iDAC 2 - Quad DSD256 / PCM 384Khz / USB DAC & headphone amp

[AMR/iFi audio]

The iDAC2 – spilling the secret sauce (part 1)

 

 

 

The iDAC2’s musicality has wooed – most recently the iDAC2 + iPurifier2 + Pro iCAN at the Fujiya AVIC Headphone Festival in Tokyo where the market is probably the most demanding on the planet.

 

 

 

 

Source: iFi Retro Stereo 50 + iPurifier2 at Fujiya Avic 2015 | DAR__KO

 

The Pro iCAN was one of the show ‘best sound winners’ so the small but mighty iDAC2 + iPurifier 2 combo deserve some credit for producing the digital-to-analogue conversion to feed the Pro iCAN.

 

 

At the show, the quite technically proficient Japanese press asked us on more than one occasion why the recently-launched iDAC2 sounded so nice. And this is a press pack that has turntables in heavy rotation so they know their onions. This gave us the idea to put finger to keyboard.

 

Source: 【候補多すぎ】ヘッドフォンç¥*アワード2015秋は週末までに公開致します！ | 目福耳福 Fujiya Avic BLOG

 

 

Bake-Off: the iFi recipe behind a really good DAC

 

The following are our thoughts only. The application of the parts budget as common industry practice is similar to what we did in the iFi iDSD nano which is more indicative of <US$1,000 DACs. Often a very large part of the budget is needed for the digital platform (after all, we need a DAC Chip that headlines and a good USB processor). Power supplies and analogue stages receive the remaining budget.

 

Hence, the three key cost areas in the electronics of a digital-to-analogue converter:

 

 

1. Digital section cost - is much the same across the board, chipsets/clocks vary but slightly and software is usually ‘off the shelf’ like the XMOS firmware and off the shelf DAC Chips of usually comparable cost. In a USB DAC the USB Processor is usually the biggest ticket item before the DAC. Clocks and their power supplies often cost as much as a DAC Chip.
   
2. Analogue section cost – from a simple double Op-Amp for 20 cent to things done much more extensively cost varies. The iDAC2 has BB Soundplus integrated amplifier and a discrete JFet and BJT Class A output stage. Further it employs C0G Capacitors and MELF Resistors for all signal positions, all this adds cost.
   
3. Power supply section cost – Often USB DACs have minimal power supply arrangements with generic 3-pin regulators. If done more extensively, then the cost is higher eg iDAC2 uses ELNA Silmic Capacitors and Active Noise Cancellation® to eliminate the USB power noise.
   

 

iDAC2 is no piece of cake

 

Things are different in the iDAC 2: we started with a larger budget, we could have spent it in a number of ways, like more DAC Chips or different ‘fashionable’ DAC Chips or fancy clocks. Instead we chose to put the extra budget where it impacts most, namely analogue stages and power supplies.

 

The iDAC2 has a similar cost digital section as with other DACs out there (and to our iDSD nano) but its analogue and power sections are more extensive hence its overall cost is greater so its pie chart area is larger.

 

 

 

 

 

 

 

The proof is in the listening.

 

Next time: Part 2. The Digital section

[AMR/iFi audio]

iDAC2: Spilling the Secret Sauce (2/6)

 

 

 

Digital section: ‘True Native’ augmented by latest gen Octa-Core Transputer

 

At the heart of AMR/iFI products including the iDAC2 micro is the same Burr-Brown ‘True Native’ chipset which we have explained before in the micro iDSD thread. This particular Burr-Brown chip offers two separate signal pathways for PCM and DSD. What this means is that one chip offers the ‘best of both formats’ as the signal quality remains native.

 

 

Choosing the right DAC topology significantly effects the final sound. We loved the dynamics and slam of the multi-bit topology (e.g. the legendary Philips TDA1541A), however when a High-Definition signal is used, the Multibit topology (actually no true multi-bit DAC chipset available yet) doesn't have the low-level linearity of the Delta-Sigma topology.

 

 

So to get the best of both worlds, we need:

 

• Multi-bit for dynamics and slam (the higher bits of the PCM data);
  
• Delta-Sigma for the low level linearity (the lower bits of the PCM data);
  

 

 

So the DAC chip we picked for the micro iDSD has the following topology:

 

• Top 6 bit: true multi-bit;
  
• Lower bits: Delta-Sigma.
  

 

 

 

 

 

 

 

 

 

 

 

 

As a result, the iDAC2 micro also supports DXD and DSD up to DSD256^#. Three Digital Filters are included; Bit-Perfect (Non-Oversampling), Minimum Phase and Standard and three selectable analogue filters for DSD.

^# DSD256 is available on Windows via ASIO DSD and with special firmware on OSX, Windows and Linux via DoP DSD

 

 

Running alongside the Burr-Brown ‘True Native’ chipset is the 8-Core XMOS. But with one essential difference – our own firmware. Such update to the XMOS Design and Firmware introduced by iFi include Star Clocking as first outlined in the iDSD micro.

 

 

 

 

 

 

 

 

 

 

 

 

The IDAC 2 implements Version 4 AMR XMOS Platform and uses the latest generation 8-Core 500MIPS XMOS1 transputer derived main processor. These processors are quite unique in their architecture and are based on a technology that once was considered to have revolutionized computing, the INMOS Transputer: which allows extremely high sample rates, supporting 384kHz/32Bit PCM and 11.2MHz single bit (DSD).

 

 

¹INMOS transputing to XMOS

 

 

The ‘Transputer’ (Trans – Com – Puter) was a pioneering microprocessor architecture of the 1980’s, intended for parallel (multi-core) computing. It was designed among others by David May and produced by Inmos, a semiconductor company based in Bristol, United Kingdom.

 

For some time in the late 1980’s many considered the Transputer to be the next great design for the future of computing. While INMOS themselves ultimately faded from the scene, their pioneering parallel computing platform is echoed in every modern PC running Dual or Quad Core CPU’s and in any Smartphone or Tablet featuring multi-core CPU’s.

 

XMOS was started among others by David May and modernised the Transputer core architecture. XMOS ‘Transputer’ Chips have since found many applications where their unique architecture outperforms both traditional CPU’s and FPGA systems, not the least in USB Audio.

 

While the digital section is very extensive and based on fairly unique items that stand out from the crowd, without a commensurate highly dedicated analogue section, the iDAC2 would not be half the DAC it is.

 

 

Next time – Secret to the ‘Analogueness’ (part 3)

 

 

 

[AMR/iFi audio]

By extension ... would that mean the iDAC Pro (or what ever its names is) will perhaps having 12% spent on the Digital section and 44% of the cost spent on Analogue and Power?

 

 

> By extension ... would that mean the iDAC Pro (or what ever its names is) will perhaps having 12% spent on the Digital section and

> 44% of the cost spent on Analogue and Power?

 

No, in the iDSD Pro we have massively beefed up the Digital section, starting with galvanic isolation and quad core DACs and quite a bit more. There are additional goodies in the digital stage - we shall disclose those once it is ready for general release.

 

The Pro is likely to have seriously impressive analogue stages too compared to the digital and power stages. For example, the 2 x GE5670s for the balanced circuit nicely add to the cost. There is also a fully discrete stage - as it is able to switch between Tube/Solid-State.

[AMR/iFi audio]

The iDAC2 – spilling the secret sauce (part 1)

 

 

 

/QUOTE]

 

 

 

Secret's out: made in The Nederlands for quite a while.

 

 

> Made in the Netherlands!

 

Yes, in the UK we are very multicultural. Anyway, a Bacon Butty without HP sauce is just totally out :-)

[AMR/iFi audio]

iDAC2: Spilling the Secret Sauce (3/6)

Analogue section: Secret to the ‘Analogness’

The analogue stage uses a Burr-Brown (by TI) Soundplus® J-Fet integrated amplifier with an added discrete Class A output buffer. The Class A buffer is a unique design combining J-Fets and bipolar transistors in a single-stage and allows the iDAC2 to handle even 600 Ohm Loads without breaking a sweat.

The ‘Soundplus®’ integrated amplifier selected for the iDAC2 easily goes up against the ‘ultimate’ audio integrated amplifiers from Burr-Brown (OPA627 and OPA827). Rather than relying on ‘fashion’ and selecting a ‘fashionable’ Op-Amp, we selected a specific Burr-Brown part because of its performance when combined with our unique Class A TubeState® output stage and in our mixed passive/active filter Analogue stage, it goes up right against the most expensive options (both in objective measurements and in listening).

We start from the datasheet specifications in the comparison table below itself which speaks for itself. Using industry benchmarks such as Input Noise and GBWP the BB Soundplus® compares well with both the BB OPA627B and OPA827….

¹ Input Noise, a measure how much noise the Op-Amp produces, lower is better

² Stands for Gain Bandwidth Product, a measure how ‘fast’ the Op-Amp is, higher is generally better

³ Stands for Total Harmonic Distortion & Noise, a measure how linear the Op-Amp is, lower is generally better

 

….BUT where we finish is by adding the extra discrete J-Fet + BJT Class A output stage to the iFi selected Burr-Brown, distortion performance is improved beyond that of the OPA627 and OPA827 from an already pretty good 0.00005% to a gobsmacking 0.000017%. This is another instance of where we go beyond the datasheet to wring a higher measured performance level.

The next question is, nice number but can one hear this in the real world?

Our answer would be a resounding ‘yes’ – we compared and we found audibly better dynamism with more difficult loads.

Next time, DirectDrive® (part 4)

[AMR/iFi audio]

iDAC2: Spilling the Secret Sauce (4/6)

DirectDrive^® look mom, no coupling capacitors

Commonly, coupling capacitors are employed to ensure the output from the DAC is free from DC (Direct Current). As the signal directly passes through coupling capacitors compared to other components in the system, these have an uncommonly large impact on sound quality. There is an after-market in coupling-capacitors. Take these rather special audiophile capacitors (pictured is the Duelund CAST Copperfoil Capacitor – yours for a cool 2,700 USD EACH). They are huge in size and expensive, sound every bit as good as their price yet a simple straight copper or silver wire invariably ‘impacts’ the signal less.

In the iDAC2 the analogue stage is direct-coupled (read: no coupling capacitors). In the iDAC2 a DC Servo is employed to ensure the output is always DC-free, however the DC Servo is implemented such as to make it effectively ‘invisible’ sonically-speaking, by using the same type integrated amplifier as for the actual signal path to perform the DC servo duties (so the DC servo is of the same level of quality) and the influence of the Servo on the signal becomes 20 times less than that of the main audio path through the circuit design and so completely disappears from measurements as well from the subjective sound quality.

In the iDAC2 we use MELF resistors as well as C0G filter capacitors (these are at least as good as polystyrene and approach Teflon capacitors for performance) for the low-pass filter. Additionally, the filter is a mixed-mode type where a passive filter first removes the unwanted very high frequency noise from the DAC output, which would be detrimental to the integrated amplifier performance before an active filter implements the final roll-off.

Dual-Mono Headphone section

The headphone amplifier is a non-trivial section either. It is an all-new dual-mono design with 350mW output into 16 Ohm and a maximum of 3.3V output available to drive high impedance headphones (> 100 Ohm). Using Direct Drive® technology it too is fully direct-coupled and MELF resistors are used for the gain setting network.

From above....

From below.....

With a well-respected pair of headphones such as the Fostex TL50-RPs, this reasonably-priced ‘plug ‘n play’ combination really shines. It even surprises one or two far more expensive setups.

Next time, Active Noise Cancellation® (part 5)

[AMR/iFi audio]

[h=1]iDAC2: Spilling the Secret Sauce (5/6)[/h]Active Noise Cancellation® gives >1,000X noise reduction

Another new feature used is the Active Noise Cancellation® power supply conditioning. First seen here in the iDAC2 micro and specially developed for USB-powered devices. They are not ‘Regulators’ in the classic sense of the word, but instead they only work on noise on the power supply voltage, without actually ‘regulating’ the voltage.

The traditional approach (see first graphic) using filters and regulators sees DC voltage lost AND noise is not comprehensively eliminated, there is still a residual level.

However, in comparison to the iFi ANC system, there is no DC voltage lost AND the power is noise-free:

We all know that USB Power from common USB ports is very noisy (usually tens if not 100's of millivolts of noise). So normally, power supply regulation is required to remove this noise. Common regulators use regulation elements in series with the powerline and require at least several 100mV to several Volt to work. In a USB-powered device this loss is not acceptable as we need at least 5V to give sensible levels for line outputs and headphones.

The all-new Active Noise Cancellation® avoids the loss of power supply voltage, while achieving similar or better rejection of power line noise as classic regulators. No extra series elements are needed. The Active Noise Cancellation® circuit cleans up the incoming USB power before it is distributed inside the iDAC2. A further such Active Noise Cancellation® circuit is cascaded after the first before supplying the headphone amplifier and analogue stage with power. A third Active Noise Cancellation® circuit is cascaded after this to give the DAC Chip the cleanest possible power source.

This multi-stage cascade achieves a reduction of USB noise by over 60dB (1,000 times) at low frequencies for the headphone amplifier and analogue stage. At higher frequencies where switching noise from switched mode power supplies is found, the noise attenuation is even greater.

At the DAC Chip the noise is reduced by over 90dB (31,600 times) for low frequencies. So even very high noise levels of several 100mV are suppressed below the self-noise of the analogue circuitry and DAC Chip, giving the iDAC 2 micro a Signal/Noise ratio of > 114dB(A) or an equivalent number of bits of > 19 Bits. Many DACs with an extra ‘zero’ on the end of the retail price do not go this far.

Next time, Power Supply – with great power come great responsibility (final, part 6)

[AMR/iFi audio]

iDAC2: Spilling the Secret Sauce (6/6)

Power Supply – with great Power brings great responsibility

 

Almost 940uF power supply capacitance directly supports the DAC Chip analogue power supply and a further almost 470uF dedicated to the digital supplies. The digital supplies incorporate three low noise series regulators for the 3.3V supply voltages of different parts of the digital section (including one dedicated for the Audio Clocks only).

 

To maximise dynamic performance, especially with bass, the analogue stage is backed by audio-grade ELNA Silmic Capacitors dedicated only to the analogue stage and additional capacitors giving a total of 1,400uF, while the Headphone Amplifier uses by almost 2,000uF in local power supply capacitance, located within a few mm of the audio circuitry supplied.

 

 

All in all, of the iDAC 2 micro printed circuit board, nearly 50% is taken up with power supply circuitry while the rest is used for all other functions. So whilst there should be discussion over the DAC chipset that surrounds the digital section, we always ask our readers to not overlook the analogue and power supply sections. These two are less 'sexy' but they are both just as important to the total sound.

 

By using such extreme and elaborate power supply circuitry, the iDAC2 micro achieves a new level of dynamic performance in USB powered DACs and a USB powered performance (objective & subjective) that previously required either battery or external mains power. This also results in why the headphone section sounds really impressive.

 

As we always say, the proof is in the pudding.

 

We hope you found these tech notes of interest. Thank you for reading.

 

[AMR/iFi audio]

On 2.12.2017 at 9:39 AM, Em2016 said:

Hello again young chaps @AMR/iFi audio

 

I completely understand that PCM stays PCM and DSD remains DSD with your True Native playback with this DAC.

 

One question:  when I look at the attached, are all incoming PCM rates up-sampled internally by the DAC up to 705/768kHz? 

 

So the format stays native, but the sample rates still undergo over-sampling/up-sampling internally? 

 

i.e. all PCM gets oversampled up to 705/768k internally and all DSD gets up-sampled up to DSD256 rate?

 

The reason I ask about the DAC internally over-sampling up to 705/768khz, is because this number appears in the attached.

 

Cheers !

 

 

 

In iDAC 2 micro's 'bitperfect' mode all digital filters are bypassed and the DAC operates at the original sample rate.

If any filter is selected, the input signal is upsampled to either 352.8/384kHz or 705.6/768kHz.

[AMR/iFi audio]

15 hours ago, Em2016 said:

 

Thanks iFi! Which filters or inputs result in internal upsampling to 705/768k specifically ?

 

 

Any sample rate from 88.2kHz to 192kHz is converted to 705.6/768kHz IF the minimum or standard filter is selected.

 

If the bitperfect filter option is selected the DAC is fed directly to the actual DAC part of the chip and converted without any upsampling and digital filtering and without repeating samples or such trickery. 

 

It operates equivalent to what has been termed 'Non-Os, Non-Oversampling, Zero-Oversampling' etc. We prefer the term bit-perfect because that describes the operational principle better.

 

15 hours ago, Em2016 said:

 

I thought ALL PCM inputs are up-sampled to 705/768k internally, when the non bitperfect filters are used - as per the block diagram from the tech notes I attached 

 

 

The diagram is a little simplified, the digital filter inside the DAC is 8x oversampling and can accept up to 192kHz audio. 

 

15 hours ago, Em2016 said:

 

Also, when DXD rates are fed to the iDAC2, do they remain DXD rates or they too are up-sampled internally to 705/768k?

 

 

DXD, DXD X 2 (and DSD) always remain DXD and are converted directly, without digital filtering, over/up-sampling etc.

[AMR/iFi audio]

On 5.12.2017 at 9:59 PM, Em2016 said:

 

One follow-up question Young Chaps, since I also have a nano iDSD and micro iDSD Black Label.

 

Does the part I quoted also apply to BOTH the nano and micro BL?

 

Can you clarify for each, for completeness.

 

Many thanks in advance

 

 

 

The internal up-sampling in all past and present iFi products with the DSD1793 DAC Chip and non-MQA Firmware (that is iDSD nano, iDSD nano LE, iONE Nano,  iDSD nano Black Label flashed with 5.2X, iDSD micro, iDSD micro Black Label, iDAC 2 micro, Retro STereo 50) operates as follows:

 

Bitperfect filter selected (all micro products only):

• 44.1kHz ... 384/768kHz -> No ASRC, no upsampling, no digital filter
• DSD64 ... DSD256/512 -> No ASRC, no upsampling, no digital filter

 
Standard/Measure Filter selected: 

• 44.1kHz ... 48kHz -> No ASRC, synchronous oversampling to 352.8/384kHz using long FIR (Transient Aligned) digital filter 
• 88.2kHz ... 192kHz -> No ASRC, synchronous oversampling to 705.6/768kHz using long FIR (Transient Aligned) digital filter 
• 352.8 ... 384/768kHz -> No ASRC, No upsampling, no digital filter
• DSD64 ... DSD256/512 -> No ASRC, No upsampling, no digital filter

 

Minimum/Listen Filter selected: 

• 44.1kHz .. .48kHz -> No ASRC, synchronous oversampling to 352.8/384kHz using ultra short FIR (minimum ringing Bezier) digital filter 
• 88.2kHz ... 192kHz -> No ASRC, synchronous oversampling to 705.6/768kHz using ultra short FIR (minimum ringing Bezier) digital filter 
• 352.8 ... 384/768kHz -> No ASRC, no upsampling, no digital filter
• DSD64 ... DSD256/512 -> No ASRC, no upsampling, no digital filter

[AMR/iFi audio]

22 hours ago, Em2016 said:

 

Brilliant summary. Greatly appreciated.

 

 

Yup, you're welcome.

[AMR/iFi audio]

2 hours ago, buonassi said:

A little late to the game, but does the iDAC2 support Audirvana+ "Direct Mode"?  

 

This is USB Audio Class 2.0 compliant deivce and as such it should work. But this software related matter than anything else, hence if you want to be in 100% sure, please ask Audirvana about this.

[AMR/iFi audio]

18 hours ago, Les Habitants said:

 

I own this combo and use it in my bedroom system, you won't regret this purchase, it punches WELL above it's weight class.

 

Very nice to hear that! Enjoy!

[AMR/iFi audio]

12 hours ago, buonassi said:

I read somewhere, can't remember where, that appliances like the schiit eitr or the iUSB3.0 have much more "apparent" effect on SQ than USB cables, and cables would only be the "cherry on top" if you will.  

 

In general, more people who tried cables and USB related tweaks would agree with what you wrote.

[AMR/iFi audio]

On 1.01.2018 at 4:07 AM, Les Habitants said:

 

I own this combo and use it in my bedroom system, you won't regret this purchase, it punches WELL above it's weight class.

 

We believe so as well

[AMR/iFi audio]

We'll leave it here... 

 

The Sound & Vision Bristol Show 2018 is in the past now. Things happened there, oh yes they did!

 
 
Yup, that's our full Pro rig on the Pro iRack.

 

 

 

Some people were impressed more than just 'mildly'....

 

And we had someone (aka The Cap'n) in there to answer the most technical questions there are.