Review and Measurements of New JDS Labs Atom Headphone Amp

[bobbooo]

No disagreement here. But I think Dr. Olive's point is that a competent headphone that can be effectively EQ to produce world class sound can be had for about $100 USD.

How about $10? https://clarityfidelity.blogspot.com/2017/03/sony-mh755-iem.html

Check out those low distortion measurements! They hardly need any EQing either, they're already very close to the Harman target.

 

[bobbooo]

I read that before too and got confused. How resolving a headphone is isn't really represented in just a frequency response graph. THD vs Frequency at various power levels (SPLs) is kind of important imo as well. There are other metrics that can be measured that I'm not going to post since I have less understanding of them. The point is I think Dr. Sean Olive is nuts

As I understand, headphones are almost entirely minimum-phase, so any other metric (apart from distortion) you measure will already be represented by the frequency response (e.g. the impulse response is just the Fourier transform of the frequency response). Even multi-driver balanced armature IEM's with all their crossovers are minimum phase, e.g. the 5-driver BGVP DM6 has a group delay below the threshold of audibility over the entire frequency range, as measured by Rtings.com: https://www.rtings.com/headphones/1-2/graph#786/2074

 

[sychan]

Here's Frost's case:

https://www.thingiverse.com/thing:3230306

I'm a little biased, but if you'd prefer a case that matches your Atom:

https://www.thingiverse.com/thing:3824172

You can get either printed by clicking "Order This Printed" and following the prompts. Personally I use Treatstock and black PLA.

From what I've seen, the JDS Atom's top is plastic. I'm a little surprised that nobody has designed a "mezzanine" enclosure for the KTB+Atom that just replaces the top of the Atom enclosure with an extra tall structure that covers up the KTB, and screws into the Atom enclosure base. That way you have a nice clean 1 piece component. I've been a little tempted to pick up a KTB and Atom and mod your case myself, but once you add up the price of a KTB and JDS Atom, it makes more sense to just pickup a DX3pro.

 

[Sean Olive]

Lets say I respectfully disagree with dr Olive here.
There is much more to headphones than FR.
Cone break up, resonances, acoustic nulls to name but a few obvious ones would be hard to simulate.
Not to mention comfort, feel, proper positioning on the head.

I do think they are doing the whole industry and themselves a good service though.
If anything, a lot of folks aren't afraid to use the word 'EQ' any more.

But "cone break up, resonances, acoustic nulls" . all manifest themselves as artifacts clearly visible in the frequency response as peaks and dips. The only thing the frequency response doesn't reveal is non-linear distortion and non-minimum phase effects. Most headphones we've measured behave as minimum phase devices up to 1-3 kHz above which cone break up and reflections wreak havoc. And we've done a study on the audibility of non-linear distortion on a small number of headphones and the audible effects are quite subtle at normal listening levels unless the headphone is poorly designed. see http://www.aes.org/e-lib/browse.cfm?elib=17441

 

[solderdude]

But "cone break up, resonances, acoustic nulls" . all manifest themselves as artifacts clearly visible in the frequency response as peaks and dips. The only thing the frequency response doesn't reveal is non-linear distortion and non-minimum phase effects.

Yes, I fully agree. Also that upto 1-3kHz there isn't much cone break-up.
My position in this is that while one can EQ frequency response to mimic that of other headphones one can get the tonal balance pretty close but yet one cannot get a HD201 to deliver the same sound quality as that of a HD650 or HD800S for instance despite the tonal balance being EQ'ed the same there will still be audible differences due to cone break up, resonances, acoustic nulls etc.

What I don't agree with is that one can take 'a' $ 100.- headphone and make it sound like a $ 1000.- headphone.
Some headphones react quite well to EQ but some simply don't.
So to me... Some $100.- headphones can be EQ'ed to sound as good as some $1000.- headphones but it is not a rule for all headphones.

Indeed while some headphones show considerable distortion at test levels (between 90 and 100dB SPL) the distortion is usually far less at listening levels. I consider distortion above 5% in the bass audible. For higher frequencies below 0.1% should be possible. Above 1% usually is not desirable.

 

[Tks]

Yes, I fully agree. Also that upto 1-3kHz there isn't much cone break-up.
My position in this is that while one can EQ frequency response to mimic that of other headphones one can get the tonal balance pretty close but yet one cannot get a HD201 to deliver the same sound quality as that of a HD650 or HD800S for instance despite the tonal balance being EQ'ed the same there will still be audible differences due to cone break up, resonances, acoustic nulls etc.

What I don't agree with is that one can take 'a' $ 100.- headphone and make it sound like a $ 1000.- headphone.
Some headphones react quite well to EQ but some simply don't.
So to me... Some $100.- headphones can be EQ'ed to sound as good as some $1000.- headphones but it is not a rule for all headphones.

Indeed while some headphones show considerable distortion at test levels (between 90 and 100dB SPL) the distortion is usually far less at listening levels. I consider distortion above 5% in the bass audible. For higher frequencies below 0.1% should be possible. Above 1% usually is not desirable.

You can get it close enough. And in some cases, better than $1000+ headphones in the sonic realm. The only reason you will never get the same exact sound, is due to driver construction, and headphone shell design (factors such as cavity size, driver angle when wearing the headphone, ear pads, headphone housing itself, size of the driver and rigidity).

All this contributes to a different “sound” at certain sound pressure levels, especially once distortion is added to the mix. So your statement about “not being a rule” makes sense in the same way no two different model cars can ever perform the same simply due to the physics of the sum of the parts when all is put together. It basically goes without saying, and as such, there was no need to say it as no one disagrees with that portion. What was disagreeable was the implication from the tone of your prior message that made it -seem- like FR isn’t pretty much the main contributing factor. And as if there are many many other properties that you allude to that can’t be taken into account for whatever reason, or are unknowns by adding in “comfort” as some sort of sonic aspect?

And the last comment about EQ, the people “afraid” of EQ only need be afraid of distortion once you EQ “up” instead of “down” in order to get he sound you want (as most software in consumer devices don’t have a pre amp setting which leads many laymen to do things like boost bass as much they can). Though I think you didn’t mean clipping issues when you spoke of people talking about EQ, I think you meant those odd paranoid people of some ignorant bygone era akin to digital detesters of which had more credence to exist with digtial’s advent, but look like living fossils when they still persist in their stance on an ideological level now almost in the year 2020.

 

[solderdude]

You can get it close enough. And in some cases, better than $1000+ headphones in the sonic realm. The only reason you will never get the same exact sound, is due to driver construction, and headphone shell design (factors such as cavity size, driver angle when wearing the headphone, ear pads, headphone housing itself, size of the driver and rigidity).

That's what I said.

What was disagreeable was the implication from the tone of your prior message that made it -seem- like FR isn’t pretty much the main contributing factor. And as if there are many many other properties that you allude to that can’t be taken into account for whatever reason, or are unknowns by adding in “comfort” as some sort of sonic aspect?

FR is the most important factor. What I disagree with is that even when you EQ headphones to sound the same tonally the one with the better driver may still have audible properties (caused by what I mentioned) that makes them still different while tonally corrected.
So the $100 with EQ to match that of a $1000 may still not sound as good.
Some EQ'ed $ 100.- may even sound better than some $ 1000.- so it isn't a rule.

Comfort isn't a sound issue but even when you EQ a $100.- with crappy comfort headphone to sound like a (more comfortable) $ 1000.- you may have emulated the sound but not the headphone.

And the last comment about EQ

EQ does not need to clip. What I meant is suppose you EQ a headphone to measure exactly flat on the exact same measurement rig,
That applied EQ may well be not what is needed IRL, It will only be perfect for that particular measurement rig.
It may well be f'ed up for real ears and when measured on any other test rig may show dips and peaks.

I believe in EQing headphones... better yet I insist.
That doesn't mean others like it as well and prefer the 'original tuning as it was intended'.

 

[Quizel]

Is it fair to say there is a limit though to price to performance on headphones? Like can a $1,000 (or your favorite one for less) be just as good as or better than a $3,000+ headphone. My gut tells me yes. For example, anything above $1,000 becomes preference and materials/build but I've never actually experience it. And if I'm wrong, perhaps the number is $4,000 is top notch performance and anything above it doesn't improve audibly.

I could also image you simply can't put a number on it because the perfect headphone for someone might have a characteristic only a specific extremely expensive headphone can do. It's just harder to imagine with my limited experience. Granted, discovering what you like and dislike is half the battle and can change over time.

I think what I really meant to say is that there's significant diminishing returns sooner or later. The critical point where this happens varies from person to person so it's hard to put a number on it. I mentioned $1,000 above probably because I'm not rich and it isn't that important to me given I'm satisfied with what a have- a few years from now I may have a completely different perspective.

 

[watchnerd]

Is it fair to say there is a limit though to price to performance on headphones? Like can a $1,000 (or your favorite one for less) be just as good as or better than a $3,000+ headphone. My gut tells me yes. For example, anything above $1,000 becomes preference and materials/build but I've never actually experience it. And if I'm wrong, perhaps the number is $4,000 is top notch performance and anything above it doesn't improve audibly.

I could also image you simply can't put a number on it because the perfect headphone for someone might have a characteristic only a specific extremely expensive headphone can do. It's just harder to imagine with my limited experience. Granted, discovering what you like and dislike is half the battle and can change over time.

I think what I really meant to say is that there's significant diminishing returns sooner or later. The critical point where this happens varies from person to person so it's hard to put a number on it. I mentioned $1,000 above probably because I'm not rich and it isn't that important to me given I'm satisfied with what a have- a few years from now I may have a completely different perspective.

There are diminishing returns to everything in audio.

As for what is "good enough", remember that when headphones get used as part of the recording engineering and mixing process, it's often with cans that are far, far cheaper than $1K, often with durability and replacement parts being one of the criteria, with mid-tier AKG/Senn being popular, but also a lot of fairly inexpensive Sony and AT cans, too.

 

[bobbooo]

Is it fair to say there is a limit though to price to performance on headphones?.

There is no correlation between price and performance of headphones at all: https://www.audioholics.com/loudspeaker-design/price-sound-quality-headphones, so the limit is technically $0 e.g. these earphones (black curve) that come free with current Samsung smartphones very closely match the Harman target (light blue curve): https://github.com/jaakkopasanen/Au...0/harman_in-ear_2017-1/Samsung EO-IG955 (AKG)

 

[bobbooo]

But "cone break up, resonances, acoustic nulls" . all manifest themselves as artifacts clearly visible in the frequency response as peaks and dips. The only thing the frequency response doesn't reveal is non-linear distortion and non-minimum phase effects. Most headphones we've measured behave as minimum phase devices up to 1-3 kHz above which cone break up and reflections wreak havoc. And we've done a study on the audibility of non-linear distortion on a small number of headphones and the audible effects are quite subtle at normal listening levels unless the headphone is poorly designed. see http://www.aes.org/e-lib/browse.cfm?elib=17441

One thing I'm curious about is whether it would be possible with EQ to simulate headphones of one form factor using headphones of another? E.g. EQing a closed-back pair to sound like an open-back pair, or even in-ears to sound like open-back over-ear headphones? As open-back headphones leak out and let in sound to and from their environment (which can change), and (presumably) from one ear to the other (acoustic crosstalk) via room reflections etc. Do these inter-aural and room-dependent characteristics fall under non-minimum phase effects? If so, does this mean it would not be possible via equalization techniques on closed-back headphones to recreate these effects, often claimed to contribute to the wider, more natural 'soundstage' (i.e. loudspeaker effect) of open-back headphones?

 

[JohnYang1997]

One thing I'm curious about is whether it would be possible with EQ to simulate headphones of one form factor using headphones of another? E.g. EQing a closed-back pair to sound like an open-back pair, or even in-ears to sound like open-back over-ear headphones? As open-back headphones leak out and let in sound to and from their environment (which can change), and (presumably) from one ear to the other (acoustic crosstalk) via room reflections etc. Do these inter-aural and room-dependent characteristics fall under non-minimum phase effects? If so, does this mean it would not be possible via equalization techniques on closed-back headphones to recreate these effects, often claimed to contribute to the wider, more natural 'soundstage' (i.e. loudspeaker effect) of open-back headphones?

Non linearities is the issue here. And there are limitations of dynamic range in the digital domain.
First of all there shouldn't be any null in the system. Damped peaks are allowed. Then the distortion will be rising if we are trying too hard to equalize eg open back in the bass. The driver may have thd shoot up over 1% even 10% and may hit clipping, which is bad. Also only single driver headphones are considered minimum phase. You can plot excess phase by using the measurred phase subtracted by the calculated minimum phase from the magnitude response. Most multidriver iems are far off in this respect. Best ones have gental rise or fall near the high frequency. The worst ones will have phase jumping all over the place, or more than 90° phase shift across the frequency range. Those can't directly be corrected by minimum phase filter. In some cases still fixable.
Then here we comes to equalization. Measurements don't give you accurate representation at high frequency over 6khz. In many cases you can hear a peak at 6khz but on the measurements it's at 8khz. Or the measurements show dip at certain frequencies but doesn't exist when putting on the head and perform sine sweep. You may likely end up getting misaligned peak in the eq. And the peaks can be very dependent to the insertion depth for iems, And the position on the head for overear.
Ideally it's all great and possible. But in real life, only a fraction of the time it works.

 

[bobbooo]

Ideally it's all great and possible. But in real life, only a fraction of the time it works.

As I already linked previously, Dr Olive and his team showed that yes, it does work a 'fraction' of the time, that fraction being 85% in their study: https://seanolive.blogspot.com/2016/04/a-virtual-headphone-listening-test.html

That's very good for an initial attempt, and Dr Olive suggests reasons for the remaining error and ways to improve it here:

"The differences between virtual and standard test results we believe are in part due to nuisance variables that were not perfectly controlled across the two test methods. A significant nuisance variable would likely be headphone leakage that would affect the amount of bass heard depending on the fit of the headphone on the individual listener. This would have affected the results in the standard test but not the virtual one where we used an open-back headphone that largely eliminates leakage variations across listeners. Headphone weight and tactile cues were present in the standard test but not the virtual test, and this could in part explain the differences in results. If these two variables could be better controlled even higher accuracy can be achieved in virtual headphone listening."

You could maybe argue that the remaining 15% error in this headphone virtualization method via EQ could also in part be due to non-minimum phase effects and the other difficulties you raised, but I can only foresee these problems being negated (if partially) through further refinement of the technique and controlling for the variables Dr Olive proposed. Of course, you will never get zero error, but I can see it getting close, within the limits of audibility.

 

[JohnYang1997]

As I already linked previously, Dr Olive and his team showed that yes, it does work a 'fraction' of the time, that fraction being 85% in their study: https://seanolive.blogspot.com/2016/04/a-virtual-headphone-listening-test.html

That's very good for an initial attempt, and Dr Olive suggests reasons for the remaining error and ways to improve it here:

"The differences between virtual and standard test results we believe are in part due to nuisance variables that were not perfectly controlled across the two test methods. A significant nuisance variable would likely be headphone leakage that would affect the amount of bass heard depending on the fit of the headphone on the individual listener. This would have affected the results in the standard test but not the virtual one where we used an open-back headphone that largely eliminates leakage variations across listeners. Headphone weight and tactile cues were present in the standard test but not the virtual test, and this could in part explain the differences in results. If these two variables could be better controlled even higher accuracy can be achieved in virtual headphone listening."

You could maybe argue that the remaining 15% error in this headphone virtualization method via EQ could also in part be due to non-minimum phase effects and the other difficulties you raised, but I can only foresee these problems being negated (if partially) through further refinement of the technique and controlling for the variables Dr Olive proposed. Of course, you will never get zero error, but I can see it getting close, within the limits of audibility.

I'm talking about 5% error here.

 

[bobbooo]

I'm talking about 5% error here.

Not sure what your point is - are you saying 5% error is too high, or that's what we should be aiming for?

 

[JohnYang1997]

Not sure what your point is - are you saying 5% error is too high, or that's what we should be aiming for?

This means Olive's paper is irrelevant to my statements. The eq they've done are too far off in my standard.
The 15% of difference is able to differentiate shit from good or OK from extremely well.

 

[solderdude]

As open-back headphones leak out and let in sound to and from their environment (which can change), and (presumably) from one ear to the other (acoustic crosstalk) via room reflections etc.

Perhaps when one is on a really small toilet with only one channel playing loudly you might just pick up something on the not playing channel.
Most likely bone conduction is worse than sound bouncing of walls meters away.

The ear is just 3cm away from the driver. When a wall is 2m away from you on both sides the sound has to travel 6m to reach the other ear.
This means the sound from the driver on the right side arrives at the left side at approx -75dB. (6dB attenuation when the distance to the source is doubled)
Assuming there is absolutely no damping going on in the rear side of the headphones nor from the walls.
I don't think anyone can hear that when music is playing in both ears.

 

[bobbooo]

This means Olive's paper is irrelevant to my statements. The eq they've done are too far off in my standard.
The 15% of difference is able to differentiate shit from good or OK from extremely well.

For some headphones the agreement was more like 95%, meeting your criteria. Saying Olive's paper is irrelevant to your argument is like a theoretical physicist dismissing early experimental evidence for the Higgs particle as being irrelevant to their argument that it can't exist, because the experimentalists' confidence level for its existence was too low. The experimentalists were eventually proved right with further experiments that built upon and confirmed their previous work. That's how science works. I'm not saying Olive will definitely be proved right, but this study is very promising for an initial attempt, and warrants further investigation in order to confirm (or deny) the efficacy of headphone virtualization.

 

[JohnYang1997]

For some headphones the agreement was more like 95%, meeting your criteria. Saying Olive's paper is irrelevant to your argument is like a theoretical physicist dismissing early experimental evidence for the Highs particle as being irrelevant to their argument that it can't exist, because the experimentalists' confidence level for its existence was too low. The experimentalists were eventually proved right with further experiments that built upon and confirmed their previous work. That's how science works. I'm not saying Olive will definitely be proved right, but this study is very promising for an initial attempt, and warrants further investigation in order to confirm (or deny) the efficacy of headphone virtualization.

It doesn't matter. All models from etymotic are less tham 6% different. And they do all sound quite different. Not even mentioning these are very well behaved iems on the market.
What Olive proved was only merely: EQ can improve sound quality. It's you that stretched the conclusion to far, my friend.

 

[bobbooo]

Perhaps when one is on a really small toilet with only one channel playing loudly you might just pick up something on the not playing channel.
Most likely bone conduction is worse than sound bouncing of walls meters away.

The ear is just 3cm away from the driver. When a wall is 2m away from you on both sides the sound has to travel 6m to reach the other ear.
This means the sound from the driver on the right side arrives at the left side at approx -75dB. (6dB attenuation when the distance to the source is doubled)
Assuming there is absolutely no damping going on in the rear side of the headphones nor from the walls.
I don't think anyone can hear that when music is playing in both ears.

Yes any effects would be very small (if audible). I've just noticed Olive's study did virtualize closed-back headphones using open-back ones with good results. I don't think they tried the other way round though. It would be interesting to if another study this way round produced the same results.

Good point about bone conduction. That might make virtualizing over-ear headphones using in-ears harder over-ears likely conduct some sound inter-aurally through the skull, whereas in-ears don't