Head-Fi.org › Forums › Equipment Forums › Headphones (full-size) › Oppo PM-1 Planar Magnetic Headphone Impressions Thread
New Posts  All Forums:Forum Nav:

Oppo PM-1 Planar Magnetic Headphone Impressions Thread - Page 15

post #211 of 2885
Achieving flat response with headphones is easier to say than achieve though. In practice, it often ends up a trade off between the response lively and open enough sounding while still appropriatly damping driver and acoustic resonances that will abound in the mid-frequency range and up.

I hear what you say about masking effect bigshot but there's a lot more to it, the transient response of the phone cannot be neglected for instance. I don't think it's that easy to prevent spurious reflections from the baffles, grills and such (not necessarily easily visible in the frequency response function) , that certainly can affect perception of image (and a reason why very open phones like the hd800 have this large perceived stage width associated to them).

Anyhow, looking forward to more impressions, esp. Relative to other gear!

Arnaud
post #212 of 2885
Quote:
Originally Posted by Maxvla View Post

The problem though is the compensation curve. Who's to say what is flat after the curve? Whose curve do you use; Tyll's, Golden Ears', Purrin's, Olive's? Headphones interact differently than a pressurized room so you can't just expect flat from a headphone like you can from speakers in a properly setup system. The tweakability of speaker placements, room construction materials, reflection management (either via DSP or acoustically) and more are all things that are mostly not transferable to headphones as they are all made the same way to fit a common denominator. Sure you can mod your headphones like Anax mod HD800, jergpad HE500s, etc., but those are not nearly comparable.

Also, aren't speaker systems calibrated with open air microphones and headphones calibrated with simulated ears embedded in a dummy head? If you were to measure your microphone calibrated speaker rig with a dummy head I doubt you'll find the result still flat.

 

Olive's can be crossed out.  The goal of that research was not to find out what "flat" was, rather they wanted to know what would be the preferred sound signature.  They showed that a flat neutral wasn't the preferred sound signature for general listening.  The standard Diffuse Field compensation is the most current for what we have that is going to be flat and has research backing it.  Tyll uses a standard DF compensation, so does Rin (I do as well when I do my rough measurements).  Golden Ear's does their own thing.

 

A lot of people make the statement that DF compensation is not flat, and they are actually right in that statement when they listen to music that has been mastered.  The more poorly recorded it is, the less flat it will sound, and the brighter it will sound (this is the idea of something that is perfectly neutral being called unforgiving).  I can see Olive's graph compensating for some of the lost warmth and bass as well as toning down the overbearing treble that results from compression (one of the reasons why might be preferred over DF). 

 

As for your last few statements about how speakers are measured, that's why the DF compensation was created, to address that exact phenomenon.  It compensates for the fact that the sound has to travel through the ear canal as well as other structures. 

post #213 of 2885
Quote:
Originally Posted by bigshot View Post

The response is flat out to 12kHz. From there it starts rolling off. We analyzed the tone sweeps by ear, so at some point our own ears probably were rolling off too. But the important thing about the Oppo's response is that it is flat from 28Hz all the way to 12kHz with less than 3dB variation anywhere in the range. Flat all the way from half an octave from the edge of human hearing on both ends. That is really remarkable and it nails everything you are going to be able to hear in music, at least with human ears!
Is this before or after compensation? Either way, I doubt it. I look forward to Tyll's and Purrin's graphs showing a certainly not flat response.
Quote:
Originally Posted by tinyman392 View Post

As for your last few statements about how speakers are measured, that's why the DF compensation was created, to address that exact phenomenon.  It compensates for the fact that the sound has to travel through the ear canal as well as other structures. 

Yes, I mostly said that to show how Bigshot is comparing apples to oranges.
Edited by Maxvla - 4/5/14 at 6:58pm
post #214 of 2885
Quote:
Originally Posted by Maxvla View Post


Is this before or after compensation? Either way, I doubt it. I look forward to Tyll's and Purrin's graphs showing a certainly not flat response.
Yes, I mostly said that to show how Bigshot is comparing apples to oranges.

 

Ah, gotcha'.  Yeah, you can't really compare graphs unless you know what compensation, if any, is used...  True that! 

post #215 of 2885
Quote:
Originally Posted by Maxvla View Post

Any chance you'll be at the May 4th Dallas meet with these?

I hadn't heard about this meet. I might attend... :-)

Now to look for the thread.

Thanks!

Mike
post #216 of 2885
My buddy is a speaker designer who is working on a new system for achieving flat response from 20 to 20 in live venues, and I've spent months and months carefully equalizing my speaker system by ear, so we know a little bit about what flat sounds like.

When we tested the response of the Oppos, we didn't measure using microphones or apply stock curves. My sound engineer buddy and I did tone sweeps a couple octaves at a time working from low to high, applying EQ corrections to flatten it out as we went. Then we went back through balancing at a higher volume level. Then we checked different spots in the spectrum to others. We double checked each other all the way. Everything was based on listening. Our accuracy might vary up to 1dB or so because we are human not machines, and at the very highest range, our ears might have rolled off. But in broad strokes, it's close enough for government work.
Edited by bigshot - 4/5/14 at 7:06pm
post #217 of 2885
Quote:
Originally Posted by bigshot View Post

My buddy is a speaker designer who is working on a new system for achieving flat response from 20 to 20 in live venues, and I've spent months and months carefully equalizing my speaker system by ear, so we know a little bit about what flat sounds like.

When we tested the response of the Oppos, we didn't measure using microphones or apply stock curves. My sound engineer buddy and I did tone sweeps a couple octaves at a time working from low to high, applying EQ corrections to flatten it out as we went. Then we went back through balancing at a higher volume level. Then we checked different spots in the spectrum to others. We double checked each other all the way. Everything was based on listening. Our accuracy might vary up to 1dB or so because we are human not machines, and at the very highest range, our ears might have rolled off. But in broad strokes, it's close enough for government work.

 

How did you test this accuracy range?  I doubt that you're that accurate.  The fact that this entire thing was done subjectively null-voids your argument that something is flat.  This isn't the first time on Head-Fi we've had people state that something was perfectly flat/neutral only to find out months later, after proper measurements were taken, that they were off by a long shot.  No matter how experienced they were.  Our hearing is not stable.  It is dynamic and ever changing.  That's the problem of subjectively measuring something. 


Edited by tinyman392 - 4/5/14 at 7:12pm
post #218 of 2885

This is the oddest appreciation thread I've ever seen :blink:

post #219 of 2885

im wondering why oppo needed ANYONE to tell them hey, flatter is better. we ran sweeps since you didn't and we're seeing dips and peaks that shouldn't be there.

post #220 of 2885
Quote:
Originally Posted by tinyman392 View Post

How did you test this accuracy range?

In the past I ran some tests on myself to determine my ability to hear different volume levels with test tones and music. With tones, my accuracy was below 1dB throughout the core frequencies. With music, my threshold was about 3dB. My sound engineer friend's ears are definitely better than mine. He equalizes with tone sweeps for a living. I think I'm pretty safe saying we are accurate to +/- 1dB in frequency ranges down to an eighth octave or so, but that is just a guess. It might be a little bit worse or better. Double those thresholds and it's still the same. Close enough for government work.

In the beta group, one of the participants ran a microphone based response that looked about the same as what I found, but then he started talking with the head Oppo engineer about the proper way to measure response and it all started flying over my head. I use tones and ears to tune. It's more straightforward, and ultimately ears are what I listen with. It isn't as accurate, but you can figure out what matters.
Edited by bigshot - 4/5/14 at 7:40pm
post #221 of 2885
Quote:
Originally Posted by DoctaCosmos View Post
 

im wondering why oppo needed ANYONE to tell them hey, flatter is better. we ran sweeps since you didn't and we're seeing dips and peaks that shouldn't be there.


+1

post #222 of 2885
Quote:
Originally Posted by DoctaCosmos View Post

im wondering why oppo needed ANYONE to tell them hey, flatter is better. we ran sweeps since you didn't and we're seeing dips and peaks that shouldn't be there.

I can't speak for Oppo, but right here in this thread are folks who seem to not believe that flatter is better!

I think a lot of compromises are made that shouldn't be made, and they are brushed away with an absolutist brush. "You can't know down to .0001% so it doesn't matter."

Oppo seemed to me to be genuinely trying to make the best headphones they could. After the first beta we reviewed, the rest of the tweaks were primarily ones involving comfort and usability, not sound. They naiiled sound quality early on in the process.
post #223 of 2885
Quote:
Originally Posted by bigshot View Post


Many headpones have a bump up in the upper mids/low high end. They do this to make them sound louder and "in your face" with electronic music. But if you listen to acoustic guitar, jazz, chamber or orchestral music, a colored headphone will never do, because acoustic instruments have a baseline response in reality that we all recognize. Electronic instruments can be adjusted to sound any way the performer wants, so that baseline doesn't exist. But even with electronic instruments, a carefully calibrated response will give you exactly the same sound the performer intended, because studio monitors are always calibrated flat. The Oppos don't have any bump or dip. They have a perfectly balanced response. You could easily add the "in your face" bump to them with a 3 or 4dB boost around 3kHz and they would sound exactly like what you are used to with colored headphones. You could also create the "fat bass" sound of some cans by doing the same around 80Hz. Specific sound signatures are just deviations from flat. If a set of headphones is capable of reproducing a flat response, it can be EQed to sound like any sound signature you want.

Contrary to common thought among audiophiles, headphones don't have soundstage the way speakers have soundstage. Two transducers shooting directly into your ear canal can only create a straight line of sound directly through the middle of the head. It can't create a 360 degree sound field like speakers. All a headphone can do is present the range of frequencies clearly and in a balanced manner, so the small aural cues that indicate distance embedded in the recording- specifically bounce back and reverb- are clear and present. A flat frequency response does this the best because coloration in the wrong frequency range can introduce auditory masking blocking frequencies an octave above. I had this vividly demonstrated to me by an engineer friend. He played some music for me and boosted the mids at 1-2kHz about 4-5dB. Suddenly, the treble around 3kHz became muffled like a pillow had been put over the speaker. I would have never guessed that a boost in one spot would affect a completely different range, but hearing was believing.

If the bounce back thud off the wall behind the snare drums sits around 1kHz, a boost at 500Hz would be disasterous to the distance cues. A typical 3kHz upper mid boost makes the high end of the treble on the cymbals disappear, so the frequencies above that have to be boosted to compensate. That's why when you see response curves of headphones, there are always images of rocky mountains on the right side of the chart. It isn't easy to balance the high end. Most headphones have to settle for kludges to get them to sound OK. The same is true of the low end... Most headphones don't have balanced extension down to 20Hz, so they boost at 80 to fill in the gap.

 

I have a pretty good idea of what a neutral headphone sounds like.  The Oppo is one of them, as is the Paradox.  I was comparing the two to each other, not to all headphones.  No headphone is perfectly flat, yet this phrase gets used every time a new headphone comes out to help add coal to the hype train.  A few headphones off the top of my head that were described as "perfectly balanced" upon launch: LCD2, LCD-X, HE-6, HD800, SR-009, etc.  Look at the graphs of the LCD2 and HD800 and tell me how those could possibly both have perfect frequency response.  Upon listen, the LCD2 is clearly lacking treble and the HD800 has too much.  I don't think the Oppo is as far off, but to suggest the FR is perfect is more of a pipe dream than reality. 

 

I'm not sure what the soundstage tangent is necessary for, but when people refer to 'soundstage' on this forum it's not hard to understand what they mean.  No, they don't throw out  a soundstage at you the way speakers do, but if you've listened to more than 1 or 2 headphones over $100 you'll understand what 'soundstage' means on Head-Fi within a few minutes.  Listen to a HD800, SA5000, K1000, or Stax Lambda and compare the way the sound space is presented to a DT-48, LCD2, SR-80, etc.  There's a very clear difference, and it's not related to frequency response.  Do you think angled drivers are for looks?  Soundstage in headphones is a function of many things: distance of driver to ear, angle of drivers, open or closed, pad design, type of driver, amount of damping on the backwave if planar, etc.

 

Most modern orthos and electrostats actually are flat to 20 Hz (or very close to it).  Dynamics typically roll off as you describe, but a sealed ortho/stat of appropriate driver area is flat.  This has been demonstrated many times.

Quote:
Originally Posted by bigshot View Post


Well there is no audible distortion. That's pretty much a given considering the planar design. Sound stage I just explained. Speed and decay are a part of the recording. PRAT is a part of the music itself. Headphones are a thousand times faster than the typical reverb and decay and a hundred thousand times faster than the rhythms in recorded music. The dynamics on the Oppos is fantastic, again because of the planar design.

I think the problem is that "flat response" isn't understood by most people because they have never heard flat reponse, and wouldn't know it was flat if they did. When you work a bit with equalizers and carefully refine response curves, you find out that "treble being comparable" most definitely *does* mean that the treble sounds the same between different transducers. Frequencies are what we hear. The balance of frequencies is what makes one set of headphones sound different from another. You can take really good headphones capable of a flat response and EQ them into sounding like any other set of headphones. I've done this myself.

Over the past few decades, technological advances have pushed distortion levels, dynamics and noise floors far below the thresholds of human perception. The big wild card is frequency response. The more you know about frequency response, the more you know about high quality sound.

Speed and decay are part of the headphone.  This is why CSDs are so useful.  A perfect headphone would have no decay and would add no extra sound to the input signal.  Physically this is impossible to achieve with moving mass, but as with everything we discuss some headphones do this better than others.  There is more to speed than being able to play a 20Khz tone.  Listen to any Stax headphone and compare it to your traditional dynamic and you'll hear exactly what a difference transient speed makes.  Remember, music is a much more complex signal than sine sweeps so the effect gets multiplied many times over.  This applies to the decay as well...just because one string on a violin may stop vibrating at a certain point doesn't mean it ends instantaneously in the headphone's cups when another string starts to vibrate.  Headphones with cleaner CSDs are usually described as clear for this reason.

 

You can NOT transform one headphone into another simply by matching FRs.  You can get much, much closer than they originally were but there is more to sound reproduction than FR.  I agree that FR is a huge deal (probably the biggest), but it's not the only thing that's important.

post #224 of 2885
The equalizing by ear done by bigshot using his calibrated speaker is why I mentioned these phones should be renamed as BigShot-1 wink.gif.

It must be an exercise of humility for the designer too, having people coming back to you with drastic changes in the voicing. If things are really like bigshot hears it, then the answer about the "target curve" is that it's neither DF or FF, but some setting representative of speakers in properly setup room. By the way, Tyll isn't using DF, he's using a curve referred by the head manufacturer as Independent of Direction, whatever that means, but it's neither DF or FF, my guess is it's more like +/-30 deg with some 5-10deg variations for averaging purposes.
post #225 of 2885
Quote:
Originally Posted by bigshot View Post


In the past I ran some tests on myself to determine my ability to hear different volume levels with test tones and music. With tones, my accuracy was below 1dB throughout the core frequencies. With music, my threshold was about 3dB. My sound engineer friend's ears are definitely better than mine. He equalizes with tone sweeps for a living. I think I'm pretty safe saying we are accurate to +/- 1dB in frequency ranges down to an eighth octave or so, but that is just a guess. It might be a little bit worse or better. Double those thresholds and it's still the same. Close enough for government work.

In the beta group, one of the participants ran a microphone based response that looked about the same as what I found, but then he started talking with the head Oppo engineer about the proper way to measure response and it all started flying over my head. I use tones and ears to tune. It's more straightforward, and ultimately ears are what I listen with. It isn't as accurate, but you can figure out what matters.

 

Again, I reiterate:

Quote:
 This isn't the first time on Head-Fi we've had people state that something was perfectly flat/neutral only to find out months later, after proper measurements were taken, that they were off by a long shot.  No matter how experienced they were.  Our hearing is not stable.  It is dynamic and ever changing.  That's the problem of subjectively measuring something. 

 

No matter how accurate you want to say you are, you just can't place a bound around like that because of the fact that the human hearing simply doesn't function like that.  Actually, none of the human senses are accurate like that.  Not your hearing, not your vision, not your sense of acceleration, temperature, etc.  We simply don't function on bounds and numbers.  We simply don't function like that and can't be bound by numbers.  Asking how accurate your hearing is to the dB is like asking how many megapixels your eyes are.  We simply don't have a bandwidth for that because of the way sound is interpreted by the brain.  It's filtered, non-directly, in a non-uniform, non-flat manner.  Even further, it's further influenced by our feelings and emotions.  It's highly unstable, very dynamic, and not an accurate measuring tool. 

New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Headphones (full-size)
Head-Fi.org › Forums › Equipment Forums › Headphones (full-size) › Oppo PM-1 Planar Magnetic Headphone Impressions Thread