I added a 10r resistor to the output of a headphone buffer I built up.
Mostly to avoid this risk of shorting during insertion.
Use 32*NE5532s. Use a different connector?
Come on guys, there are amplifiers out there intended to drive speakers that will withstand SC. It's something many consumers expect.
If other people want to build headamps with high output impedances I can only deprecate it, I can't legislate against it. But none of the points raised so far lead me to think any different.
Commercial amps have SC protection, also most chips have SC protection, just like the NE5532.
But many diy amps don't have, as ppl like to make them as simple as possible. And they will pop in the second you sc the output.
Ppl. don't want to build high output imp amps. That is the question if it is good or bad. I just mentioned that sometimes it is only there for protection.
Few years ago there were trends to say that 1/8-1/5 of the headphone impedance should be used on the output.
Interesting link, dramatic charts. But I am not sure if he was measuring distortion or "impedance distortion". Funny he didn't provide acoustic measurements.
The only value of low output impedance, I think, is when the sound is reflected in the back EMF of the voice coil. Even then, we are talking about damping, not feedback.
With woofer, the effect on bass resonance is dramatic (not that there aren't folks who prefer the tubby sound to the damped sound). Not at all sure (yet) about consequences for headphones or any specific model. Are the little drivers in headphones just mini speakers?
Ben
I did found - more or less - the same: the higher the amplifier output impedance, the higher the THD you measure *** on real loads ***; I can't measure acoustic THD, but it would be very interesting to find a correlation (or no correlation at all, maybe) between, say, electrical and acoustical THD.
What do you mean by impedance distortion? Non-linear I-V relationship?
L.
I just made up the term, hoping others would recognize it was a guess in the dark.
The measurement was the same as the set-up for impedance measurement. The weird stuff found at the spot where the distortion measurement was taken (which is fair enough to call THD) is electrical jitter at that spot.
If I recall, buried in the text is the remark that the acoustic distortion is a whole lot less than what I called the "impedance distortion" and that amp output impedance was a lot less influential in acoustic deterioration than electrical deterioration.
I hope somebody who isn't just guessing in the dark will post wiser thoughts than mine.
It would be really nice to come to some trustworthy conclusions about output impedance. For sure, the majority of jacks marked "headphone" on power amps and boutique amps - not to mention 10,000 Cmoy-type amps - are preceded by a non-trivial sized resistor. Is that bad for quality?
Ben
Low impedance dynamics (not orthodynamic) headphones that I own like Foxtex (TH900) or Audio Technica full sized (A900, W1000X) are quite sensitive to output impedance.
There may be distortion in frequency reproduction, but that's not the main concern for me, it's the bass impact/control that suffers the most when damping factor is not good enough. 2 ohm of output impedance is the upper limit that I accept. (Well, my Yulong A18 is 2.2ohm - and I could live with it
)
My Beyerdynamic DT770 32ohm seems less sensitive - or it's already too "boomy" to notice the improvement.
I suggest to try Violectric HPA-V100/V200 amps for their bass control. They may not be everyone taste for other reasons (hot treble, frontal soundstage), but it has become my benchmark for the bass impact/control of low impedance dynamic headphones. Well, The Wire is quite good in that region too. There are exception thought, O2 is subpar in bass reproduction despite the low output impedance, imho
.
From Violectric HPA-V200 technical data:
output impedance of <0.06 ohm / damping factor 400
I totally agree that most headphones amps design should really be looking for getting as close as 0 ohm as possible. There are few cans that were made for that damn st*p*d 110ohms IEC recommandation... but they're much less common than the one that that benefit from low impedance, specially with the ever growing IEM market.
My 2 cents as a headphone junky.
There may be distortion in frequency reproduction, but that's not the main concern for me, it's the bass impact/control that suffers the most when damping factor is not good enough. 2 ohm of output impedance is the upper limit that I accept. (Well, my Yulong A18 is 2.2ohm - and I could live with it
)My Beyerdynamic DT770 32ohm seems less sensitive - or it's already too "boomy" to notice the improvement.
I suggest to try Violectric HPA-V100/V200 amps for their bass control. They may not be everyone taste for other reasons (hot treble, frontal soundstage), but it has become my benchmark for the bass impact/control of low impedance dynamic headphones. Well, The Wire is quite good in that region too
. There are exception thought, O2 is subpar in bass reproduction despite the low output impedance, imho
From Violectric HPA-V200 technical data:
output impedance of <0.06 ohm / damping factor 400
I totally agree that most headphones amps design should really be looking for getting as close as 0 ohm as possible. There are few cans that were made for that damn st*p*d 110ohms IEC recommandation... but they're much less common than the one that that benefit from low impedance, specially with the ever growing IEM market.
My 2 cents as a headphone junky.
As a newbie in electronics, i designed my own headphone amp. It is designed for flat response, mono input, for guitar playing, plugging it into my AKG K550(wich is "flat", too).
At the time i read something about that 120 Ohm standard, and I decided to compare 3 impedances : 147, 122, and 100 Ohm.
The best sound i was able to obtain from this circuit and this headphone was with 122 Ohms.
Lower impedance gave me more basses and the sound quality was degraded in the lower end.
Higher impedance conducted to a loss of 'punch' and of basses.
O2 employs parallel opamps in buffer stage. The O/P resistors (1R) prevents the opamps fighting each other for current. The JRC4556 can withstand momentary or continuous sc (of course within its datasheet specified range).
FWIW, the effect of non-zero output impedance varies considerably. Beyer DT235s (32 ohms) are hardly bothered, while Sennheiser HD239s with the same nominal but far more variable impedance very quickly acquire a considerable bass bump. When I had some HD598s (also with highly variable impedance, from under 60 to well over 200 ohms), I felt that 33 ohms already made them a bit too bassy, while a ~0-ohm output was just about right. And let's not even get started with multi-driver BA IEMs, some of which appreciate well under 1 ohms of Zout.
By contrast, a number of people feel that some higher-end models like Sennheiser HD800 or Beyer T1 are too thin without some extra output impedance; Sennheiser's own HDVD800 employs a fixed 43 ohms for this reason (which is a bit disappointing - some customization options would have been nice, especially considering what the thing costs).
In some cases we are only talking a dB or so of boost. Do keep in mind, however, that it usually is quite broad and as such quite easily audible in a direct comparison.
By contrast, a number of people feel that some higher-end models like Sennheiser HD800 or Beyer T1 are too thin without some extra output impedance; Sennheiser's own HDVD800 employs a fixed 43 ohms for this reason (which is a bit disappointing - some customization options would have been nice, especially considering what the thing costs).
In some cases we are only talking a dB or so of boost. Do keep in mind, however, that it usually is quite broad and as such quite easily audible in a direct comparison.
Quite strange because high impedance drive on a loudspeaker may impair its frequency response but generally, it decreases non linear distortions (THD).
It has often been proved and discussed here, search for "current drive".
Telephone engineers used to have terms and circuit layouts unique to their area and mysterious to people otherwise educated in electric matters (generally thought to be intentional).
I'm finding the something similar among headphone specialists. In particular, using "impedance" of an amp to mean something like "how loud your headphones will sound" and conversely, using headphone rated impedance as if it related to how loud the headphones will sound for a given input power.
For sure, there is a necessary relationship among output impedance, loudness, voice coil windings, and boominess (or loudness in that bass sense).
Ben
It's more complicated than that, actually. You can model a dynamic driver as a transformer-coupled resonator. So how much impedance response varies on the primary side will depend on mechanical damping in driver plus enclosure, coupling strength, and resistive losses. You often find closed headphones with rather flat impedance response, while open ones with little inherent damping in the drivers may show significant peaks. (You can tell a closed model has strong coupling when all kinds of small resonances show up on the impedance graph.)
This model also tells you that it is possible to make versions of one driver that basically differ only in impedance (but have the same power sensitivity), just by varying primary, err, voice coil turns and wire diameter while keeping voice coil mass constant. Beyerdynamic make extensive use of this nowadays, with e.g. the DT880 coming in 32, 250 and 600 ohm versions (plus the Custom One Pro at 16 ohms). They're not totally identical sounding but fairly close. As you might expect, a given amount of amplifier output impedance would have a stronger effect on the 32 ohm model when compared to the 600 ohm one.
I've never seen any Thiele-Small parameters (TSPs) posted for headphone drivers, but basically those and the underlying models should apply just the same as for speaker drivers.
The amount of coupling indirectly figures into sensitivity (along with such things as driver-ear distance etc.). You may find two different specs here, dB SPL re: 1 Vrms (sensitivity) or dB SPL re: 1 mW (efficiency), usually at 1 kHz. The dB/mW spec essentially derives from speaker driver specs but is of rather limited practical use (well, a 120 ohm output approximates constant power for 16..600 ohms, so it may be useful there). When you've got essentially-0-ohm outputs, the dB/V spec is more useful, as you can directly tell how loud things will go for a given output amplitude. My little headphone level calculator spreadsheet makes use of this.
Thanks for a very good explanation. Now I understand your earlier post.
You are saying that the damping situation for loudspeaker drivers is, in theory, the same for headphone drivers. In headphones, you are saying electrical damping does, in theory, influence frequency response apropos resonances (and as also influenced by coupling, etc).
I believe you are also saying that if we find, in practice, no large resonance bumps in the impedance curve of a set of phones, they are not likely to be influenced much about output impedance.
In your earlier post, you also offered examples of headphones that are influenced and not.
For all the rock 'n roll verbiage and heartfelt declarations about breaking-in timing (which some authorities claim is nonsense) that we find in headphone reviews, you'd think somebody would include remarks about sensitivity to output impedance.
Ben
Exactly. The whole setup is basically treated as a complex voltage divider of Rout and Zload, which means voltage at the load becomes
Vload(f) = Zload(f) / (Zload(f) + Rout) * Vout
if our output is an ideal voltage source Vout with a series resistance of Rout added. So if you have an impedance graph taken with the headphones mounted on something broadly resembling a head, you can calculate what sort of frequency response variation to expect. There's a few sources for such graphs, like IF datasheets and GE reviews.
(BTW, taking these graphs usually involves the same kind of computation, just backwards, i.e. solved for Zload(f). If Zout(f) is known, you can also use this for determining a pretty decent estimate of Rout with a splitter cable and some means of comparing frequency response deviation loaded vs. unloaded, like via RMAA.)
Alas, this seems to be going over most audiophiles' heads. Finding out what sort of output impedance a certain device has can be hard enough; not uncommonly a peek into the service manual or a measurement is needed. Ideally you'd be able to solder up a little perfboard adapter with an assortment of resistor values selected by jumpers that you can add to a known near-0-Ohm output. It doesn't even have to be as fancy as that though, just as long as you've got outputs with known near-zero, medium (33-100 ohm) and high (220-470 ohm) output impedance. With some luck, a portable player, a soundcard or two and an integrated amplifier or receiver will provide such a selection.
sgrossklass -
Clear. Thanks. I would only differ in saying you don't even need a perf board - just a handful of alligator clip leads, 3 resistors, some plugs you can take apart, and free REW software and anybody can get a good picture of the output impedance of their amp (only one or two measurements are needed) and the impedance curve of their headphones. Easy.
But there is something in your earlier post that is a difference between speakers and phones that needs serious discussion: I can't imagine anybody advocating reducing the electrical damping factor for a loudspeaker*. But with phones, it is routine to say the sound (meaning the bass) is adjustable to taste through rasing or lowering output impedance.
That may be the reality of headphone state of the art. But it doesn't seem wise to traffic in resonances in phones any more than in loudspeakers.
Ben
*T-S modeling software does exactly that... and it seems to me as ill-conceived to optimize box damping as it would be to "optimize" electrical damping.
Clear. Thanks. I would only differ in saying you don't even need a perf board - just a handful of alligator clip leads, 3 resistors, some plugs you can take apart, and free REW software and anybody can get a good picture of the output impedance of their amp (only one or two measurements are needed) and the impedance curve of their headphones. Easy.
But there is something in your earlier post that is a difference between speakers and phones that needs serious discussion: I can't imagine anybody advocating reducing the electrical damping factor for a loudspeaker*. But with phones, it is routine to say the sound (meaning the bass) is adjustable to taste through rasing or lowering output impedance.
That may be the reality of headphone state of the art. But it doesn't seem wise to traffic in resonances in phones any more than in loudspeakers.
Ben
*T-S modeling software does exactly that... and it seems to me as ill-conceived to optimize box damping as it would be to "optimize" electrical damping.
The only difference between an Etymotic ER-4S and ER-4P is an inline resistor. Etymotic even sells it as an aftermarket conversion accessory. Most consider the audible difference significant. New headphone amps appear to strive for low output Z, for example some devices from Fiio are rated 0.25 ohm.
I guess headphones (esp around-ear and in-ear) have lots of damping from the confined air space and perhaps other design issues constraining resonance Q and providing degenerative feedback. That's unlike old-fashioned cone loudpspeakers that get little air damping, horns excepted*.
Therefore, it possibly is reasonable to "adjust" electrical damping to headphones in order to achieve a favourable sound, unlike with loudspeaker drivers.
That design approach doesn't make good sense to me as a student of woofering, but just one of the audio world's compromises.
Historical footnote: decades ago, there were discussions of the optimum electric damping for loudspeakers. Maybe in the future when loudspeakers can handle sound without major compromises, it will be a non-issue.
Ben
*which is why some of us are keen to develop motional feedback systems
Therefore, it possibly is reasonable to "adjust" electrical damping to headphones in order to achieve a favourable sound, unlike with loudspeaker drivers.
That design approach doesn't make good sense to me as a student of woofering, but just one of the audio world's compromises.
Historical footnote: decades ago, there were discussions of the optimum electric damping for loudspeakers. Maybe in the future when loudspeakers can handle sound without major compromises, it will be a non-issue.
Ben
*which is why some of us are keen to develop motional feedback systems
Even in the loudspeaker world there are people who like amplifiers with low DF - think simple tube amps.
It's just a lot more troublesome because speakers tend to have crossovers for 2 or 3 ways, and it's them that make their impedance response quite a bumpy ride (hence the effect becomes a lot less predictable). The closest equivalent to conventional speakers in the headphone world would be higher-end multi-way IEMs, some of which are known to be extremely fussy (a UE triple.fi 10 Pro dips down to 6.5 ohms and goes up to about 60 ohms, for example). For a single wideband driver, a DF of about 10 generally is ample.
When it comes to driving speakers, standardizing on virtually zero output impedance was sensible - it minimizes losses to boot. Imagine your 100wpc amplifier would be dissipating half its output power internally, what a waste. The issue of power delivery has never been quite as pressing in headphones, as they tend to go plenty loud on a mere milliwatt.
Minor quibble: not sure even tube enthusiasts embrace high output impedance because even triode tube amps have lots of damping. Not sure speaker performance would differ much between a DF of 4 and 1,000,000 (esp when you consider other resistances in the speaker circuit).
Ben
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.