[KeithPhantom]

The problem is that the difference he is talking about is less than the manufacturing tolerances of mid range cans. He could find a model he likes, but the copy he ends up with might be slightly off. I think the only solution is to buy somewhere that you can return them for exchange until you find a set that is "just right". It's like the Princess and the Pea.

Interesting, are we talking about >3 dB differences between channels?

 

[bigshot]

He says it's 1.5 dB and only below 100 Hz.

 

[KeithPhantom]

He says it's 1.5 dB and only below 100 Hz.

Technically, we are pretty much deaf at those frequencies, and by the amount he is citing, even with test signals would be pretty hard to hear, and even harder with music. I would recommend to request for a replacement and then perform a blind test.

 

[bigshot]

Or maybe his measurements are off.

 

[castleofargh]

The problem with IEM is that the amount of subs will be directly impacted by how good the seal is. So measuring them in a coupler might not show what's actually happening in the ear.

What's sure is that we're typically a lot less sensitive to low freqs amplitude compared to mid range. I remember easily and consistently coming within 0.5dB at 3 or 4kHz when trying to EQ by ear with test tones, while often having variations of up to 4dB in the subs. So as I already said to @carmatic, I at least would not expect to notice small imbalance in the low end while casually listening to music. And last time I EQed to match a graph he provided, I indeed couldn't really tell by ear.
So it's a puzzling issue that he's having. But at the same time, with Etymotic IEMs it's trivial to insert them into the opposite ears or even to swap the cable's sides(and I or others probably told him to try when he first mentioned the issue in the ER4 thread). so I think we can rule out something like a difference in insertion because of the shape of one ear, or the amp being the main cause of a big imbalance.

 

[bigshot]

He said it's the same with two different models and brands which is even more puzzling. I think he is making a mistake somewhere. Hard to know where without being there administering blind tests to him.

 

[carmatic]

He said it's the same with two different models and brands which is even more puzzling. I think he is making a mistake somewhere. Hard to know where without being there administering blind tests to him.

Just to clarify, I have not made a measurement of my new Audio Technica IEM's yet due to being away from my computer (I am using my phone)
my 1.5dB correction in the 30-60hz range was when I was using my ER-3XR , I will need to bring my ATH-E70 to my computer to be able to apply the equalizer to it and see what it looks like

 

[carmatic]

The problem is that the difference he is talking about is less than the manufacturing tolerances of mid range cans.

you mention 'mid range' ... does this mean that tighter tolerances do exist at higher price points? I really want to avoid the situation where I put money into something which costs many times more than what I currently own, only to find that they are just as unmatched
I'd like some kind of guarantee, or a specification where the drivers are matched down into the low frequencies

 

[bigshot]

you mention 'mid range' ... does this mean that tighter tolerances do exist at higher price points?

I was part of a group that gave feedback to a company designing a model of headphone in the $1200 range. I got to speak with the designer of the cans. He told me that the difference between high end cans and ones in the $600 range wasn't better components or different types of designs, it was manufacturing tolerances. Every set of headphones that comes off the line varies a little bit, and they are tested to make sure they measure within the acceptable range. He said that most midrange headphones can deviate up to 3dB, which is the just detectable difference (JDD) for response deviations under music. The high end cans we were working with had a tolerance range of 1dB. He said the added cost was to cover all of the rejects that came off the line and couldn't be sold.

So the answer is, yes higher priced headphones have tighter tolerances. But even there, it is still possible that you could end up with a 1.5dB difference between two different drivers, and the odds would be that the deviation would be in either the very highest frequencies, or the very lowest. That small of a difference isn't supposed to matter. I'm pretty picky about response, and I only worry about differences of 2dB in the sweet spot in the upper mids. Below 100Hz, it isn't critical except with speakers where you get into interactions with the room. And if you jump from one model of headphones to another, the overall response curve is going to change drastically, and the overall response is more important than tiny variations from unit to unit.

My advice is to either not worry about it, or buy your headphones from a dealer who lets you audition the specific set you are buying. Then listen to them and if they are balanced for you by the luck of the draw, go ahead and buy them. I don't think you'll find any headphones rated to be accurate from sample to sample to 1.5dB below 100Hz. Your only guarantee would be to check them before you buy them.

 

[KeithPhantom]

does this mean that tighter tolerances do exist at higher price points?

HifiMAN tries to match drivers within 0.1 dB from each other, you would like to look into that.

 

[bigshot]

I think you have the decimal point a digit off there. If they say that, they are prevaricating. Unless they measure everything and wait for perfect matches to come up by random selection. They must have a warehouse full of drivers waiting for a match.

I'm sitting here thinking about this, and I can't fathom how you'd go about even measuring cans coming off the line with that kind of accuracy, much less manufacture them to a tolerance like that.

 

[KeithPhantom]

I think you have the decimal point a digit off there.

As I said, they try and they do a good job at it. This is a graph of the Susvara, and they seem to do a good job:

 

[bigshot]

There is a 2dB difference there at 500Hz. and a bigger one between 700 and 800. How accurate is measurements below 100Hz? I would think that is the hardest to measure.

 

[KeithPhantom]

There is a 2dB difference there at 500Hz. and a bigger one between 700 and 800. How accurate is measurements below 100Hz? I would think that is the hardest to measure.

But on average they are pretty well matched, I would say below 1 dB for most of the frequency range shown. It is extremely hard to impossible to have perfectly matched drivers just because of the physics involved. Also, remember that measurements have resolution limits (this rig had a limitation with THD, It cannot measure below 0.02%).

 

[bigshot]

I would say if you don't listen to test tones a 3dB deviation is acceptable. Do they say "averaged across the whole response"? Because obviously if you factored in above 10kHz, the whole average would be thrown off. I'm trying to remember what the Oppo guy told me when he said they were shooting for less than a 1dB deviation. He might have said below 10kHz or something like that. I don't remember. I wish I had capped that discussion forum (but of course I was contractually obliged to not do that.)