[GoldenSound]

In the first post @iFi audio mentions the DAC can go in NOS mode. I love NOS sound and my main DAC is a TDA1541A tube DAC so is this BurrBrown DAC chip able to work and sound like those old NOS chips?

The Diablo doesn't have the 'NOS'/'Bitperfect' mode that the iDSD signature has.

Though it's not ACTUALLY NOS, it just provides an output extremely similar to NOS. The chips are not native PCM converters like an R2R ladder, the native accuracy of the converter is 6 bit and so running it truly NOS would yield a maximum dynamic range of 36dB.
The simple explanation is that the BB/TI 'advance segment' chips have an architecture where the top 6 bits are converted 'as is', and the lower 18 bits are converted using a delta sigma modulator.

Some products like the T+A DAC200, Teac UD501 etc have a 'NOS' mode on these too, where the initial oversampling filter is bypassed, but you still have to go through the second stage oversampling/modulation else the DAC effectively wouldn't work properly.

So you still get an output that looks kinda like NOS with this approach:

though can do the same with a fully oversampling DAC if you use a sample + hold filter anyway, the ADI-2 for instance:

If you want genuine NOS you have to have a converter that is able to convert a PCM sample natively at its native bit depth without further digital processing.

 

[MSXX]

Currently, I have just the original Diablo here. I ordered the Diablo 2 to compare it. Why? Because on paper, the Diablo 2 is so close to be my perfect device (portable, powerful, Bluetooth, nice build quality, desktop mode). The only thing I am missing is Xbass, but with my current headphone and IEMs I do not need it. My plan was to sell my desktop and mobil setup and have just 1 device which can handle everything and every use case.

Since the last 2 pages, I am not sure anymore if the Diablo 2 is such device. It seems it has more technical flaws than I expected


I would not put my hand into the fire for my statement. I just remember, but maybe I am wrong. But I am sure it was somewhere in the ifi announcement thread (if true) --> https://www.head-fi.org/threads/ifi-audio-idsd-diablo-a-portable-reference-done-our-way.951615/

I am in the market for the same thing and I will also sit this one out. Thanks for a very illuminating discussion.

It seems to me that the Diablo 2 if not flawed doesn’t really improve on the last design. It don’t care about new features such as Bluetooth but more interesting in a better amp/dac.

I will put my hands in the hornets nest and claim that I sometimes wish ifi would spend a little more time on development then launching new products with a thousand features. My ifi zen stream could definitely use some of that - it’s a mess and very unreliable to use.

 

[Thorsten Loesch]

In the first post @iFi audio mentions the DAC can go in NOS mode.

Yes. As in "Nitrous Oxide".

All marketing fluff. On the old Diablo the gain/power modes were labelled like on iDSD Micro as "eco", "normal" and "Turbo".

They simple changed the names to "normal", "turbo" and "nitro". So the "normal" mode has been elevated to "turbo" and what used to be turbo is suddenly "nitro", without actually delivering more power or sounding different.

I love NOS sound and my main DAC is a TDA1541A tube DAC so is this BurrBrown DAC chip able to work and sound like those old NOS chips?

IN PRINCIPLE, Yes. I'm also a fan of the TDA1541 and I designed the iFi products to get as much as possible of this sound, using chip's remaining in production.

In practice there is no switch to select filter modes on either Diblo and some other bits that were needed for Non Over-Sampling mode were deleted on the original Diablo as it was explicitly not to have selectable digital filtering and/or Non Over-Sampling.

Depending how the new XMOS chip has been implemented it might be in theory possible to have a separate firmware working Non Over-Sampling, iFi have to answer to that and to if they would offer such a firmware.

Is the BB chip used the same as the one in other DAC/AMP from iFi

Yes.

and does it sound the similar when using the line out of the Diablo2?

I would expect so, the circuitry is the same as the original Diablo which in turn shares a design pretty much used all across the iFi line.

Is the line out dedicated with very good parts in its signal path or since most users use Diablo like device as DAC/AMP then it’s not worth using high quality parts for the line out?

The line out is decent enough. The original iDSD micro used OPA1642 J-Fet Input Op-Amp's in the analogue stage. The Diablo, as a balanced out was needed and thus double the number of Op-Amp's switches to an OPA1679 which is Mosfet input. Sonic difference between both are very slight, even though the OPA1642 costs nearly triple for a dual compared to OPA1679 for a QUAD.

I’ve been looking for a « battery transportable » DAC for a long time to feed my Woo WA8 transportable tube amp but I usually dislike the sould of ESS and AKM DAC chips so I’ve been waiting for someone to finally make an R2R transportable DAC (not dongle), doesn’t seem like this will ever happen!

If you want NOS sound with a BB Chip look for a second hand iDSD micro Black Label (or Signature/Finale if the price is right). It is all around the more versatile product, can select Non Over-Sampling mode for the DAC and it has more of a "tube sound" Headphone amp (the Diablo was designed for a very "dry" sounding Headphone amp).

You do lack the BT connection and if that is really important obviously the iDSD micro BL/Sig/Fin doesn't have this.

Thor

 

[Thorsten Loesch]

The Diablo doesn't have the 'NOS'/'Bitperfect' mode that the iDSD signature has.

Yup.

Though it's not ACTUALLY NOS, it just provides an output extremely similar to NOS. The chips are not native PCM converters like an R2R ladder, the native accuracy of the converter is 6 bit and so running it truly NOS would yield a maximum dynamic range of 36dB.
The simple explanation is that the BB/TI 'advance segment' chips have an architecture where the top 6 bits are converted 'as is', and the lower 18 bits are converted using a delta sigma modulator.

There are few interesting things to note. Delta Sigma and Multi-Bit are very different and have what I'd call "complementary flaws".

A Multibit DAC has very good resolution at high levels, the closer we get to zero the worse it gets. On the other hand, delta-sigma converters (as in pure single bit) are best at low levels and degrade the closer levels get towards full scale.

These days we get a variety of "hybrid" DAC architectures with a relatively opaque internal operation. Back when the specific system in use by most older Burr Brown DAC's was patented and put into production there was nothing like it.

So you get the low level performance of DS with the high level performance of MB.

The digital filter is really a completely different issue.

Let's not forget, a "Class D" Amplifier is a first or second order DS modulator usually at 250 - 500khz max.

Some products like the T+A DAC200, Teac UD501 etc have a 'NOS' mode on these too, where the initial oversampling filter is bypassed, but you still have to go through the second stage oversampling/modulation else the DAC effectively wouldn't work properly.

Yes, a 3rd order modulator with 5-Level output. This means in effect each phase of the DAC analogue output can be at "center" or at +1 or at -1, once the two phases are summed we get -2/-1/0/+1/+2 Values. There is however just a digital domain DS modulator, no further digital filtering.

In principle any DAC with the right BB Chipset can operate "Digita Filterless".

So you still get an output that looks kinda like NOS with this approach:



though can do the same with a fully oversampling DAC if you use a sample + hold filter anyway, the ADI-2 for instance:

The ADI-2 also uses oversampling DAC IC's. They just have different digital processing and analogue side, that in effect allow them to match the NOS waveform more closely whereas the BB Chip's evidence some ultrasonic noise.

If you want genuine NOS you have to have a converter that is able to convert a PCM sample natively at its native bit depth without further digital processing.

NOS, that Non Over-Sampling is literally a marketing term. In technical terms using "Digital Filterless" is more accurate.

The next question is what actually produces the characteristic "NOS Sound", in other words which part of the objectively poor performance is responsible for the sound characteristic.

Is it the presence of signal correlated ultrasonic images that mostly consist of high frequencies (the "unfolding" claimed for MQA uses this effect to create an apparent extension of the audio bandwidth past the limit set by the sample rate, Pioneer used a similar system called "Legato Link" to claim the restoration of ultrasonic content removed in the recording stage).

Is it specifically the multibit DAC used?

Is it the SINC rolloff of around 3dB at 20kHz?

Is it the Impulse response free from ringing?

Is it the label "NOS" on the front panel?

A combination of some/all of the above?

Something not included in the above but related to the operation of the human hearing where the hearing system forms one "interference grid" and the NOS Audio signal forms a complementary grid (I'm really trying to express a 4D effect in 2D metaphors here)?

Answers on a postcard.

Thor

 

[Thorsten Loesch]

It seems to me that the Diablo 2 if not flawed doesn’t really improve on the last design. It don’t care about new features such as Bluetooth but more interesting in a better amp/dac.

The "transportable" DAC Range of iFi (I note the xDSD/Gryphon as portable) all go back to a 2014 design (the original iDSD micro in silver) and have seen minimal changes. Where there were changes they were either forced by parts obsolescence or by marketing requesting feature changes (which led to the Diablo) without really improving the product.

Now I wish I could claim that my 2014 product design was so perfect, it is still valid. Yet even at the time I was aware of some compromises driven by a number of factors, so had there been a true 2018 model, it would have been very different.

I will put my hands in the hornets nest and claim that I sometimes wish ifi would spend a little more time on development then launching new products with a thousand features.

Around a decade ago this was the case. The products were few and they were designed with a lot of thought and tested thoroughly. Over time iFi decided that the best way to higher sales is not to make superior quality products but just more of them. A new product will easily give a quick rise in sales, while keeping sales of a mature product high requires a lot more effort.

I basically quit over a number of issues, but the trigger was literally burnout from too many parallel projects.

My ifi zen stream could definitely use some of that - it’s a mess and very unreliable to use.

I suggested a similar product but based on a commercial streaming platform that would place the software development/maintenance burden on an outside company that ONLY does streaming hardware/software (and had more people in software only than iFi had headoffice + R&D staff) and add the hardware parts that make iFi unique.

It seems after I left someone decided to do the software for this in house was a good idea. Software done right is a lot harder than hardware done right (ask Toyota about their "acceleration when the car thinks you need it" feature).

The best you can do is complain loudly and hope the software get's fixed eventually.

Thor

 

[GoldenSound]

There are few interesting things to note. Delta Sigma and Multi-Bit are very different and have what I'd call "complementary flaws".

A Multibit DAC has very good resolution at high levels, the closer we get to zero the worse it gets. On the other hand, delta-sigma converters (as in pure single bit) are best at low levels and degrade the closer levels get towards full scale.

For clarification I'm meaning delta sigma as in any DAC that requires/uses delta sigma modulation. Regardless of to what bit depth it modulates to. Personally I view things as somewhat binary. It's either a native PCM converter or it isn't. Though there are of course a myriad of different approaches once DS modulation is included including the various multibit switched-resistor, thermometer coded or other topologies etc.
So I'm including most modern multibit architectures from AKM, ESS, BB etc. In fact I'm not actually sure if there are any currently available pure 1-bit solutions aside from the discrete implementations from some manufacturers? (I'd be curious to know if that's not the case though! Would be pretty interested in trying one.)

A Multibit DAC has very good resolution at high levels, the closer we get to zero the worse it gets. On the other hand, delta-sigma converters (as in pure single bit) are best at low levels and degrade the closer levels get towards full scale.

Technically this would depend on the modulator performance though right? At least in the sense that whilst there are challenges with 1-bit converters outputting close to full scale, they also are limited in their small signal amplitude/phase accuracy by a combination of the speed they're running at and the noise shaper/modulator performance. (Though of course usually more accurate at low levels than what could usually be done with a native PCM approach/R2R ladder, at least until some recent developments)

Chord DACs are an interesting one there as the additional compute power available and the extremely high speed of operation allowed @Rob Watts to implement a modulator with over 300dB dynamic range.
In fact purely for academic interest, with massive modern computing power tools like HQPlayer allow for pure 1-bit DSD modulators that can exceed 350dB dynamic range. Here's an FFT of DSD512 output from HQPlayer's ASDM7EC-Super 512+fs modulator for instance (with a few PCM filter/input options compared):

The ADI-2 also uses oversampling DAC IC's. They just have different digital processing and analogue side, that in effect allow them to match the NOS waveform more closely whereas the BB Chip's evidence some ultrasonic noise.

Yeah my point there was that the ADI-2 is oversampling, and in fact in the manual they detail that filter a bit more. They list it as 'NOS' in the menu but more correctly describe it as a 'super slow' filter in the manual. It's an example of how one can get a 'NOS-like' result without needing to be NOS.

There are some other DACs which have a 'NOS' option but provide quite a different result. Gustard on a few of its products has the option to bypass ESS's initial FIR filter, but because of the way in which the later stages of the chip work, the output IR doesn't look like NOS:

(Gustard X26 Pro 'NOS' IR)

NOS, that Non Over-Sampling is literally a marketing term. In technical terms using "Digital Filterless" is more accurate.

The next question is what actually produces the characteristic "NOS Sound", in other words which part of the objectively poor performance is responsible for the sound characteristic.

That's the tricky bit, and where some of the debate around what 'counts as NOS' stems from.
For those simply seeking a 'NOS sound', I think you can get most of that using an oversampling zero-order-hold filter regardless of the DAC itself. As it'll create the same effects as a NOS DAC for the most part. The only parts it won't replicate being differences due to the DS modulator and the DAC architecture itself. One could also argue that 'NOS Sound' is contributed to by some of the nonlinearities of older R2R DACs in particular as well. Though some new R2R DACs actually exceed most delta sigma DACs in dynamic range and distortion performance. Those ones have to take approaches that aren't exactly 'affordable' of course though.

Personally though, I rarely ever actually listen in NOS. My interest in a true NOS DAC is because I want to be able to take full advantage of the highest performance PCM reconstruction and noise shaping tools available, which means the DAC ideally needs to actually NOT be running its own noise shaper on top even if it's not technically 'filtering'.
With AKM DACs for instance I always got better results using a higher performance DSD modulator +1-bit output vs using the DACs internal multibit modulator/converter.

 

[Thorsten Loesch]

For clarification I'm meaning delta sigma as in any DAC that requires/uses delta sigma modulation.

I think we are getting into a OT veer with minutiae that are highly intriguing to us two but make almost everyone else glaze over.

99% of the people just want to know how it sounds and not why.

Maybe we chat over a beer or two if you ever visit Thailand?

Thor

 

[Terr0rSandmannLuxx]

Just arrived today. Lets see how they compare ... I will test the 2 Diablos with an Expanse and Tia Trio.

 

[tramubla]

Just arrived today. Lets see how they compare ... I will test the 2 Diablos with an Expanse and Tia Trio.

How is the build quality of that volume lock slider in your opinion? Does rattle at all when disengaged and does it hold firmly when engaged?

 

[McCol]

How is the build quality of that volume lock slider in your opinion? Does rattle at all when disengaged and does it hold firmly when engaged?

I've been using the Diablo 2 for a few days and have found no issues with the volume lock

There is no rattle and it holds very firmly.

 

[iFi audio]

Just arrived today. Lets see how they compare ... I will test the 2 Diablos with an Expanse and Tia Trio.

Looking forward to your thoughts and impressions of the Diablo 2!

Cheers!!

 

[iFi audio]

How is the build quality of that volume lock slider in your opinion? Does rattle at all when disengaged and does it hold firmly when engaged?

I also have no issue with the Volume Lock, it is nice and secure and holds very well.

Cheers!!

 

[boxster233]

Finally demoed this @Audio46 . Clean, powerful sound. This works as both a large portable and a desktop solution. Really phenomenal sound stage with Final D8000 and detail with Dan Clark Expanse.

 

[BS5711]

Finally demoed this @Audio46 . Clean, powerful sound. This works as both a large portable and a desktop solution. Really phenomenal sound stage with Final D8000 and detail with Dan Clark Expanse.

Does it still have the battery current limit problem of the original Diablo when pushed hard into a low impedance low sensitivity load like the Expanse ?

That is the unit will cut out when the current limit protection is triggered.

 

[Terr0rSandmannLuxx]

Does it still have the battery current limit problem of the original Diablo when pushed hard into a low impedance low sensitivity load like the Expanse ?

That is the unit will cut out when the current limit protection is triggered.

Hm. I have both here, the Diablo and Diablo 2 as well as the Expanse and I can not reproduce the issue you shortly described. Or maybe it kicks in far beyond normal hearing level.
But what I can say is that I can not hear very loughtly with the Expanse, because both Diablos are getting shrill and exhausting. Which is usually the case when the amp does not have enough power. I assume that this is a sign that both Diablos does not deliver the ~5W but far less. I can set my D90SE + ifi Pro iCan Sig to the same db volume (Measured with a sound pressure meter) and the same song sounds much more pleasant on the desktop stack. Sure sure, apple vs pears, but just saying.
I had the same hearing experience with the Fiio Q7, which should have 3W on paper. But the Q7 was a bit less engaging compared to the Diablos. With low and medium hearing level, I like the Diablos more compared to the Q7 from Fiio.

Since I owned every ifi micro iDSD device in the past years (exept the Diablo X, which was an absolutely unnecessary product?), I had the hope that the Diablo 2 is a step up in sound quality and power deliverey. But unfortunatly this is not the case. It sounds the same or almost (maybe a bit more resolving) the same as the original Diablo. Is this bad? I do not think so, I really like the Diablo and for me it is still one of the best (or maybe the best?) portable DAC/Amp device out there. And the Diablo 2 is a smart improvement with features I was missing a lot on the original Diablo. I like the Bluetooth option, the integrated IEmatch (I used an external ifi 4.4 IEmatch with the original Diablo when using IEMs), volumen knob lock, USB-C for data and charging and the desktop mode. It is not mentioned by ifi as far as I know, but ifi improved the annoying channel imbalance with the volumen knob. I do not say it is completly gone, but much better and no problem in daily usage anymore. At the end it is a clever further development, but not the jack-of-all-trades (I used Google translate here, in Germany we call it "Eierlegende Wollmilchsau") that many fans were waiting for. Is the Diablo 2 worth the extra 300€? Due to all the additional improvements and added features, I would say yes. But if we are honest, both Diablos are just to expensive for what they are. I would see the original Diablo at around 749€-799€ and the Diablo 2 at around 899€-949€.

PS: ifi, I still miss the Xbass, even if I do not need it for my Expanse and Tia Trio But your Xbass is a killer feature in many of your products and the community like the feature a lot