Watts Up...?

[AnyDave]

Unfortunately not. I will be doing Singapore CanJam in early April though, with London CanJam in July. My next USA show is Socal CanJam, followed by Dallas CanJam.

Ah, that's too bad. Perhaps I'll make one of the CanJams and see you there.

 

[Jawed]

Just to re-cap there are broadly two forms of filters - causal filters (where the output depends solely on the current and past values) and symmetric acausal filters (where the output depends upon future and past values, and when the coefficients for samples offset into the past or the future are identical). Symmetric acausal filters have exactly the same output from an impulse whether the time is forward, or run in reverse or backwards in time. An example of a symmetric acausal filter is an symmetric FIR filter, like the WTA filter. Examples of causal filters are pretty much all analogue filters, and digital IIR filters (like the EQ in Mojo's UHD DSP), in short causal filters are real time filters.

FIR means finite impulse response - the output from an impulse stops once the impulse has past through all of the coefficients. An IIR filter is an infinite impulse response - as the filter feeds back to the input, it can create an output for an infinite amount of time, so long as the filter has infinite resolution.

The issue that's important here is that symmetric acausal filters pre-ring and post-ring, that is you get a response before the impulse. Causal filters on the other hand only create an output when an impulse occurs, or after an impulse has happened.

For a perfect interpolation reconstruction filters, it MUST be an infinite sinc function filter, and this is an ideal symmetric acausal filter. In reality, we need to reconstruct the timing of transients as accurately as possible, and that's what the WTA filter uniquely does, so that it reconstructs the timing of transients as accurately as possible given the limitations of real processing. But it will pre-ring; but that isn't a problem, as an appropriately bandwidth limited impulse will show minimal pre and post ringing with a WTA filter, and as the filter gets longer, the less ringing it will show. And if there was ringing, it's at an inaudible 22kHz (for 44.1k sample rate), so pre-ringing for an interpolation filter isn't an issue at all, in spite of what the rest of the audio industry says.

But if we are filtering in the audio domain, then pre-ringing is an issue, because pre-ringing artifacts would be in the audio bandwidth and hence audible, and it's clearly unnatural to have a response before the impulse. So it's very important to use causal filtering or IIR for EQ.

I want to avoid confusion related to EQ, per se, since a DAC does not need EQ to function.

In the past you discovered that adding a WTA 2 stage improved sound quality, as before that your DAC designs only had a single stage of WTA. WTA 2 effectively changed one IIR stage (assuming there's a set of IIR stages in the DAC, following WTA) as seen in Hugo 2 and later DACs that have the option to disable WTA 2, where the IIR stage is reconfigured to "connect" with WTA 1 if WTA 2 is turned off.

By adding a WTA 2 stage you effectively increased the reversibility of the DAC. It reduced the errors introduced by causality, amongst other things.

In theory you could create a DAC that uses WTA exclusively to produce 2048FS, either going in steps, WTA 1, WTA 2, WTA 3 say, or working in a single step. In this scenario, reversibility is maintained up until pulse array noise-shaping/modulation and the analogue filtering of the output stage.

So, what I'm asking is specifically about the sound quality problems associated with digital stages and a lack of reversibility. When WTA 2 was added, it seems possible that some of the sound quality benefit came from reversibility. In theory transient timing errors were reduced because of the increased degree of reversibility.

So is it possible to identify the sound quality problems associated with reversibility, independently of the use of a WTA stage (which is intrinsically reversible)?

There is one FIR that you've created that's not reversible, that's the video mode filter in Hugo M Scaler.

 

[Rob Watts]

I want to avoid confusion related to EQ, per se, since a DAC does not need EQ to function.

In the past you discovered that adding a WTA 2 stage improved sound quality, as before that your DAC designs only had a single stage of WTA. WTA 2 effectively changed one IIR stage (assuming there's a set of IIR stages in the DAC, following WTA) as seen in Hugo 2 and later DACs that have the option to disable WTA 2, where the IIR stage is reconfigured to "connect" with WTA 1 if WTA 2 is turned off.

By adding a WTA 2 stage you effectively increased the reversibility of the DAC. It reduced the errors introduced by causality, amongst other things.

In theory you could create a DAC that uses WTA exclusively to produce 2048FS, either going in steps, WTA 1, WTA 2, WTA 3 say, or working in a single step. In this scenario, reversibility is maintained up until pulse array noise-shaping/modulation and the analogue filtering of the output stage.

So, what I'm asking is specifically about the sound quality problems associated with digital stages and a lack of reversibility. When WTA 2 was added, it seems possible that some of the sound quality benefit came from reversibility. In theory transient timing errors were reduced because of the increased degree of reversibility.

So is it possible to identify the sound quality problems associated with reversibility, independently of the use of a WTA stage (which is intrinsically reversible)?

There is one FIR that you've created that's not reversible, that's the video mode filter in Hugo M Scaler.

Rather than talking about reversibility, it's more accurate to talk about the requirements to have full sinc function coefficients, as that's the important issue. The issue is sinc reconstruction, not whether it's symmetric FIR or not. It's only full sinc that will perfectly reconstruct.

You are right to mention reconstruction to 2048FS - but it turns out that the whole reconstruction issue to 104MHz is very much more complex than I have talked about so far. This is an issue I will be discussing further in the future.

 

[Jawed]

@Rob Watts an idea for Quartet Scaler: give it a mono mode with selectable channel. The user can then configure two QMSs, one for the left channel and the other for the right channel. 8 million taps...

 

[icnaiwhimor1975]

are there any technical limitations that prevent a DAC with 200 pulse area elements?

like 100 per channel

Or does it make no sense?

20 elements per channel - sounds impressive
but why not add even more..

 

[Daorom]

are there any technical limitations that prevent a DAC with 200 pulse area elements?

like 100 per channel

Or does it make no sense?

20 elements per channel - sounds impressive
but why not add even more..

I think there was a mention about radio interference created by many elements.

 

[Jawed]

are there any technical limitations that prevent a DAC with 200 pulse area elements?

like 100 per channel

Or does it make no sense?

20 elements per channel - sounds impressive
but why not add even more..

There was a time when Rob talked about Davina using 40 elements per channel, January 2018:

Doubling the number of elements halves the correlated jitter error (reduces random jitter error by 3dB), reduces noise by 3dB, improves high signal THD (that's very complicated to explain!) and improves small signal resolution by 6dB (but that is outweighed by other benefits as the noise shaper design is different, and that changes resolution by 50dB in the case of Dave against Hugo 2).

The Davina ADC project - because I wanted 135dB dynamic range - uses 40 elements as the pulse array reference for the ADC.

April 2016:

To answer all the questions it's yes,yes,yes,and finally yes.

For example Davina ADC will have 40 elements as that was the only way I could get the 350 dB noise shaper small signal resolution performance, and have Dave levels of distortion.

Its actually a complicated mix of different factors - generally, all things being equal, you get 3dB lower noise for doubling the number of elements, and a 6 dB improvement in signal resolution - assuming the noise shaper time constants do not change.

Rob

 

[icnaiwhimor1975]

There was a time when Rob talked about Davina using 40 elements per channel, January 2018:



April 2016:

as far as I understand "Davina ADC" is an ADC
this is what takes an analog signal at the input and turns it into a digital signal at the output
at least the abbreviation ADC hints at this

besides, as I understood - that this project was not implemented

I am only interested in DACs
I hear tangible advantages of 20 elements per channel compared to all the others

and I wonder why since 2015 the progress in measuring is on 20 elements per channel
in qbb76 there were 16 of them per channel
in DAC64 there were also quite a lot of them

 

[Jawed]

as far as I understand "Davina ADC" is an ADC
this is what takes an analog signal at the input and turns it into a digital signal at the output
at least the abbreviation ADC hints at this

besides, as I understood - that this project was not implemented

I am only interested in DACs
I hear tangible advantages of 20 elements per channel compared to all the others

and I wonder why since 2015 the progress in measuring is on 20 elements per channel
in qbb76 there were 16 of them per channel
in DAC64 there were also quite a lot of them

An ADC such as Davina needs to have a DAC inside it.

Rob is continuing to work on Davina. A lot of the motivation is to make DACs better.

 

[CIEM Patient Zero]

3 months ago there was a AMA (Ask Me Anything) with Bowers & Wilkins on Reddit. I'll try to post some comments, but it ended up not being too popular as Audio users are a tiny demographic of Social Media, add to that B&W is niche:

B&W == AMA - Andy Kerr of Bowers & Wilkins

?: Been rocking 803N for awhile now and thinking of updating to the D series though noticed the B&W sound has changed over the years, any plan to go back to the more warmer sound?

B&W: Think I've answered this one earlier but to come back to it – I don't disagree with you, we perhaps pushed it a little too far with early Diamond series models, especially D3. Working on feedback from reviewers, customers and studio colleagues, we think we've moved back towards a warmer tonality with some of our newer products (D4, 600 S3) and it lookalike the reviews are bearing that one out. But there's also something new to do! Perhaps try a listen to a pair of 805 Signature when you get the chance? That might give you some good insight into where we're going these days.

? A good number of people are saying that you need to break in a speaker for a certain period of time to deliver their full potential.

Does a speaker fresh out of the assembly line have a difference in sound compared to a speaker used for 1000 hours ?

B&W: I wouldn't necessarily say thousands of hours are required to 'break' a speaker in – but it's true that speakers do tend to require some use to operate to their best. That relates mostly to physical elements such as the stiffness of the suspension and the transducer surround. We also find that if speakers have been left in cold conditions for some time (perhaps in transit), they tend to sound more natural after they've been given time to come up to room temperature.

?: what role do materials play in your design philosophy and are there any new finishes or materials being explored for upcoming projects you can share?

B&W: Improvements in materials are hugely important to use from both practical acoustic and mechanical perspectives and also of course from an aesthetic viewpoint. We have constant, ongoing research into new transducer materials, into the impact of improved materials (eg aluminium) and processes on cabinet behaviour and so on: you can see evidence of that in, for example, the added aluminium bracing used in 800 Series Diamond, or the Aerofoil cones (made with carbon fibre skins) on some of our bass drivers.

We're continually looking at what's next in all of those regards. Can't share any details, unfortunately!

?: I'll kick it off just to get things rolling.

As Bowers & Wilkins has a significant history in this market, hopefully your experience can help shed light on the future of the space.

My question is, where do you see the audiophile market growing in the next 5 years? Where do you see it shrinking or maybe there being less emphasis?

B&W: I think it's stable, rather than growing. I think the biggest challenge we all face isn't so much the cost of components – of course, that's definitely an issue! – but the cost of real estate. For those of us fortunate enough to own our properties and have sufficient space, a large pair of floor standing speakers is something to aspire to, but I worry that the younger generation, for whom every square centimetre (or inch) of space is hugely expensive, might be less keen on the idea of surrendering a large part of their home to big speakers and the associated gear.

Now don't misread me, I think there's always a space for big speakers for certain buyers. But I think the challenge for us as an industry is also to demonstrate the potential for high quality in smaller, perhaps less space-hungry solutions. Active products could be an interesting part of that, but also EQ-enabled, DSP-driven speakers that are less sensitive to room positioning / room constraints etc.

?: Hi, Andy!

Been a HUGE fan of B&W since I first heard the Nautilus 802 in the late 90s. I watched all of your factory tours on YouTube and hope to do one in person one day.

I’ve owned many B&W speakers over the past 20 years and I currently have a pair of 800D2 in my living room. Before that I owned a pair of 802D1.

I was curious if you could speak to the evolution of B&W’s “house sound” over the last few generations. I auditioned both the 802D4s and 702S3 last year and found them to sound different than my older generations.

Wondering if I’m making that up or if there’s been a shift sometime after 2010.

Thank you for doing this AMA!

B&W:
It's an interesting observation. I don't think we're trying to move away from our roots in terms of what we like in terms of overall presentation, but at the same time I think it's fair to say that some of the more recent technologies we've introduced have some different tonal characteristics to older products. The most obvious example is Continuum, which we think sounds 'cleaner' than the older yellow cone material we were famously associated with. That material had what we used to call a midrange 'quack', which isn't present in the newer models.

We have also received comment from owners and from studio professionals that we've taken on board and incorporated over the years into products. A good example would be the D3 series of 800, which had (we felt) massively more resolution but which in hindsight could perhaps be a bit too 'hot' in the upper registers. We think we've improved that notably in the D4 era.

https://www.reddit.com/user/bowers-wilkins/

I found a video of a young Rob Watts and John Franks working behind a Hi Fi counter:




NSFW (Not Suitable For Work), Music Appreciation after an Apocolypse:

 

[Daorom]

Interesting, that amplifier Chord Suzi will play as cleanly as TT2?

 

[ecwl]

Interesting, that amplifier Chord Suzi will play as cleanly as TT2?

Somehow I doubt it.
Granted, I don't have Suzi or TT2
But I do have DAVE and Chord Etude. Both are based on/related to TT2 and Suzi technology.
I would say Etude is not as transparent as DAVE. Except DAVE can't drive my speakers so I need the Etude.
I even debated whether to switch to a TT2 just so that maybe the TT2 can drive my speakers loud enough (probably can since I don't listen loud)

That said, different people listen for different things. So I can see some people preferring Suzi over straight out of TT2. But since I don't own Suzi or TT2, I really shouldn't comment. not totally sure.

 

[Daorom]

And what do you say about Eversolo F10? + TT2.
Is it clean enough DAVE for Susvara?

 

[Triode User]

Somehow I doubt it.
Granted, I don't have Suzi or TT2
But I do have DAVE and Chord Etude. Both are based on/related to TT2 and Suzi technology.
I would say Etude is not as transparent as DAVE. Except DAVE can't drive my speakers so I need the Etude.
I even debated whether to switch to a TT2 just so that maybe the TT2 can drive my speakers loud enough (probably can since I don't listen loud)

That said, different people listen for different things. So I can see some people preferring Suzi over straight out of TT2. But since I don't own Suzi or TT2, I really shouldn't comment. not totally sure.

In a slightly different but similar vein, I preferred and found it to be more transparent to have Dave + Pass Labs XA30.8 power amp driving speakers compared to TT2 on its own.