|
Post by gns on Jun 11, 2007 9:25:56 GMT
I'm often asked if my Solo amp will drive a particular headphone or another. The truth is, having not tried every headphone on the market, I can only give an answer on those cans that I've actually tried.
Wouldn't it be great if there was an IEC standard that told you?
Well there is, but will it tell you?
IEC 61938 was written many years ago (in 1996 to be exact) to deal with matching between audio equipment. It is not the law, but it gives some guidance.
Section 10 is a very small section that deals with headphones, and in table 8, it gives "general purpose" matching values between headphone amplifiers and headphones.
It recognises that headphones can have impedances between 8 and 2000 Ohms, and to be able to provide a good match volume wise, it suggests a driving impedance of 120 Ohms from the amplifier, which also should have an output, before that impedance, of 5V rms. This output being just before clipping.
Analysing this you find that with the 2000 Ohms extreme, this load will get virtually all the output voltage as the attenuation of the 5V output gives 4.7 volts.
With an 8 Ohm load the output voltage will be just 0.3125V.
So how come both the 2000 Ohm can and the 8 Ohm can sound like they're at the same volume?
Here's the math...
2000 Ohms: output 4.7 volts
P out = V2/ load
= 4.72/2000 = 11mW
8 Oms: output 0.3125 volts
P out = 0.31252/8 = 12mW
Virtually the same!
Now let's try that for 32 Ohms (Grado's/Sennheiser Street Wear...)
P out = 12/32 = 31mW
...and now 300 Ohms (HD600/650/250)
P out = 3.572/300 = 42mW
...and at 100 Ohms (Some AKGs)
P out = 2.272/100 = 51mW
From the calculations you find that the extreme impedances work, but are not that effective, but most reader's beloved cans fall into the 30-300 Ohm impedance range, so from a "general purpose" angle, this form of constant current drive does work.
As Mike commented the other day "Some headphones require current and some voltage". IEC61938 demonstrates that high impedance cans are voltage driven and low impedance cans are current driven. But will that be so in practice?
I guess that all depends on whether the headphone manufacturer takes a blind bit of notice of IEC standards?
I know if I was a headphone manufacturer, I would, because it gives me the biggest possibility of market share. Most integrated amplifiers with a headphone jack provide a similar driving impedance derived from the speaker outputs.
So in answer to the original question, the Solo with its self impedance-adjusting output was designed to IEC61938 section 10, and should drive all headphones to very similar levels. The difference in the 30-300 Ohm range is only 2dB!
What surprises me is that most dedicated headphone amps don't work like this and simply give a voltage output, then have to use gain switches to match different impedance ranges.
As for how a particular headphone sounds, that can't be determined by the math, and that's why this forum and others like it exist, to share the listening experience.
Share on!
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jun 11, 2007 10:00:22 GMT
Graham In SiliconChip Studio Series Headphone Amplifier, they also mention IEC 61938 specifying that headphones should be driven by a 120R source. If you are interested in a little light reading, email me and I will forward the article. SandyK alexkethel@optusnet.com.au
|
|
FritzS
Been here a while!
Sound of Blue Danube
Sound of Blue Danube
Posts: 1,364
|
Post by FritzS on Jun 11, 2007 15:24:55 GMT
Dear Graham, .... self impedance-adjusting .... what's the influence of some impedance jumps at different frequencies? Some frequency/impedance curves look likes as the skyline of the alps If some headphones like "zero" Ohm of the amplifier-impedance - can you drive them with good results too or in other words what output-impedance have the Solo? In my current WNA MKII headphone amp I use Holco H2 15 Ohm resistors (mainly for short circuit protection) and BlackGate NX 1000 uF 25V caps (to protect my headphones from DC voltage) in series to the headphone.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jun 11, 2007 23:06:18 GMT
FritzS That first statement seems very innocent, but is right on the money. They also delve into that with a graph in the previously mentioned article, if you are interested.
SandyK
|
|
|
Post by gns on Jun 12, 2007 1:54:08 GMT
Think about it... Sennheiser have 400 outlets in the UK (or so the rep said). They are not exclusively selling to the head-fi or rock-grotto fanatic like us. They are selling to people with amplifier headphone jacks, CD headphone jacks, booth headphone jacks. What do Sennheiser base the foundations of their designs on? Do they buy every piece of equipment in existence just to make sure they're OK? I doubt it. Now let's broaden it to ALL headphone manufacturers. I'd guess a good number will have seen IEC61938? I'd also guess the impedance frequency curves are meant to be like that Why? If they weren't I don't think they'd sound all that good. I reckon the sound would be mostly inside your head. If you take a look at a constant power 90 degree phase filter, you'll see it does a similar thing to the phase - in and out instead of up and down. This type of filter is used to create spatial effects, and was also used by some mono BBC radio stations to make stereo records sound more natural (and not by many commercial AM stations on grounds of cost or simply not knowing about them). So if you drive these headphones from a zero impedance you can bet they'll sound different, but that may not be what the designer intended. I'll let you into a little secret - the Solo output impedance is 120 Ohms.
|
|
|
Post by gns on Jun 12, 2007 2:29:59 GMT
SandyK,
Just seen the argument for low impedance drive in the article you sent me.
UTTER CRAP! (IMHO)
Who is this Peter Smith?
Put 47 Ohm resistor in series, increase output to deliver same power - the entire amp is operating under different characteristics, providing a higher output, being fed with a higher input (different volume control setting?) - distortion is bound to be different!
Nothing to do with the Cans in my opinion.
If you think headphones have odd frequency plots, try phono cartridges. Adjust the circuit to make them flat and they sound SHITE!
(I don't mean the RIAA bit)
There's more to transducers than the simple argument presented by this article. If it were only that simple I'd be a billionaire right now.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jun 12, 2007 2:57:35 GMT
Graham Peter Smith is someone who doesn't like criticism. For example in the main preamplifier he used OPA2134 which have very low DC offset. He used a couple of unnecessary coupling capacitors. In an email he conceded that in hindsight he should have left them out. Did he post any such advice in the magazine for the readers? No. He also had long relay PCB traces with minimal separation between L & R. BTW, I only use 32ohm Sony headphones. For what it is worth, a friend left out the Zobel network, fitted BC549/BC559 O/P devices, and says it makes a quite reasonable preamplifier. Just as well your design doesn't have to do ridiculous things like switching diodes in and out every time you wanted to use different impedance headphones! Regarding phono cartridges, in the past I used standard RIAA, and later corrected minor response deviations at the low end with Sound Forge 9. SandyK
|
|
|
Post by gns on Jun 12, 2007 17:01:47 GMT
Well, I must admit sometimes not taking criticism all that well. Mike could tell you one or two things about that (but Mike - better not, old pal...) A zobel network is used to load the output of an amplifier at high frequencies in a bid to stop it going unstable (oscillating). It is usually only needed when driving "hefty" loads like loudspeakers because the speaker "system" is like an open circuit at such frequencies. The oscillation usually doesn't destroy the amp but I managed to do "wi-fi" before wi-fi was ever thought of, by leaving the zobel out... ;D In my opinion, headphone amps are just preamps with line-driver outputs, and as such, should not need the zobel network. But if the buffer stage can't keep up with the op-amp driving it, then by "miller effect", the thing will go unstable. In the design you sent me you'll see a small resistor in series with the op-amp output. It acts as a RC filter with the output stage capacitance, rolling off the HF. The op-amps function is to turn off the output transistors, while the current sources (Q1 and 3) turn them on. The BD139/140 while being quite fast, need quite a lot of drive because their hfe is poor. The net result is that the output phase is going to be too far out at high frequencies, requiring further HF roll-off, and that is provided by the massive 1.2nF cap across the NFB resistor R1, giving a -3dB bandwidth of 133kHz. Now that's OK, but it's in the negative feedback network, which makes fast edges clip the voltage rails, and when that happens you need a zobel network to try and filter it out. I include the page containing the circuit here...
|
|
|
Post by PinkFloyd on Jun 12, 2007 17:57:47 GMT
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jun 12, 2007 21:52:43 GMT
Graham Thank you for the explanation.It is very much appreciated.The main advantage of being a forum member is that you are able to learn from other people who may have a much greater depth of knowledge in a particular field.You will also have noticed that in the modified version, the chosen output devices have an FT of 200MHZ, combined with a minimum HFE of 100 at 100mA. The main reason I forwarded the article was to elicit a response from you about their use of the Zobel instead of the standard's recommendation. BTW, do you agree with the "Blameless Amplifier" philosophy as presented by Douglas Self? I see that it has come in for quite a few negative comments in another forum recently. Thank you again. Alex
|
|
|
Post by gns on Jun 12, 2007 23:00:50 GMT
www.dself.dsl.pipex.com/ampins/dipa/dipa.htm#1If you're asking if I agree with this philosophy, the answer is yes! It is the complete opposite of the adverse comments on Doug Self that I've heard in rumour. Circuit technology like this is what makes music, and exists in the op-amps I select in one way or another, and that's why I'm a bit stubborn in trying some suggested alternatives, because they don't feature these musical techniques. This approach is refreshingly sane and the total opposite of Mr Smith's. However, having contributed an article to Everyday Electronics many years ago, you wouldn't "give-away" your life's work for the money project magazines pay, so you're not going to get anything really fantastic from a magazine. But the above website is ROCKET SCIENCE, and I will have to PDF it in case it gets deleted.
|
|
|
Post by PinkFloyd on Jun 12, 2007 23:22:04 GMT
Not likely to happen, it's been "online" for years "THREE NON-EXISTENT DISTORTIONS. Audio engineering suffers from misinformation, disinformation, and downright lying more than most fields of endeavour" Classic!!!!
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jun 12, 2007 23:27:35 GMT
Graham Do you agree that although the current mirror may produce identical currents in both legs, that TR2 has very slightly less collector current available to it, due to the bias being supplied to the VAS stage? This will result in a difference in potential between the collectors of the input pair. Thus our perfectly matched differential pair are no longer perfectly balanced. If you can find time to scratch yourself, you may be interested in the remarks by Carlos at the bottom of the page in the attached linkIn particular the comments about the perceived sound quality of the "secret modification" SandyK www.diyaudio.com/forums/showthread.php?s=&threadid=103157&perpage=10&pagenumber=5
|
|
|
Post by gns on Jun 12, 2007 23:59:01 GMT
With collector resistors in the differential stage the collector load will be imbalanced, but here the current mirror prevents that. However, the base current for the VOS flows in one transistor and not the other, so there is an imbalance in transistor current and because that effects Re (intrinsic emitter resistance) the gain should increase in that leg, but then the base impedance of the VOS effectively lowers Rc which should take it back to where we started. So I guess it could be argued that it is still "blameless", but I don't feel like doing lots of the math right now. The VOS (voltage amp) consists a high beta darlington, so the base current is very small anyway.
The only way of obtaining true symmetry here is to follow a typical op-amp technique and have two voltage amplifiers, each being driven by each input differential pair's collectors, and then make it symmetrical right to the output.
|
|
|
Post by gns on Jun 13, 2007 0:04:51 GMT
PS. That's why op-amps make so much sense over discrete implementations in preamps where signal paths need to be short, and the additional complexity required for perfect symmetry could easily lead to compromise in signal paths. What is going to be the worst evil?
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jun 13, 2007 0:10:08 GMT
Graham What if you inserted diodes wth the appropiate voltage drop in series with TR3 collector OR replaced the s/c across b & c of the current mirror with a 1N4148, as well as varying by a small amount the emitter resistance of the VAS so that the difference between the differential pair collectors was LESS than 5mV with this very well matched pair ? Did you read Carlos's comments in the attached link ?
|
|
|
Post by gns on Jun 13, 2007 0:39:05 GMT
It's about current matching, not voltage.
Admittedly, the current mirror looks odd legged because of the shorted BC, but that's how they work - taking the output from the one that isn't. Adding a diode in the collector makes no difference to the current. Placing one in the mirror's BC connection would introduce an unwanted thermal effect, destabilising the current mirror vs temperature.
Don't understand what this Carlos is on about as I cannot see the diagram, because I am not a member of that forum.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jun 13, 2007 1:04:42 GMT
Graham This was before current mirrors were in general use - Way back in 1987 I asked the editor of Electronics Australia" why in a preamplifier with a very well matched differential pair, after a period of time, did the soundstage markedly improve and overall sound was "sweeter"The improvement, however slowly vanished. I added that each time I became aware of the phenomenon, I took voltage measurements, and each time , the difference between the collectors was <5mV." He did not reply. In late 1989 Silicon Chip published a 4 page article on mods I had made to their Studio 200 Amplifier. My observations wer published, but nobody commented on what I was trying to say, only criticised the way that I enforced this balance. In the meantime there have been many amplifies of varying topology built by a friend and myself, usuing the basic method I described previously. It has also been used in a few Class A preamplifiers with similar results. I recently posted a thread in this forum regarding a comparison between several amplifiers and preamplifiers , including AKSA55 and a AU$6,000 Krell preamplifier. Carlos was privately given this information and has done many simulations of the D.S. front end implementation etc. This recently resulted in him trying the various methods, and he posted his subjective findings. Yes, I know that engineers usually hate the word "subjective" Alex
|
|
|
Post by gns on Jun 13, 2007 2:25:39 GMT
By doing the same thing to both sides of the equation you should get perfect balance (x-y=0) TR12 draws (0.6V/1k)/TR12's hfe mA from TR2. Therefore bleed off the same current from TR3. You could use a resistor alone for the two VBE drops, or even better a resistor dealing with one VBE drop in series with a diode in thermal contact with TR12, and if you make the resistor adjustable, you can tweak it bang-on when warmed-up, and the diode (1N4148 will do) will track TR12's tempco. Without subjectivism there'd be no measurements to be objective about.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jun 13, 2007 3:42:11 GMT
Thanks Graham Very much appreciated. We will try to implement your suggestions in the new SC Class A 20W when we try it out. , als o see if existing can be readily alterd . PCB pattern permitting. P.S. Thank you for not calling me a "pajero" Alex
|
|
FritzS
Been here a while!
Sound of Blue Danube
Sound of Blue Danube
Posts: 1,364
|
Post by FritzS on Jul 17, 2007 16:21:43 GMT
Hi, I call Sennheiser (Germany) about in IEC61938 recommand 120 Ohm.
Sennheiser tells me that a good headphone amp for Sennheiser headphones should be have an output resister from 0 ... 10 Ohm - otherwise the bass will rise.
The answer from AKG coming later too .....
|
|
|
Post by PinkFloyd on Jul 17, 2007 19:03:41 GMT
Hi, I call Sennheiser (Germany) about in IEC61938 recommand 120 Ohm. Sennheiser tells me that a good headphone amp for Sennheiser headphones should be have an output resister from 0 ... 10 Ohm - otherwise the bass will rise. The answer from AKG coming later too ..... Jan meier has always offered Variable output impedance, one very low and the other 120 ohm and I quote: "Normally a headphone amplifier has a very low output impedance which gives the amp tight control over the headphone action. However, many headphones have been sonically optimized to be driven by an output impedance of 120 Ohm. This output impedance is an international standard." From another one of his articles: "The headamp has two sockets for connection to headphones. Both sockets will provide different sound characteristics. One socket has a very low output impedance and gives the amp tight control over the headphone action. However, many headphones have been sonically optimized to be driven by an output impedance of 120 Ohms and may sound better when connected to the other socket. Generally, the low impedance socket provides a clean sound whereas the high impedance socket yields a warmer sound. Use the one you like most. There is no risk of damage to your headphone by connecting it to either socket. You can also use the sockets to connect two headphones simultaneously. However, the volume produced by the high impedance socket will be slightly lower than that of the other socket." Sure, some headphones have been designed to be driven by an output impedance of 120R but which ones?? I too asked the guy at Sennheiser UK a couple of years back and he said "as low as possible, optimally 0 ohms" he didn't mention the bass rise phenomenon though Fritz... this is interesting. I wish all headphone manufacturers would actually state what standards the're designing to in the specs.... Most of the old headphones were to 120R but it seems most of them nowadays are favouring 0R.... it would be nice to get a definitive answer from all the headphone manufacturers, you've emailed Karl Peschel at AKG yes? He's the man to ask. Most people who have experimented seem to agree that their headphones perform better when driven from a low impedance source.... not always the case and some prefer 120 ohm but it certainly is dependant on what headphones you're using.... not all of them are created equal.
|
|
FritzS
Been here a while!
Sound of Blue Danube
Sound of Blue Danube
Posts: 1,364
|
Post by FritzS on Jul 18, 2007 11:04:35 GMT
Yes I have Karl Peschel is in semi-retirement - anyhow he answer K701 are constructed for low impedance output amplifiers.
But the rise of impedance at the resonance frequency is unincisive, so the effect to the sound is low.
The K701 works good (without diversification in sound) with all the current headphone-jack output-impedance.This is a corresponding translation from the mail I get ...... I hope I can spoke with him face-to-face
|
|
|
Post by PinkFloyd on Jul 18, 2007 12:18:00 GMT
OK so far we have Sennheiser saying "low impedance" and now we have AKG also saying "low impedance", seems they are no longer designing to the old 120R standard.... all we need now are answers from Beyer, Audio Technica, Grado etc. etc. I'll ask Grado, will go do that now. Cheers Fritz
|
|
|
Post by PinkFloyd on Jul 18, 2007 12:43:55 GMT
Have e-mailed Grado, Sennheiser UK and Beyer. Think I already know the answer from Sennheiser and Grado but Beyer may well reply with "120 ohm" I remember Jan Meier has visited the Sennheiser factory quite a few times and they told him that the Sennheisers are designed to sound as intended with an amp as close to zero ohm output as is possible. No real need to ask Grado seeing as the RA-1 amp has a zero ohm output impedance, I think that tells us what their 'phones have been designed for no harm asking though! IEC 61938 is one thing but how many headphone manufacturers actually conform to it? this is the question..... so far it seems Sennheiser, grado and AKG don't and all favour as close to zero ohm as possible.
|
|