[nikongod]

I would say same or slightly higher cost. 
 
If you wire the boards up differently to take better advantage of CMRR the cost is basically the same. I should note that I havn't tried with any of the popular amps (B22, M3) so it may not be possible without causing instabilities.
 
On the more expensive front: Transformers make all of this this very easy. I wish more people would use input transformers. For what people spend on 4-channel builds a pair of input transformers would barely make a dent in the wallet. They are very "not" measurements first, but have so many nice things going for them. ground-loop proofing is pretty slick IMO.

 

[jcx]

but xfmr also pretty much demand blocking Cap selection too - inducing audiophile anxiety

 

[sbradley02]

I see what you are referring to. My comparison would be to a single board per channel solution with double the supply voltage and a fully differential input section. This would seem to be cheaper with around half the parts count (though likely some more expensive parts).
But I admittedly don't have any hard data on cost comparisons. The bridged units I have seen just seem awfully expensive for what is being delivered.
I don't have enough familiarity with input transformers.
Quote:

I would say same or slightly higher cost. 
 
If you wire the boards up differently to take better advantage of CMRR the cost is basically the same. I should note that I havn't tried with any of the popular amps (B22, M3) so it may not be possible without causing instabilities.
 
On the more expensive front: Transformers make all of this this very easy. I wish more people would use input transformers. For what people spend on 4-channel builds a pair of input transformers would barely make a dent in the wallet. They are very "not" measurements first, but have so many nice things going for them. ground-loop proofing is pretty slick IMO.

 

 

[Steve Eddy]

 
Quote:

And I believe (correct me if I am wrong) that cancellation at the load depends on having essentially perfectly matched amplifiers for the two phases.

Yes, any difference between the two amplifiers would convert common mode noise to differential mode noise.
 
se
 
 
 

 

[Steve Eddy]

Quote:

but xfmr also pretty much demand blocking Cap selection too - inducing audiophile anxiety

What blocking cap?
 
se
 
 

 

[jcx]

do you trust your $100 of Permalloy to any source's possible DC offset? - we are talking "audiophile" here where such niceties as source/preamp output AC coupling may have been sacrificed at the altar of some Guru's irrational prejudice
 
of course you do know how to demagnetize - not everyone will

 

[SP Wild]

Quote:

but xfmr also pretty much demand blocking Cap selection too - inducing audiophile anxiety

Whats an xfmr?

 

[nikongod]

 
Quote:

do you trust your $100 of Permalloy to any source's possible DC offset? - we are talking "audiophile" here where such niceties as source/preamp output AC coupling may have been sacrificed at the altar of some Guru's irrational prejudice
 
of course you do know how to demagnetize - not everyone will

Outputs with DC on them screw everything up. I think most DIYers are very conscious of this. 
 
If the amp had DC coupled inputs or was DC coupled end to end any DC on the inputs screws everything up too. 
 
You could always eliminate some cap(s) in the DAC/preamp by transformer coupling that too.
 
Quote:

Whats an xfmr?
 

Its an abbreviation for transformer.

 

[SoupRKnowva]

what a great thread. i think i finally understand the difference between just a bridged balanced amp, and a differential balanced amp. i was always curious, cause when the b24 came out, some people were asking if they could lower the gain enough to use it with headphones instead of the b22, since it was fully differential.
 
Now what im curious about is that in order to reject the common mode noise at the amp, wouldnt you have to amplify the difference between the two phases, therefor producing a single ended output? so i guess what im asking is how do you have a differential amp that has 4 phases going in(2 channel balanced operation) and yet still have 4 phases(2 channel balanced operation) going out the output?
 
 

 

[Steve Eddy]

Quote:

do you trust your $100 of Permalloy to any source's possible DC offset? - we are talking "audiophile" here where such niceties as source/preamp output AC coupling may have been sacrificed at the altar of some Guru's irrational prejudice

Ultimately I see that as the responsibility of the owner. The product should be labeled to clearly state that the source should be free of any significant DC offset and what the maximum output should be.
 
 
Quote:

of course you do know how to demagnetize - not everyone will

 
A CD or downloadable file could be made available for that.
 
se
 
 

 

[Steve Eddy]

Quote:

Originally Posted by SoupRKnowva /img/forum/go_quote.gif
 
Now what im curious about is that in order to reject the common mode noise at the amp, wouldnt you have to amplify the difference between the two phases, therefor producing a single ended output? so i guess what im asking is how do you have a differential amp that has 4 phases going in(2 channel balanced operation) and yet still have 4 phases(2 channel balanced operation) going out the output?
 

Just look at the B24.
 
It's fully differential from input to output.
 
Of course common mode noise rejection is really most important at the input so you don't necessarily have to go the fully differential, B24 route. You could have a transformer coupled input which would take care of common mode rejection at the input, the amplification itself could be single ended, and drive an output transformer to get a balanced output.
 
se
 
 
 
 

 

[Yoga Flame]

I'm still seeking further understanding. Please correct me where I am wrong.
 
Common mode noise is noise that is added equally to both the + and - wires (for balanced), or both the + and ground wires (for single ended). When a headphone transducer rejects common mode noise, it does so equally well with both balanced amps, and single ended amps.
 
But is it the case that common mode noise is the same for both balanced and single ended signals? Certainly that is true for noise from electromagnetic interference. What about noise introduced by the amplification circuit itself? In a single ended amp, only the + signal is amplified. So noise from that stage will not be rejected by the transducer. In a balanced amp, any noise injected into the + signal will also be injected inversely into - signal. So that will get cancelled out. Correct?
 
Lastly, how would each of the following compare in terms of transparency (in theory, of course):
 A. balanced DAC -> drop half -> single ended signal -> single ended amp
 B. balanced DAC -> opamp -> single ended signal -> single ended amp
 C. balanced DAC -> transformer -> single ended signal -> single ended amp
 D. balanced DAC -> balanced signal -> bridged balanced amp
 
 E. balanced DAC -> balanced signal -> fully differential balanced amp
I've listed it from worst to best according to my uninformed opinion. Did I miss anything?
 
Thanks.
 

 

[nikongod]

Quote:

 
But is it the case that common mode noise is the same for both balanced and single ended signals? Certainly that is true for noise from electromagnetic interference. What about noise introduced by the amplification circuit itself? In a single ended amp, only the + signal is amplified. So noise from that stage will not be rejected by the transducer. In a balanced amp, any noise injected into the + signal will also be injected inversely into - signal. So that will get cancelled out. Correct?
 
Lastly, how would each of the following compare in terms of transparency (in theory, of course):
 A. balanced DAC -> drop half -> single ended signal -> single ended amp
 B. balanced DAC -> opamp -> single ended signal -> single ended amp
 C. balanced DAC -> transformer -> single ended signal -> single ended amp
 D. balanced DAC -> balanced signal -> bridged balanced amp
 
 E. balanced DAC -> balanced signal -> fully differential balanced amp
I've listed it from worst to best according to my uninformed opinion. Did I miss anything?
 
Thanks.
 

You can theoretically create a system where a single ended amp will appear to have common mode rejection ratio. It actually has more to do with the source than the amp. If you read back a few pages the "key" to balanced signal lines having CMR is not that they are "balanced" per-se its that all impedances to ground or other references, capacatances, etc are equal on both wires. Impedances/resistances to ground are key. If the source has an EXTREMELY low output impedance a single ended line will have CMR. The problem is that getting a low enough output impedance to make this work is a lot of work. A balanced interconnect has CMR with a source with any output impedance.
 
Regarding junk introduced by the amp its self.
This is theoretically a problem for balanced amps, although at the same time a blessing. 
Its a problem because "random noise" is twice as bad on a balanced amp. Unless your talking about a phono stage random noise quickly fades in importance (although headphones are tricky!)
It is a blessing because CMR works all over the place. a single ended amp has a given amount of PSR (power supply rejection - how well the amp rejects noise floating around the power supply) When you add CMR to PSR you get a very pronounced combined effect. More complex amps with constant current sources, and voltage followers all over the place dont benefit as obviously from this, but the effects still add up very nicely. Comparing a SET tube amp to a fully differntial push-pull tube amp using the same power supply demonstrates a night and day difference as far as PSR is concerned. 
 
A through E: 
Pretty much agree. Some would rank the opamp balanced-se converter (B) as better than the transformer (C). The difference comes down to: better measured performance with an opamp VS ground loop proofing with the transformer. Other considerations are that the opamp is generally less expensive by about half the cost, perhaps even cheaper if its on a big PCB with a shared power supply with the rest of the DAC. Transformers on the other hand dont require a power supply, you just wire them in and they do their thing. Pick your poison. 

 

[sbradley02]

I would also go with E as the best (and ideal IMO), though we have been debating here whether there is any benefit to a balanced output at the final stage.
Anyone have any schematics as to what a balanced output stage would look like (as opposed to what I am used to, a standard push-pull differential output, referenced to ground)?
Would be interesting to compare and contrast. The bridged amp topology is obvious, this bit is less so.
 
EDIT:
Did some checking around and it seems like the answer is, it depends. In the case of a transformer coupled output, like the McIntosh (solid state) and possibly the Audio Research, the output transformer appears to act as a balanced to single ended conversion stage. 
For OTL I found what is described as a true balanced output: http://www.tubecad.com/august99/
Would anyone who knows tube circuit design care to say what they think of this configuration? It looks interesting and I wonder if some of the commercial headphone amps use a circuit something like this. This is just a balanced output stage and not a completely balanced amplifier.
 
Thanks
 
 

 

[KingOfTheWild]

Wow! I guess I need to stay out of the Sound Science section of the forums . . . I didn't understand any of this at all (don't bother trying to explain . . . I'm sure that you won't be successful in educating me)