A future range of course. It would not be feasible physically or , I suspect, economical, to retrofit it to Rossini and Vivaldi. There isn’t the space in the chassis and it would require replacement of the front panels to provide the 1/4" and XLR sockets. The cost of those front panels from supplier to dCS is substantial and if one adds the conventional fivefold increase to reach shop price … !
1 Like
Anup, I am glad that you used the thinking emoji to conclude. Succesful crossfeed circuits which include Expanse ( at the top end) are based around amplitude and phase behaviour. In turn the algorithms chosen may depend on their success upon known amplitude and phase characteristics of the analogue amplifier circuit that they feed. I also need to use 
as we just don’t know.
3 Likes
Exactly what I was wondering when I posted that reply 
1 Like
Definitely in the (fascinating) weeds, but here’s a chap who knows a thing or two about the processing we’re talking about:
Worth the hour in my opinion.
Disclaimer: he’s a friend, too. But he doesn’t give me a fiver for saying so 
If the Expanse crossfade DSP is taking place in the FPGA and through the dCS DACs across the line-up, then a provision at the back panel could be an extra analog headphone-only line out (balanced XLR and unbalanced RCA), to be connected to a separate (dCS or non-dCS) headphone amp.
A dCS headphone amp would give the advantage of best adaptation between the two.
1 Like
I am using a 3in x 3out XLR switch to connect all my gear.
In go the Rossini, a Neukomm CDA126S DAC and an empty connection for my testbed.
Out it goes to my Neukomm PA18 monoblocks and two headphone amps (Highamp Sirius for STAX, Bryston BHA1).
The switch was custom built by Dodocus Audio-Umschalter. The switch is completely passive using high quality switching elements. The blue LED does not come from the switch.
Dodocus are very flexible and custom build switching equipment. They can also do remote controllable, relays based etc.
Now if all the DSP processing in the Bartok is done digitally it would seem easy to implement the same thing across the product range, as the Bartok has essentially the same board as Rossini.
I would vastly prefer to keep the headphone amp separately. Otherwise STAX and Co will not be able to use the feature.
@Andrew maybe something for Rossini 3.0?
1 Like
Thanks Rudi, that looks like the perfect interface for using a separate headphone amp. But then we still need Rossini and Vivaldi to be able to output a normal line level, or an Expansed line level out. The Bartók puts Expanse only out from its headphone connections at the front.
As long as Expanse is purely digital, you could select off/on it like you do different filters. Should be doable via software.
I don’t assume you would want to listen to Expanse and Normal at the same time via two different outputs
1 Like
So, by a switch in software/ Mosaic you either route a Expanse-modified line output signal, or a normal line level output signal, from the line level outputs at the back. These are connected to a switch box as you have indicated. The box outputs to either a headphone amp, or a stereo power amp, or 2 mono power amps.
This would also work with my Benchmark HPA4, without the switch box.
… except as Pete suggested, it’s possible the Expanse DSP algorithm has to take the built-in Amp section characteristics into consideration - we won’t know unless someone from dCS comments
(Speaking of which, if dCS does do that, I wonder if Expanse 2.0 on the Bartok could pre-compensate for any analog non-linearity in the Amp section… or maybe it already does that? 
).
Thanks for posting that. This is precisely what I need in my headphone rack, or a version of it anyway. I’ve been using a nice Hattor passive pre, but it does more than I need it to. All I need is a quality switch. Inquiry sent.
They will get busy after the weekend. Inquiry sent too.
1 Like
If you mean the Benchmark HPA4, looks like they do exactly that, but in the analog domain; compensating for non-linearities with their “Feed-Forward Error Correction”
1 Like
As in the “white glove treatments” by Bob Stuart. I think only two were ever done. Probably more effective to use a linear amp from the beginning.
Not quite. Bob’s “White Glove” MQA transfers were required because of the old-world mixing used for those masters, requiring them to be compensated individually “by hand” (hence the “white glove” term).
The pre-compensation I was suggesting is pretty much exactly what the Benchmark guys are doing (as I quoted in the post above), but done in the digital domain with DSPs. Wadax’s reference DAC adopts a similar strategy to mitigate non-linearities in their system chain.
While this might be new on the Audio world, it’s been done for decades in the Telecom space with DSPs applying pre-compensation and FEC to mitigate non-linearities, in frequency, time, and polar domains.
1 Like
Headphone transducers are notoriously non-linear transducers operating in sub-optimal acoustic chambers. As such, a linear amp (of which there are many) is only one of the factors—and an oversold one at that, in my view—to take into account the effectiveness of the presentation of the system at the listener’s ears. It may well be that a linear amp presents an easier case around which to build DSP calculations, but it guarantees little more than flatness or neutrality in the outcome. Doesn’t mean it will be bad, only that it doesn’t guarantee anything else. And on many headphones, it can be less than ideal. Let alone ears. Accuracy is easy in hifi today. And linearity is merely one factor in final SQ. Granted, linearity in the amp might well be the best place to have it. Buy by itself? Otherwise, scores of midfi solid state devices from DACs to amps would sound as good as many devices known to be less linear.
Wadax Atlantis was one of two exotic European DACs I had wanted, but was unable, to demo. Anyone here heard it?
Headphone transducers have their own typical equalisation, this is a design decision of the manufacturer.
[A] factor that has to be taken into consideration is the fact that modern headphones
are designed to present a range of audio experiences with a wide range of source material
and listening environments while being fed simply from the original stereo signal; different
headphones differ in a number of characteristics, not least overall equalisation, where, for
example, a manufacturer may arrange for their headphones to give a “warmer” sound
by rolling off some HF or boosting at lower frequencies. A loudspeaker-to-headphone
conversion system has to function effectively whatever the headphone manufacturer’s
design decisions, and whatever are the listener’s purchasing decisions based on what they
like to hear.Good Headphones already take into account the frequency component
of Head Related Transfer Functions
From: https://www.dcsltd.co.uk/support/dcs-expanse-white-paper/
The more linear the headamp is, the more it respects the design decision of the headphone in question.
That is at best tautological. I am well-versed in HRTF theory. As I said, linearity in an amp is a good thing, but it is not the only thing. And it is why something like the Benchmark might quite simply not sound as good as another headphone amp. Linearity is not the only thing. Headphone designers design to a certain goal. Same with headphone amp designers that voice their amps using certain headphones. With apologies to JRRT, there is no “one amp to rule them all.”
Think about it Erno. Your claim—that the most linear amp is the best for all headphones, which is actually the claim of your previous post—is simply not credible. And it is also at odds with what you wrote previously:
I agree with you there. We know you love your Benchmark. Enjoy it, but please stop trying to prove it’s the best amp for every headphone. It’s not even close, considering the reference-class cans it’s not suitable for driving. 