New Posts New Posts RSS Feed - Headphone amps
  FAQ FAQ  Forum Search   Events   Register Register  Login Login

Welcome to the Graham Slee Audio Products Owners Forum

 

Open to all owners plus those contemplating the purchase of a Graham Slee HiFi System Components audio product and wishing to use this forum's loaner program: join here (Rules on posting can be found here)

This website along with trade marks Graham Slee and HiFi System Components are owned by Cadman Enterprises Ltd


Headphone amps

 Post Reply Post Reply Page  <12
Author
dvv View Drop Down
Regular
Regular
Avatar

Joined: 12 Jan 2008
Location: Belgrade, Serbi
Status: Offline
Points: 95
Post Options Post Options   Thanks (0) Thanks(0)   Quote dvv Quote  Post ReplyReply Direct Link To This Post Posted: 15 Jan 2008 at 3:36pm
@John C
The better to find you, John. Because both Graham and I think so, we argue for the fun of it and because we both learn from each other. The day I stop learning is the day I start dying, that's my motto, I love learning from people who know. But, no matter what we say, in the end, only the resulting product will make any difference to speak of.
 
@Leo
Some years back, I tested Graham's Solo for TNT-audio.com site, I think the texts are still there in the log. I liked it, but somehow felt it could have been taken a step further. Graham and I started exchanging messages furiosly, I know we both loved it, and lo and behold, some months after the first sample, Graham sends me a Mrk II version.
 
Item by item, he followed my suggestions, and in terms of the electronics, took them intially and then improved on them. And I must say, I was ever so pleased. Not because of vanity, but because here was a man willing to at least investigate what he had been told in goodwill. Later on, as you can expect, Graham returned the favor with some very solid advice, which I took very seriously and was the better for it.
 
From this small example, and others I have experienced, I came to the conclusion that true audio lives at the outskirts of the audio industry, it rests with small companies still dedicated to good sound first, not sales figures first. I KNOW Graham to be such a person, which is why I am here, even though in a very short while, we could well become competitors.
Back to Top
leo View Drop Down
Moderator Group
Moderator Group
Avatar

Joined: 13 Jan 2008
Location: uk
Status: Offline
Points: 53
Post Options Post Options   Thanks (0) Thanks(0)   Quote leo Quote  Post ReplyReply Direct Link To This Post Posted: 15 Jan 2008 at 4:21pm
Well whatever was done then is now working a treatSmile
Looking forward to reading your posts dvv and nice to have you on here

I'll be building a Kumisa III next to try out so will be interesting to see how this one compares, its all good fun!
Back to Top
dvv View Drop Down
Regular
Regular
Avatar

Joined: 12 Jan 2008
Location: Belgrade, Serbi
Status: Offline
Points: 95
Post Options Post Options   Thanks (0) Thanks(0)   Quote dvv Quote  Post ReplyReply Direct Link To This Post Posted: 16 Jan 2008 at 10:11am
Originally posted by Graham Slee Graham Slee wrote:

... For the 10mW into a high impedance headphone of 300 Ohms (Sennheiser HD600/650/250/etc) an rms voltage of 1.73V is required and the rms current will be just short of 6mA. This can be easily done by small signal transistors or even the ouput of an op-amp (operational amplifier IC). However, a small signal transistor can't handle a dead short that could be possible inserting a phones plug into a headphone jack. Therefore some series resistance is required. And on hearing shouts of "what about the damping factor?", I must ask why? The tiny headphone transducer hasn't much mass at all, and compared to an average 8 inch loudspeaker drive unit, you may as well call it none. At this point I could be accused of over simplifying the argument, but I'm used to that, and I wouldn't want to loose our less technical readers too early in tech-speak. ...
 
Graham, I agree the actual moving mass of the headphone transducer is very, very low, I image around the same as a classic 25 mm (1 inch) titanium dome, say 0.2 grams. If ruled by that, I'd go for your argument. However,  one must also look at what one's Q factor of the headphones looks like under different types of drive.
 
On the lab bench, .i.e. on a 'scope, you can actually see the difference between a damping factor of say minus 5 (VERY typical with standard voltage dividers in integrated amps) and damping factor of sau 30:1. That Q factor moves from the hated neighbour grandma to the coveted Halee Berry.
 
But much more than this, my ears tell me that with a decent damping factor I get better definition, I get crisp, well defined and not artificially rounded-off bass lines, and so forth. In a nutshell, I get better definition. I am not saying it's a revelation, but it can be heard quite easily on say Sennheiser HD595 cans, or Grado SR-60 cans for that matter.
 
The point being that indeed, some series resistance is well advised as a general precaution, but if you use 60W output devices as I do, that series resistance can be rather low, 1 Ohm or even less. You could even use something like 2 or 3 Ohms, because even with an uncommonly low can impedance of say 22 Ohms (some Panasonic cans I once tested), with a 3 Ohm series resitor you still have a damping factor of no less than 7:1. With a 3 Ohm series resistance, Motorola/ON Semiconductor's MJE 15030/15031, fed by +/- 15V (typical for quality op amps), will be limited to around 5 Amps - and they are rated at 6 Amps (8 Amps transient). I assume even a slow blow fuse will give up the ghost by then.

Quote When driving lower impedance headphones like the 32 Ohm Grado's and 25 Ohm Denon's most headphone amps that opt for a "power amp" approach, will use a gain switch to reduce the output voltage so as to place the volume control in the same respective place. However, by inserting the right value of output resistor, an approximately similar volume can be obtained with headphone impedances from 8 to 2,000 Ohms, and gain switching is not required. And this means small signal transistors can be used instead of power devices. This very approach was adopted by the IEC (International Electro-technical Committee of Geneva) under CEI/IEC 61938.
 
And for the worse, I feel. Those people are great when it comes to general electrical safety, but they suck when it comes to quality audio.
 
That said, I must admit I adhere to their view that for typical line outputs, a series 120 Ohm resistor should be used. They got that one right as a fluke.

Quote The Solo MC was designed to that standard and in common with many headphone amplifiers, the Solo uses an op-amp with an output buffer, which in the Solo's case is a class AB output stage using small signal transistors. And going on its reputation and popularity, it works!

It does, and then some - no argument there. But that's not to say one can't do even better.

Quote However, there are those who will deny what their own ears tell them, and insist IEC61938 is wrong. This could be due to personal bias, or maybe brainwashing by the doctrines of other manufacturers. However, it could be that some headphones are designed for the most popular headphone amps on the market - and unfortunately, the Solo, for some reason, doesn't rank among those. One very popular headphone amplifier actually uses 33 Ohm output resistors. Therefore will some lower impedance headphones sound better fed that way? I haven't found that to be the case with any of the Grado's, but perhaps we need an alternative design to cater for such an eventuality?
 
Leave the niche makret, Graham, you/we are not strong enough financially to get into that. Much development work, by default very low sales - hateful terms in audio, but they are the harsh reality we live in, like it or not.
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Retired

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 16298
Post Options Post Options   Thanks (0) Thanks(0)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 17 Jan 2008 at 10:31am
There are pros and cons with any circuit designed to work with a transducer.
 
It could be argued that a power-amp stage is best to drive headphones, especially low impedance headphones, but most of these arguments forget, or don't consider the actual output voltage to the headphones. These are incredibly low compared to loudspeakers. 1mW into 30 Ohms equates to less than 200mV. To hear the subtleties of a classical piece which is commonly accepted as requiring the ability to resolve at -60dB, the signal voltage is less than 0.0002V (200 micro volts) using a headphone with a good SPL of 96dB. It is not very much more than this with a less efficient headphone. Are power transistors designed to have good loudspeaker driving characteristics, going to acurately resolve voltages below 1mV?
 
Your answer may be yes. But in what configuration? Class AB? No. Because that takes it right into the depths of crossover distortion. Even with the most carefully optimised class AB output stage with maximal loop gain (lots of negative feedback) the signal is swamped by this distortion. The loud bits may be crisp and carry the weight, but the subtleties will be grossly distorted.
 
And this is exactly why most power-amp-like headphone amp output stages resort to class A, or class Ab (class B with a gross amount of quiescent current).
 
I am not saying this is wrong in all cases, but output transistors must be selected carefully. Often what looks to be a good transition frequency (eg 60MHz) can translate into a bandwidth limited implementation. When in class A conduction, the collector current is going to be very high indeed to reduce crossover effects to what's desired. Taking a look at the datasheet of the transistor may not reveal the true nature of the device - I have found many of the important characteristics are left out. How can a good designer design without being aware of these characteristics?
 
Transition frequency, as I've said elsewhere, is the frequency the transistor can reach at unity current gain - eg, with no current gain - the equivalent to a piece of wire bypassing the device. It has no worth at all until it is used in conjunction with the Hfe of the device which states the devices current gain.
 
The transistor's high frequency "cut-off" is determined by Ft/Hfe. In class A, which means lots of collector current; to maintain a good bandwidth transfer function the full Hfe cannot be taken advantage of otherwise its bandwidth is effectively reduced. This is because of the devices input capacitance. It may look like it's easy to drive at low frequencies, but the highs have to negotiate a much heavier load.
 
This requires a pre-driver stage capable of driving the capacitive load, and that itself adds complications. Either that or copious ammounts of negative feedback requiring an extreme amount of open-loop gain pushing op-amp or discrete VAS (voltage amplifier stage) capabilities to their limits. Or both!
 
It should also be understood that negative feedback is not the cure-all, but has become a crutch many designers rely on.
 
Ultimately we end up with a very complex power hungry amplifier to deliver 1mW to a headphone on the premise that "it will sound better" which, after all is a subjective assumption based on what other people would have us believe.
 
It is this "what other people would have us believe" that constrains progress in audio, branding anyone who thinks differently as a deceiver. Unfortunately all that does is confuse people's minds, and this community isn't here for that.
 
Different properly argued implementations (and I mean properly argued in such a way that deception can be detected) must be able to exist otherwise there can be no progress. I have faced this very same argument time after time, but there is a saying: "the proof of the pudding is in the eating". The Solo's success to date suggests that it is a good alternative approach, and that's exactly what it is - a good alternative approach that works.
 
And my reason for doing it so?
 
Exactly for the reasons set out in this reply. I have researched the weighty arguments for a power-amp based approach. It would seem that people will always believe the only thing that works in hi-fi is the sledgehammer to crack a nut approach. It may work, and in a few cases work well, but so can the opposite. Everything will always have a downside to be unearthed by those having the time and ability to do so. I identified the downside of the power-amp method. I have highlighted its disadvantages here, as dvv has highlighted the disadvantages of series resistance.


Edited by Graham Slee - 17 Jan 2008 at 10:48am
That none should be able to buy or sell without a smartphone and the knowledge in how to use apps
Back to Top
dvv View Drop Down
Regular
Regular
Avatar

Joined: 12 Jan 2008
Location: Belgrade, Serbi
Status: Offline
Points: 95
Post Options Post Options   Thanks (0) Thanks(0)   Quote dvv Quote  Post ReplyReply Direct Link To This Post Posted: 18 Jan 2008 at 8:24am
Rather than dissect the text piece by piece, I'll try to give a summery answer.
 
Yes, to have the power devices work properly, I do have to (if I want truly good sound, that is) use a pair of drivers - we both know that to be effectively mandatory. However, while it is a further complication, it also has its benefits, such as increased bandwidth, lower output impedance, etc.
 
To make sure all is well, one does indeed have to select the drivers and the output devices MOST carefully, not only on their own, but just as much in tandem. And I agree that using 60W power transistors may well seem like an overkill.
 
But consider this. To use the KISS (Keep It Simple Stupid) principle without much ado, one is most likely to opt for an enriched bias design. The only true question here, as far as I can see, is what you take as your worst case load, much the same as if one was designing a power amp (and this IS a power amp. albeit with a low power output). For example, one could say that quality cans start from 30 Ohms upwards, others will say there are units out there as low as 22 Ohms (some Panasonic model, I forget the designation), and others yet will insist that for some obscure reasone we include switchboard 8 Ohms cans as well.
 
The other side of that is headphone effciciency, which does vary rather wildly, although of course, it's still way below speaker variance (and general demands). Combining the two can change the requirements rather significantly. On one extreme, we have very benign, high impedance, high efficieny cans such as for example Beyer cans, ond on the other extreme we have some obscure models, from even more obscure sources, which are most demanding (in headphone terms) and very low efficiency designs. Good designers will always want to work with as much headroom as they can reasonably have, which adds still more, etc.
 
The argument regarding Negative Feedback (NFB) is, unfortunately, all too true, however it is not mandatory. One can design a discrete amp with programmed and strictly controlled open loop gain, which can go a long way in reducing distortion to very low levels, allowing the NFB to be what I feel it should always have been, icing on the cake, not a cure-all, but simply ironing out a clean and well tailored shirt, so to speak. I find that 14-20 dB (5:1 ... 10:1) NFB is all one really needs if one has done one's homework well.
 
Electronics is exactly like audio - there is no one, sure-fire way to get there, there are always oh-so-many alternatives. That's probably why it's so much fun.
 
Lastly, over the decades I have been up to my neck in audio, I have observed many times that circuits using higher than usual supply voltages somehow tend to sound at least that little bit better than circuits with lower supply voltages. Assuming equally competent design work, it's obvious that circuits with higher supply voltages will have more headroom, and it seems to make a difference even when one rationally knows that what is really needed is just a say +/2V swing, way below the circuit's capabilities. I frankly admit I have no ready explanation for this effect.
Back to Top
 Post Reply Post Reply Page  <12
  Share Topic   

Forum Jump Forum Permissions View Drop Down

Forum Software by Web Wiz Forums® version 12.01
Copyright ©2001-2018 Web Wiz Ltd.

This page was generated in 0.109 seconds.