3rd February 2020
  #20
Lives for gear
Quote:
Originally Posted by jontornblom View Post
Ok, I’ve put those two guys on my Ignore List.
Please tell him I said thank you.
Old 3rd February 2020
  #21
Gear Maniac
 
jontornblom's Avatar
Quote:
Originally Posted by enginefire View Post
Surely being scientific and quantifying the actual difference that is happening could provide meaningful insight into your electronics problem? Perhaps not to you, but perhaps to others here from whom you are seeking solutions?
Yes, I agree. But the problem with people not understanding what I'm after is that the scientific route I'm taking is asking about inductor behaviour regarding temperature, not whether or not there is a difference in the sound.

Do you understand what I mean? The meandering route this conversation goes something like this:

OP: I saw these strange moving green lights in the northern sky when I was in Alaska. I think they are called Aurora Borealis? What causes that?

Answers: Are you SURE you saw strange moving green lights?
Were you on mushrooms? People can often imagine things when they are tired or stressed. If you didn't take a photo of it, why should we believe you saw them?

So you can see that this has taken a totally different path than it should have and there has been some pretty annoying stubborness that instead of just answering my question, people are questioning whether my query is valid in the first place. They aren't scientific questions, they are questioning the validity of my observations in a forum where we are all supposed to be able to trust what our ears tell us in an A/B test. If we can't trust our ears in an A/B test, aren't we useless as audio professionals?

Now onto what I'm actually after.

In the most basic sense, Inductors connected in series form a low pass filter. The shoulder frequency of the LPF depends on the inductance, which has a frequency dependent impedance, known as Inductive Reactance. If the inductance changes, so does the inductive reactance, and hence the shoulder frequency changes.

That is something that happens. It's no mystery or anything 'voodoo.' The question is about whether designers are able to mitigate these effects through clever circuit desgin or not. That's my question, that's what I don't know. It's totally scientific and is independent of whether or not I heard a change in the source audio. If so, now I have to look into whether the design is correct and if I need to take any measures to avoid the change in shoulder frequency. The secondary question is why this is only happening at night. Is it extra heat from the higher mains voltage?

Does this make sense?

SO I'm asking for someone WHO ACTUALLY UNDERSTANDS MY QUESTION to give me an answer. The internet is rife with people who aren't interested in being helpful, but just trying to prove how smart they are, and that they are smarter than you. I don't have time for that kind of nonsense.

Last edited by jontornblom; 3rd February 2020 at 04:53 PM..
Old 3rd February 2020
  #22
Lives for gear
Quote:
Originally Posted by jontornblom View Post
I'm asking for someone WHO ACTUALLY UNDERSTANDS MY QUESTION to give me an answer.
That’s an SOS to all the therapist-engineers and priest-engineers. Please help this person!
I’ve got to go... my inductors are heating up, and I need to prepare a hot beverage.
Old 3rd February 2020
  #23
Gear Maniac
 
jontornblom's Avatar
Arrow

I've been looking more into this and have found something rather compelling in one sense, but also not compelling in another:

taken from http://npl.csircentral.net/54/1/42.pdf

"It is clear from both the methods that inductors take 10 to 12 hours to stabilize against change in temperature and the temperature coefficient is different for different inductors of same value."

This fits the timeline, which is encouraging. However, whether those changes due to heat are significant enough to produce an audible change in the cutoff/shoulder frequency is still unclear to me. The inductors alone may not be enough but that's not the only possible factor.

The next step will probably be to open the unit and take a measurement of the power rail voltages in the morning and evening. I would guess that the PSU's linear voltage regulator should be taking care of these voltage discrepancies on the input, but there IS something happening so I must keep looking for the cause.

Maybe I'll borrow a small thermocouple/meter from a friend and monitor the internal temperature of the unit as well.

I'm sure there are other variables that I haven't considered. But what else would change besides temperature and voltage?
Old 4th February 2020
  #24
Lives for gear
 

Extremely unlikely that inductor values or power supply variation has anything to do with this. Best thing is to carefully measure for any difference, then base further investigation on what is found.
Old 4th February 2020
  #25
Gear Maniac
 
jontornblom's Avatar
Quote:
Originally Posted by David Kulka View Post
Extremely unlikely that inductor values or power supply variation has anything to do with this. Best thing is to carefully measure for any difference, then base further investigation on what is found.
Hmm. Yeah, I think those differences I’ve been discussing are like 0.001% changes, so I’m skeptic all of that as at least the sole cause.

It sounds like the 6 dB/octave LPF moves down from 15 kHz to 13 kHz. A null test verifies this by listening and looking at span but doesn’t really provide me with any new information than a comparative listen. A null test is just a different way to hear a difference and isn’t really more quantifiable. Unless I’m missing something?

But I’m definitely open further investigation. Any other ideas on what is causing it?

Any ideas on which measurements to perform?
Old 4th February 2020
  #26
Lives for gear
An IQ test?
Old 4th February 2020
  #27
Lives for gear
 
Radardoug's Avatar
 

Are you saying this is entirely repeatable? In which case do some test files with tones. Record tones across the spectrum. Record the same tones when faulty. If you do it carefully, you can then do the null test and it will give you precise results with respect to frequency. Ideally use the same master file, and then do digital copies through the process.
Also, just a general question, do you have any training in electronics?
Old 5th February 2020
  #28
Lives for gear
 

No human can hear a .001% change. 1%, sometimes. Be an engineer and measure this, then let us know what you find. If the difference is really there then plan old steady state tones will probably reveal it. Of maybe try pink noise. Measure the differences between the two signals when nulling them.

Guessing and estimating will probably just take you down a path of wrong assumptions and missed clues.

Last edited by David Kulka; 5th February 2020 at 08:33 AM..
Old 5th February 2020
  #29
Lives for gear
 

Many years ago I owned a powerful solid-state amplifier from a manufacturer that's well-renowned for their stellar bench performance and traditional engineering prowess . . . but after a time I came to the idea that its sonic performance was inconsistent. From listening, I concluded that it was warmup related, but for some reason it seemed to be different depending on whether or not I simply turned it on with no input, or played music through it right away . . . putting my hand on the heatsinks across the back I could tell that it was at different temperatures at different times, but that was about it. I had enough of a listening impression over enough iterations to be somewhat confident that I might be hearing something, so I swapped it out for a (completely different) amplifier and put the big heavy beast in a corner of my shop, where I dorked around with it and made measurements here and there over the course of about a year . . .

This amp had IIRC sixteen output devices, spread across four separate identical heatsink extrusions, for EACH channel. There were also two driver transistors per channel, placed one each on half the heatsinks . . . and a single bias-sensor (Vbe-multiplier) transistor placed on one of the heatsinks that had a driver. So there were three classifications of heatsinks -- two with four output devices only, one with four outputs and a driver, and one with four outputs, a driver, and a bias sensor. On top of this, the right-channel heatsinks were directly next to the large potted power transformer.

Anybody who's spent time designing/optimizing discrete bipolar audio amps can see where this is going . . . the optimum bias window for a class B amp can be pretty narrow, and there are a total of six sets of junctions that affect it thermally (two pre-drivers, two drivers, and two groups of outputs) -- and thermal compensation for all of it relies on the junction of a single TO-92 transistor. The pre-drivers' dissipation is very small, and they're buried within the chassis, the driver dissipation is pretty much constant, and the outputs' dissipation is highly dependent on the signal and load. Chasing down this road on the test bench revealed that it was possible to put the amp into a few different steady states, where the outputs were either all a bit over-biased, all under-biased, or a combination of the two, sometimes different between the two channels . . . and given the total mass of the whole thing (it weighed over 100 pounds), the thermal time-constants involved were extremely long, on the order of several hours. Even though at no time did the amplifier not "meet specification", the differences were clear and repeatable visually on the distortion residual, once an appropriate test methodology had been devised, and I knew what I was looking for.

The question of audibility is also a complex one, and if one is honest with oneself, it cuts both ways. This amplifier was expensive, and it had an ethos that exuded quality, longevity, and stability . . . looking at it through the lens of boilerplate audio amplifier engineering, you'd be hard-pressed to find any sort of corner-cutting in its design or construction. There was definitely a period of time where I thought many times "hmmmm . . . not sounding like I remember", but then swept it under the observational rug with "naaawwww . . . it can't be." Eventually I started putting my hand on the top of the thing on a regular basis to check temperature, and at some point there was this "Es Muss Sein!" conclusion, and the balance tipped and I had problems un-hearing it. When this happened, the machine became useless to me for its intended function, as I could no longer be in the mental place where I had confidence in its performance. It's certain that by the end I was hearing things that weren't there . . . but I'm also confident that on some level, under some conditions, I was noticing a genuinely audible artifact . . . otherwise I would have never gone down the road of analysis in the first place.

My point in this story is that there are a whole series of long, tedious steps between the observation of the phenomenon, and having even a reasonable, technically valid working hypothesis of what might be, or have been, the causal mechanism. BTW from a scientific standpoint, a reasonable hypothesis is all that I ever achieved with that amplifier - I didn't even really set a toe down on the first inch of the long journey to what could be considered "proof". If you do have analyzable recordings that correlate with what you observed, that's certainly a leg up . . . but as far as understanding it goes, you simply don't have enough data to get out of the "wild speculation" phase.
Old 5th February 2020
  #30
Lives for gear
Quote:
Originally Posted by kirkus View Post
as far as understanding it goes, you simply don't have enough data to get out of the "wild speculation" phase.
You are truly a pro for not naming the amp you came to not like.
It is acceptable for anyone to stop using any piece of gear, for whatever reason or completely insane, unsupportable whim.
It isn’t necessary to go on GS and try to get the community to support and help prove your wild-ass science (fiction) constructions. If Alcoa is the wrong tinfoil for your mixing hat, switch to another brand and get back in the studio. You don’t have to prove anything to anyone.
But if you are trying to prove something to the community, the word “proof” implies that you will provide convincing evidence that has at least some scientific validity.
And by “you”, I do not mean Kirkus. He is doing what we all do. I don’t usually have time for scientific trials. If gear doesn’t sound right or feel right, even in an undefinable way, I explore door #2 .
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump