[sbradley02]

I guess we'll have to agree to disagree. I can't ever countenance using design kludges instead of just doing the design using established methodology. Headphones in general just aren't that demanding of a load. Using the one example of a "balanced" amp I have specs on, it has a 25V swing. Using bipolar supplies (which I believe this unit doesn't) means you need a transistor with ~ 12V CE - trivial to find excellent devices with that rating. Audio power amplifiers often run with rails in excess of 40V (and I have yet to see any power amp which is floating ground, though they may exist). You mentioned OTL tubes. Lets look at some extreme cases for the heck of it. I used to play around with electrostatic speakers. There are both commercial and home brew OTL electrostatic amps out there, and we are talking about outputs in excess of 1000V. High voltage is what tubes do best. As a really extreme example I designed a high voltage output stage using a single tube (not for audio purposes) with a 7000V output swing and a slew rate of 60,000V/uS.
 
I maintain that there is no valid design rationale to double parts count in so-called balanced designs.
 
Quote:

Id disagree with many parts of that.
 
--snip--
What if your active devices are somehow limited in max voltage swing? OTL tubes run into grid current VERY soon for example.
 
 

 

[nikongod]

If you are this easily convinced that you are right, why start the thread?
 
Balanced amplifiers are established design. This is nothing new or unique to headphones, just the fringe of popularity.
 
OTL electroststic amps are an interesting point, could you show me a single ended one? 
 
There are plenty of balanced power amps. This includes amps that are *not* 4 channels slapped together in the fashion of most headphone amps. I mean 1 channel for both phases, 'cause thats fun. Its a shame the world went cheap with williamson, Blumlin was a better designer but williamson was cheaper. 
 
On the very very high voltage swings front: radio? Although the voltage swing & slew rates are indeed impressive I think that is apples and oranges. 

 

[sbradley02]

I started the thread because after reading a white paper the design didn't make sense to me and was looking for the reasoning behind it. The claimed benefits (for example higher swing and faster slew) can readily be obtained with more conventional designs. Kwkarth's explanation still makes the most sense to me. If I can find another reason for the design that couldn't just as easily be obtained through a more conventional circuit I would be happy to change my mind on this. So far I haven't seen any presented.
 
To continue I would like to try to define terms since some of these names can be used for different purposes. Correct me if any of this is in error. I will try to use terms as they are most commonly used in the industry.
 
Push-pull amplifier: By far the most common. At least two devices (or more in parallel) must be used in the output stage. One device handles the positive waveform, the other the negative. Class (A, B or AB) defines the degree of overlap in the two phases. Solid state can be complementary symmetry with N-type and P-type devices or quasi-complementary with two devices of the same type. Tubes must use a phase splitter. Inputs can be balanced or unbalanced. If it us a true balanced input the input stage should be configured as a differential pair which provides the advantage of common mode noise rejection. Balanced inputs are pretty much universal in pro audio and fairly common in high-end audio. Ground referenced output.
 
Single ended amplifier: A single device (or multiples in parallel) handles both positive and negative signal portions. Must be class A by definition. Again inputs can be balanced or unbalanced. Extremely inefficient but it has its adherents. Output is ground referenced.
 
Bridged amplifier (this is what we are referring to as a balanced amp in this thread): Two completely separate amplifier sections per channel. Since they are run with opposite signal polarities, when one amp is running positive the other will run negative. I believe that this can be described as source/sink. Input can be either balanced straight in or unbalanced in with a phase splitter (I have bridged amps by just building a dual op-amp circuit, one run inverting and one run non-inverting - crude but it works). Since the amps are not differential configured there is no common mode noise rejection. Ground must be floating. I imagine that the amplifier sections can be either push-pull or single ended, the white paper didn't state, but I'd suspect push pull as this is more common.
 
Now that terms are defined, lets find some OTL designs. I know there are many more out there, these popped up with a few minutes searching.
 
One the classic designs is the Futterman. Here is a schematic for the OTL-3. Push-pull, unbalanced input. Definitely ground referenced:
http://www.mcmlv.org/Archive/HiFi/OTL3.pdf
 
 
Headphone OTL amp with balanced input and push-pull output. Ground referenced so this is not a bridged amplifier:
http://www.diyaudio.com/forums/headphones/180740-balanced-input-headphone-amplifier.html
 
Doesn't seem to be as much info on OTL electrostatic amps online as there used to be, but perhaps more germane to the discussion would be an electrostatic headphone amp:
http://gilmore2.chem.northwestern.edu/projects/gilmore_prj.htm
I find this one especially fascinating in terms of this discussion. People have called moving coil headphones balanced devices. No they are not. An electrostatic headphone (or speaker) is. Note that here we have an actual balanced load, yet the input is unbalanced and the output is in effect push pull (one could also argue that the output is balanced since there is no common signal conductor - thoughts?). What we don't see here is a bridged amplifier design, even though we are driving an actual balanced load. As an aside my fantasy headphone rig would be a truly portable electrostatic setup (and not the in-ears once sold by Stax). This doesn't seem to be possible unfortunately.
 
And yes, you caught me out on the HV tube design, it was a signal generator and used a transmitter tube called a planar triode, an extremely nifty and compact device. While it is apples and oranges compared to audio design, my intended point was that the design should drive the parts selection and that an appropriate device can be found for the task. The devices should not drive the design.
 
Quote:

If you are this easily convinced that you are right, why start the thread?
 
Balanced amplifiers are established design. This is nothing new or unique to headphones, just the fringe of popularity.
 
OTL electroststic amps are an interesting point, could you show me a single ended one? 
 
There are plenty of balanced power amps. This includes amps that are *not* 4 channels slapped together in the fashion of most headphone amps. I mean 1 channel for both phases, 'cause thats fun. Its a shame the world went cheap with williamson, Blumlin was a better designer but williamson was cheaper. 
 
On the very very high voltage swings front: radio? Although the voltage swing & slew rates are indeed impressive I think that is apples and oranges. 

 

[kwkarth]

Quote:

The last time I quoted something incorect that you wrote you deleted it. It was tangential to the discussion in the LCD2 thread where we were chatting about helmholtz resonators: you said that the DIAMETER of the tube effected the resonant frequency. 
 
Lets see how it goes this time.
 
Speakers are balanced. Its the cables that are not.

If you want to create trouble, do it in your own thread.  Speakers are not "balanced" devices.
 

 

[MrGreen]

Quote:

If you want to create trouble, do it in your own thread.  Speakers are not "balanced" devices.
 

I'm pretty sure push-pull electrostats are.
 

 

[sbradley02]

I wanted to examine this claim also, especially since it references hard data and therefore should be easy to define. It this can be backed up with real world examples, it would be very interesting. First, a properly designed push-pull amplifier should have vanishingly low distortion. For example, here are extensive measurements made on a SS amp that costs 1/4 of what the cheapest balanced designs I have been able to find do: http://www.leckertonaudio.com/2011/01/uha-6s-harmonic-distortion/
 
So lets say for some reason we have a pair of push-pull or single ended amplifiers with mediocre measurements (I am not sure why the we would be starting with units that don't measure well, but lets ignore this for now) in a bridged configuration. What seems to be claimed here is that the two amplifier sections would produce distortion products out of phase and would therefore cancel at the load. Is this indeed the reasoning behind this claim, or is something else going on? If so the amplifiers would have to be matched to an astonishing degree. And measurements would show this result. Do you have any example measurements we could look over?
 
EDIT: After thinking this over a bit I came up with a possibly more plausible mechanism for a reduction in distortion going to a bridged design. If you look at the competitor amplifier in the comparison above it exhibits an extreme hockey stick distortion curve well within expected operational conditions. Bridged operation reduces the output power by half and also the perceived load by half, which could push the amplifier back out of the hockey stick portion of the curve.
 
I have often seen the claim that great sounding amplifiers should not  measure well. I don't buy it (yes some measure poorly but I don't see this as a requirement for good sound). And yes I am going to use speaker amplifier examples because that is what I have more experience with.
For tube amplifiers it doesn't get much better than Audio Research. Here's an example: http://www.audioresearch.com/Reference110.html. 0.3% THD at full power which is exceptional for a tube amp, and .03% at 1W which is excellent by any standard, and in the headphone power range.
 
For solid state amplifiers the Classe' consistently get great reviews and I have owned one of their amps. An example:
http://www.classeaudio.com/omega/omega-mono.htm
.002%, again very low.
 
I would love to see some specs posted on some of the more highly regarded (and expensive) headphone amplifiers, it would be very informative to contrast and compare.
 
Thanks
 
Quote:

Id disagree with many parts of that.
 
A balanced or bridged amp will only measure like a single ended (drives the load from both side, very few people use the term properly) if the single ended amp measures very well. If the single ended amp has significant even order distortions the bridged and SE versions will measure very differently.
 
--snip--

 

[Steve Eddy]

Quote:

Originally Posted by sbradley02 /img/forum/go_quote.gif
 
I find this one especially fascinating in terms of this discussion. People have called moving coil headphones balanced devices. No they are not.
 

Could you elaborate on this?
 
Thanks.
 
se
 
 

 

[kwkarth]

Quote:

I'm pretty sure push-pull electrostats are.
 

Yes, you are right, some electrostatic headphones are a true bipolar/balanced load where charge on the stators on either side of the statically charged diaphragm are modulated with the audio signal.  I don't know of any dynamic "speakers" that are balanced.  There is a voice coil with two wires coming out, one for each end of the one voice coil wire and that's it.
 

 

[jcx]

2-terminal passive devices do a really good imitation of "balanced" - they only respond to the difference V across their 2 terminals, they don't care at all about common mode V
 
any real physical system has additional capacitance to to the rest of the world - in dynamic headphones this C has negligible effect at audio frequencies, and is pretty well balanced (equal on each terminal) for typical construction
 
dynamic headphones are really good "balanced" loads - circuits like Pass' SuSy that require balanced load will "play nicely" with dynamic headphone loads
 
 
 
oddly enough electrostatic headphones are poorly balanced loads - due to the cable C being a large fraction of the transducer impedance and the wire-to-wire fringing C is quite unequal for the typical flat cable with center bias
 
if you have noticeable equal series impedance in +/- stator outputs the unequal + vs - wire's cable C will roll off the +/- drive unequally

 

[kwkarth]

Quote:

2-terminal passive devices do a really good imitation of "balanced" - they only respond to the difference V across their 2 terminals, they don't care at all about common mode V
 
any real physical system has additional capacitance to to the rest of the world - in dynamic headphones this C has negligible effect at audio frequencies, and is pretty well balanced (equal on each terminal) for typical construction
 
dynamic headphones are really good "balanced" loads - circuits like Pass' SuSy that require balanced load will "play nicely" with dynamic headphone loads

Would you elaborate on this please?  Where is the "ground" reference in your balanced description?  What is the practical purpose of referring to the obvious symmetry of an audio signal connected to a voice coil?  How would one call it unbalanced other than to create a common ground for a two channel system, or use chassis ground as return path?  In so doing, in a headphone amplifier context, what have we accomplished in terms of real world practical operational advantage?  Although I suspect you're trying to use semantic games to make some sort of point, I want to make sure you get a chance to explain your statement better.  Remember, the goal here is to try and help us all understand the technology of our hobby batter in order to enable greater enjoyment from our hobby.
 
Thanks.

 

[jcx]

the semantic game was already in play - any time you discuss "balanced" in audio
 

 
Quote:

...the main 1st point is to understand that the term "Balanced" has different meanings/usages to different audiences - and even in one domain may be "overloaded" and its meaning change with context
 
professional audio often uses balanced signaling differential transmitter/receivers and balanced impedance transmitter/receivers/cabling - Bill Whitlock has pointed out the distinction between differential signaling and the role of balanced impedance interfaces for common mode interference rejection
 
some high end consumer audio components have adopted the pro "Balanced" signal interface for the source (most often DAC) analog out to the amplifier input
 
"Balanced" headphone drive is more often about opposing polarity signals to each driver and is less often clear about the balanced impedance (which is arguably inconsequential in headphone drive) - in loudspeakers amps that drive opposing polarity to the loudspeaker terminals are call "Bridged Amps"
 
Locally we also use "balanced headphones" (or balanced headphone cabling) to distinguish between the standard TRS and any of the several cable connectors that give separate connections to each of the 2 headphone driver's 4 wires - however nothing says that just because you have 4 wires available that you have to apply "balanced" opposing polarity drive signals - but most here would consider it cheating to call a amp "Balanced" that wasn't Bridged (or use balanced output xfmr) even if you could only plug in "Balanced (cable/connector) headphones
 
 
it is not uncommon for headphone amplifiers that are advertised a Balanced in, Balanced Output to be built of 2 single ended amps for each channel and they do not have common mode rejecting "true differential" inputs and absolutely require opposed polarity "balanced" input signals and use the source shield gnd as signal ground

 
so as long as you have "balanced" 4 pin cabling with 2 wires to each 2-terminal dynamic driver you can consider dynamic headphone drivers in the "balanced impedance interface" sense as really superior differential sensing, common mode rejecting "receivers"
 
but any 2-terminal device really doesn't care whether the V drive source is "balanced" in the equal, opposing, +/- polarity sense
 

 

 

[Steve Eddy]

Quote:

Originally Posted by kwkarth /img/forum/go_quote.gif
 
I don't know of any dynamic "speakers" that are balanced.  There is a voice coil with two wires coming out, one for each end of the one voice coil wire and that's it.
 

How does that description differ from the description of the primaries of the input transformers that I use? There are two wires coming out, one for each end of the one primary winding and that's it.
 
Yet the inputs of my input transformers are indeed balanced.
 
se
 

 

[kwkarth]

Quote:

the semantic game was already in play - any time you discuss "balanced" in audio
 
so as long as you have "balanced" 4 pin cabling with 2 wires to each 2-terminal dynamic driver you can consider dynamic headphone drivers in the "balanced impedance interface" sense as really superior differential sensing, common mode rejecting "receivers"
 
but any 2-terminal device really doesn't care whether the V drive source is "balanced" in the equal, opposing, +/- polarity sense

Right, so, again, to help us all practically apply this to our hobby, there is no inherent technical or performance advantage to calling a headphone amp "balanced."
The important potential performance advantage is in isolating common L+R return path from cross channel modulation, which can accomplished in a number of ways which do not need to be called "balanced."  Right?  Thanks.

 

[sbradley02]

http://en.wikipedia.org/wiki/Balanced_audio
The purpose of balanced inputs and outputs is common mode noise rejection. That is it. For this to work, you must have a device that takes the sum of two signals of opposite polarities, i.e. a differential pair: http://en.wikipedia.org/wiki/Differential_pair
I can't see how noise that is conducted equally to both terminals of a moving coil speaker (whether you are using a bridged amplifier or a ground referenced amplifier doesn't make any difference here) results in the elimination of said noise.
I have done home audio installs. It is a guideline to never allow a speaker line to parallel a power line less than a foot away (preferably further). Failure to follow this can result in 60Hz conducted into the speaker line. The speaker will not, under any circumstances, reject this 60Hz interference. Not with a push pull amp, not with a bridged amp, not with any amp.
The definition of balanced signals is well and precisely defined in the industry. Changing the meaning of those terms is counter-productive.
 
Bridged amplifiers are not balanced. Headphones (with the possible exception of electrostatics) are not balanced. Neither will reject common mode noise.
If this is in error, and you have examples of these devices able to reject common mode noise, please provide that so we can go over it further.
 
Quote:

How does that description differ from the description of the primaries of the input transformers that I use? There are two wires coming out, one for each end of the one primary winding and that's it.
 
Yet the inputs of my input transformers are indeed balanced.
 
se
 

 

[jcx]

 
Quote:

...I have done home audio installs. It is a guideline to never allow a speaker line to parallel a power line less than a foot away (preferably further). Failure to follow this can result in 60Hz conducted into the speaker line. The speaker will not, under any circumstances, reject this 60Hz interference. Not with a push pull amp, not with a bridged amp, not with any amp.
... 

you're not clear on common mode vs differential in that instance
 
the only way line noise couples to speaker cabling with audible result is through differential coupling - a time varying magnetic field from the outside induces a V difference when it "threads the loop" of the +/- speaker cable wires - just like a air core transformer
 
twisting wire helps for mag fields that are uniform over several times the twist pitch rate because the field induces mostly canceling V in each opposite orientation loops of the twist - star quad or coax can be even better
 
but if you measure the common mode V at the end of these high (differential) canceling cable constructions you still measure a V difference induced by the changing mag field that is equal in both wires and gives no output from the speaker