Ifi Neo iDSD2 Review (2) – Does A Lot And Does It Right
I recently received a Neo iDSD2 unit from Ifi, with the purpose of assessing it and developing a second opinion following the one previously published by Kazi, here.
Neo iDSD2 is a modest-footprint desktop DAC-AMP unit with a very articulated features set, such that usage flexibility, even before than output quality, is no doubt one of its reasons-to-be. It retails for 899,00 GBP in UK, or for 948,00 € on the various EU Amazon regional portals.
In this Article
At-a-glance Card
| PROs | CONs |
| Spectacular amplification quality | Lack of balanced analog input |
| Clean, detailed yet musical tonality | Not for top-challenging planars |
| Support for analog input (too) | Some range compression on Nitro gain |
| Significant output power | A higher tier power supply would be appropriate |
| Well implemented Turbo gain | |
| Advanced BT 5.4 implementation | |
| XBass II | |
| XSpace | |
| IEMatch | |
| External clock sync support |
I am not writing these articles to help manufacturers promote their products, even less I’m expecting or even accepting compensation when I do. I’m writing exclusively to share my pleasure – and sometimes my disappointment – about gear that I happen to buy, borrow or somehow receive for audition.
Another crucial fact to note is that I have very sided and circumscribed musical tastes: I almost exclusively listen to jazz, and even more particularly to the strains of post bop, modal, hard bop and avantgarde which developed from the late ’50ies to the late ’70ies. In audio-related terms this implies that I mostly listen to musical situations featuring small or even very small groups playing acoustic instruments, on not big stages.
One of the first direct consequences of the above is that you should not expect me to provide broad information about how a certain product fares with many different musical genres. Oppositely, you should always keep in mind that – different gear treating digital and analog sound in different ways – my evaluations may not, in full or in part, be applicable to your preferred musical genre.
Another consequence is that I build my digital library by painstakingly cherry-pick editions offering the least possible compression and pumped loudness, and the most extended dynamic range. This alone, by the way, makes common music streaming services pretty much useless for me, as they offer almost exclusively the polar opposite. And, again by the way, quite a few of the editions in my library are monoaural.
Additionally: my library includes a significant number of unedited, very high sample rate re-digitisations of vinyl or open-reel tape editions, either dating back to the original day or more recently reissued under specialised labels e.g. Blue Note Tone Poet, Music Matters, Esoteric Jp, Analogue Productions, Impulse! Originals, and such. Oppositely, I could ever find an extremely small number of audible (for my preferences) SACD editions.
For my direct and conclusive experience, virtually none of the above is available on the most common streaming platforms, which makes their offering of high sampling material apriori pointless for my purposes. Quite simply, then, I don’t use streaming services as I find them not up to my requested quality.
My source gear is on its turn selected to grant very extended bandwidth, high reconstruction proweness, uncolored amping. Which in particularly means that TWS and any other form of lossless-based wireless gear is apriori off my selection panel.
And finally, my preferred drivers (ear or headphones) are first and foremost supposed to feature solid note-body timbre, and an as magically centered compromise between fine detail, articulated texturing and microdynamics as their designers can possibly achieve.
In terms of presentation, for IEMs I prefer one in the shape of a DF curve, with some very moderate extra pushup in the midbass. Extra sub-bass enhancement is totally optional, and solely welcome if seriously well controlled. Last octave treble is also welcome from whomever is really able to turn that into further spatial drawing upgrade, all others please abstain.
Structural description
Externals & Internals
I’m not going to spend words on mere product description, which you can find on Ifi’s official pages (here), and even better you can read on Kazi’s aforementioned piece, here.
Input
Neo iDSD2 accepts digital input via 4 different channels:
- USB
- COAX S/PDIF (i.e. copper digital cabling)
- Toslink S/PDIF (i.e. optical digital cabling)
- Bluetooth
Based on a QCC618X chip, BT’s implementation is particularly significant. Supporting protocol level 5.4, it offers aptX Lossless, aptX Adaptive, Sony LDAC and LDACTM, HWA LHDC, regular aptX, AAC and SBC codec support.
It’s probably worth making a couple of en-passant comments here.
One: the most recent/advanced of such codecs (aptX Lossless, LDACTM…) are yet currently supported by just a few source devices. For many, their presence on Neo iDSD2 is more like a future-proof thing than an immediately usable feat.
Two: LDACTM is an LDAC variant focusing on connection stability at the cost of un-avoidable automatic resolution modulation. From the audiophile standpoint, then, LDAC is to be preferred whenever possible.
On a less technical, more practical level Neo iDSD2’s BT implementation is so solid and complete that those who are more accustomed than me to initiating audio streaming from their various wireless devices will surely perceived a big value from it.
The wired USB channel supports up to 32bit/768KHz PCM, and up to DSD512, and single and double speed DXD. Impressive. Lastly, and a bit more ordinarily, copper S/PDIF (COAX) input supports up to 192KHz, and optical (TOSLINK) will reach 96KHz.
Neo iDSD2 is equipped with a high quality internal clock (indeed, a very high spec clock for a device in this price category). Additionally, however, it supports connectivity to an external clock source via a 10MHz External Sync Clock connector.
Again, on the practical level: this means that those who already invested into a high quality standalone clock system can add the Neo iDSD2 its system, furtherly improving timing coherence (e.g.) between their CD transport and the Neo.
In addition to digital input, Neo iDSD2 also supports analog input – which means it can be used as a pure amplifier, when connected to an upstream DAC of sort.
Sadly, the sole analog channel available is a single-ended one, in form of a 3.5mm jack port – so a “full balanced” line cannot be put in place between a DAC supporting balanced output and the Neo iDSD2. Too bad.
Output options
Neo iDSD features both Phono outputs (on the front panel) and Line outputs (on the back panel).
In both cases, both a single ended and a balanced options are implemented.
Line Out ports follow 2xRCA (single ended) and 2xXLR (balanced) standards, Phono Out ports support 6.25 (single ended) and 4.4mm (balanced) jack connectors.
Notable features
Neo iDSD2 carries many of those features Ifi users are acquainted to, and rightfully love.
Reconstruction filters
First and foremost, Neo iDSD2 offers 4 easy-selectable reconstruction filters:
- Minimum Phase, slow roll off
- Standard, modest filtering
- Bit Perfect, i.e. no filtering
- Gibbs Transient-Optimised, aka “GTO”, offering minimum filtering and automatic 352,8/384KHz upsampling
I already covered Ifi’s particular implementation of the Gibbs filter within my Nano iDSD BL and Micro iDSD Signature reviews, here and here. You may also want to check Ifi’s own whitepaper, here. Or, my full article about reconstruction filters, here.
The four filters are all available within the active firmware, you can switch-on the one you prefer by simply selecting it either via the main front panel knob, or via the remote control. Which one is best to use depends on the digital material being processed, and on the user preferences.
Gain levels
Neo iDSD2 offers 4 different GAIN levels:
- IEMatch : -16db
- Normal: 0dB
- Turbo: +8dB
- Nitro: +16dB
Again, these are easily selectable either from the main panel, or via the remote control.
The Nitro setting (+16dB) offers a whopping power boost, at the cost however of a perceivable dynamic range compression. As some of my few readers may remember I’m particularly sensible to wide, relaxed sonic ranges so I recommend resorting to Nitro only in really exceptional cases.
Turbo gain (+8dB) is wonderfully implemented: I can hardly appreciate any compression when switching it on, so for my tastes it’s no doubt the go-to “High Gain Option”, when needed, on Neo iDSD2.
Finally, the IEMatch option (-16dB) can prove nothing short of invaluable when pairing Neo iDSD2 with ultrasensitive and/or ultra low-impedance drivers. This, both to avoid hissing and to win back a wider volume / smoother volume control.
The option is not labelled “IEMatch” by chance: it’s in facts an IEMatch circuit built-into the Neo iDSD2. The “IEMatch Gain” option not only applies the mentioned negative gain to the signal, but it also “shadows” impedance as follows:
- It presents connected drivers with an Output Impedance of 7 ohm
- It makes the amp section perceive a Load Impedance between 24 and 33 ohm, depending on the actual load impedance
For those not familiar with IEMatch, please refer to my dedicated piece about it, here. You can also read about how it is implemented inside other IFI models here, here and here.
XSpace and XBass
As you may already know, both are particular filters offering interesting sound “refinements”, very nice to apply in quite frequently encountered situations.
XSpace is a complex filter resulting, as the name hints, in a sensation of improved sound spatiality, which can be very precious especially in conjunction with acoustic music.
The filter mainly operates in terms of added crossfeed (partial blending of left and right channels) to simulate room speakers output, with the addition of some touch-up on the trebles, to add air.
Talking about XBass now, it’s firstly important to note that on Neo iDSD2 the filter has a twofold implementation, which is probably why they labelled “XBass II” this time.
You can activate:
- XBass Bass (blue X) : a filter enhancing the sub-bass and some bass frequences
- Xbass Presence (green X) : this one pushes the high mids more forward
- Xbass Bass+Presence (red X) : this option activates both filters at once
These are very, very useful to help with drivers and/or tracks needing a stronger punch down low, or showing a bit depressed high mids, or both.
For those like me familiar with the Mojo 2, XBass II is a sort of a skinned-down implementation Mojo 2’s DSP system.
Lastly, but importantly, XSpace and XBass II are not digital filters, but are entirely implemented in the analogue domain, i.e., they are realised by means of electrical ciruitry working on the output signal path, after the DAC has made its job.
Who cares? You should, as such approach, by definition, avoids artifacts such as aliasing, or rounding errors, produces smooth, continuous effects, does not impact on resolution, does not require any digital down- / re-sampling and does not introduce laterncy. In a word: it’s just better.
And in facts they do sound wonderfully well, and are really useful in so many occasions. Very, very good nice-to-haves.
Power supply
Neo iDSD2 comes bundled with an IFI iPower 2 unit, which is an above-decent entry level switching PS – as a matter of fact the base model in Ifi’s power supplies offering.
So this is good, in a sense, but also less than ideal. For a device like the Neo iDSD2 a higher quality PS is very significant to its output signal quality, and that’s why considering Neo iDSD2’s price tag, as a user I would definitely be more pleased to find an iPower-X inside the box, really: its improvement vs the lower cost sibling is not small, all for just a very modest retail price difference…
MQA Rendering
Does anybody care about MQA anymore? I never did of course, and I will spare the “…told ya” part here.
Suffice it to say that Neo iDSD2 is a “MQA Full Decoder”.
Singers/players/bands/publishers record their tracks, and eventually release their albums. Prior to the digital music distribution era, there could be very little doubt about whether the music we were listening to was the “original” version of that album as its creator/publisher intended or not; if we had a legit copy of that LP or of that CD, that was it.
In the digital music distribution system, instead, the end user has no “solid” way to make absolutely sure that he’s receiving an unaltered version of those tracks. For what he knows, he might be getting a subsequently remastered, equalised, anyhow manipulated version of that album.
The MQA offers a way to “certify” this. An “MQA Studio” track is a file which containes some sort of “certification codes” that guarantee that track is indeed “the original” as released by the authors. A sort of digital signature, if you wish. Anyone might process, EQ, remaster, etc, that track, and re-encode it under MQA but the new file wouldn’t carry the original author signature anymore.
“MQA Original Sample Rate” (a.k.a. “MQB”) tracks are MQA Studio Tracks for which a further certification is given that not even the mere sample rate has been altered (in particular: oversampled) compared to the “original version” as released by the authors.
Any MQA-capable device can play back all MQA encoded tracks, but only MQA Full Decoders are able to identify such additional “digital signatures” and tell the user “hey, this is an original track” or not.
Ifi GO Bar, Gryphon, HipDac-2 and Neo iDSD2 are all Full Decoder devices. Ifi GO Link, HipDac, Micro iDSD Signature, Nano iDSD Black Label are Renderers.
Between parentheses: HipDac and HipDac-2 being virtually identical in terms of sound capabilities, power, etc, with the sole major difference represented by their different MQA capabilities, offered me the interesting opportunity to check the differences on a quite similar if not virtually identical situation and I could tell a quite obvious SQ improvement when listening to a few particular tracks just Rendered (HipDac) or Full Decoded (HipDac-2).
That said, I don’t personally care about MQA, nor about any of the existing digital distribution catalogues for that matter, due to the fundamental lack of good editions of the music I prefer on there.
Sound and power
In terms of tonality / timbre, I found Neo iDSD2 to carry clearer voicing compared to other Ifi models. Yes it still reveals a modest warmth but I think I can safely say that in terms of tonal tuning it’s a step aside from what many would call “Ifi house sound” – and, if you ask me, this is good news.
From the amplification standpoint, Neo iDSD 2 is no doubt powerful device, although possibly a bit less so than the marketing jargon might try to construe.
Official spec sheets report the following max (peak) values:
- Balanced 4.4mm: 3.5V / 19.5V max. (12Ω – 600Ω Headphone)
- UnBAL 6.3mm: 4.5V / 9.50V max. (12Ω – 300Ω Headphone)
And (more interestingly) the following RMS values:
- Balanced: >19.3V/620 mW (@ 600Ω); >12.7V/2530 mW (@ 64Ω)
- UnBAL: >9.7V/157 mW (@ 600Ω); >9.4V/1390 mW (@ 64Ω)
The latter (RMS values) are those which “count”, and they are …high. How much high?
My collection does not include far-edge-insensitive planars (80ohm 104dB/mW Focal Utopia, or Kazi’s HE-6SE V2, etc) so I’m not really interested, and more prosaically unable, to directly check if Neo iDSD2 can cope with those. I trust Kazi: it doesn’t. The good news, at least for me, is : I don’t care 🙂
My collection does however include “commonly found” source-picky units, and those are in the end the paragon stones I use when assessing sources, like I’m doing now.
HD800 (SDR Mod)
The Neo iDSD2 can agilely power my Sennheiser HD800 (300 Ω, 102 dB/V), for example, and it can do so on Turbo gain, i.e. without the need to resorting to that range-compressing Nitro I mentioned above.
While I have the HD800 on, let me add that XBass Bass (Blue X) and XSpace filters both pair marvelously with the decent digital sampling of Atlantic’s R1-512581 edition of Coltrane’s Giant Steps I have in my collection.
The crossfeed introduced by XSpace is just about right to bring Coltrane’s sax closer to the stage, back from that artificial far-off left side placement it has on the iconic “Mr P.C.” track, and most of all it sounds “natural”, “realistic” in doing so.
The XBass Bass (Blue X), in its turn, adds some pleasing, and again organic, realistic body to actual mr P.C.’s (Paul Chamber) superfast standup bass hand plucking.
Sure, I can (and normally do) obtain a very similar effect by actionating upon Roon’s MUSE DSP – offering both (parametric) Crossfeed and (even more) sophisticated PEQ options. Neo iDSD2 however just makes it all happen at a mere press of a button, regardless of the source, and it “just works right”. Commendable, indeed.
E5000
Another very significant pairing experiment is represented by final’s E5000.
As many know, these babies carry both quite low sensitivity (94dB/mW) and impedance (14 Ω), and (which is not apparent on their spec figures) an irregular impedance curve.
They are tuned using a combination of 5 or 6 internal dampening techniques, aiming to obtain a smoothly balanced tonality, at the cost however of a significant amping level need.
Such E5000 “nature” is so very uncommon on the market, possibly almost unique, which construes a sort of justification for the fact that so many, if not most sources are not even designed to care about delivering the high current E5000 needs on a load of such a low value as E5000.
Many are the sources (amps) that can indeed deliver such output power, and much more, but on higher impedance loads, not at 14 ohm or below. That’s ultimately why the E5000-test is so significant in many occasions. “My amp makes my planars sound fantastic, so if E5000 sound dark on it, it cannot be the amp’s fault” – wrong: indeed it can be and often is the case.
Neo iDSD2’s spec sheet “hints” us to be optimistic as it explicitly indicates a (max) 3.5V swing on 12 Ω load, that is, circa 1W (peak value) on such a low impedance region. Veeery uncommon, thus very promising.
And, indeed it works ! Even on Normal (0dB) gain, Neo iDSD2 literally dominates E5000.
Firstly, it “opens” E5000 tonality taking away that typical darkness (or worse) they offer when subject to insufficient amping. Furthermore, Neo iDSD2 reveals nothing short of fantastic separation and layering, and a damn rare-to-get (from E5000) clean visibility onto low volume voices and details.
In such result proweness, the Neo iDSD2 is on the same page with my current reference desktop gear – the Questyle CMA-400i – which can drive E5000 the same way on its Low Gain (0dB) setting.
Indeed, listening very carefully and critically I should say that the Neo iDSD2 is even ahead by a small bit when it comes to readable layering (as I can perceive “more air” between the layers) while the CMA-400i remains unbeaten as for microdynamics rendering. Still really, really enough said!
Incidentally, as soon as I switch to its Standard Gain (+10dB) position the CMA-400i has E5000 significantly loose detail and microdynamics subtleness, and “glues” some of the layers together.
When I switch Neo iDSD2 to Turbo Gain (+8dB) only some of the same happens, which is consistent with my above note regarding Neo iDSD2’s Turbo Gain not coming with too penalising compromises in terms of range compression (unlike what happens on the CMA-400i).
Lastly, I took some time to experimentally double check what should actually be quite obvious, that is, Neo iDSD2’s proweness in driving E5000 descends from the good deeds of its amplification stage.
To that aim I ran an AudioQuest interconnect (incidentally: a Big Sur 2xRCA-3.5 cable) from my CMA-400i’s single ended Line Outs to Neo iDSD2’s 3.5mm analog input, and tested the three options:
- Transport -> USB -> CMA-400i -> E5000
- Transport -> USB -> Neo iDSD2 -> E5000
- Transport -> USB -> CMA-400i -> Analog i/c -> Neo iDSD2 -> E5000
As expected, the third case is very similar to the previous 2. I get some of CMA-400i’s microdynamics, and some of Neo iDSD’s “air between layers”.
Andromeda 2020
Last school case for today is Campfire Andromeda 2020, featuring very low impedance (12.8 Ω) and ultra high sensitivity (123dB/mW). So easy to drive “loud”, yet so difficult to drive “right”.
I’ll make the story very short here, as this case corresponds exactly to Kazi’s article findings regarding another Campfire model, the Holocene.
Neo iDSD2’s “IEMatch Gain” perfectly matches the Andromeda challange: gain is lowered thereby making volume control smoother, the amp sees a “more comfortable” load impedance, no hiss comes off the drivers, all this without the beefier output impedance producing distortions, especially in the bass. Simply put: it just works!
Oddity in the specs?
Quick test: How many of you noticed an “oddity” – or at least an uncommon value – in the output power specs I reported above (and below, too)?
Let me make the case more apparent:
| Load / Max Power | Balanced output | Single Ended output |
| 12 Ω | 1021 mW | 1685 mW |
| 32 Ω | 5551 mW | 2832 mW |
| 600 Ω | 650 mW | 184 mW |
Can you see it now? Balanced Output power value on 32 Ω loads and more is much higher than Single Ended Output – that’s commonly the case comparing Balanced vs Single Ended, isn’t it? – but on a load of 12 Ω it’s the other way around !!
How uncommon…. But why ?
I asked ifi, and they clarified that this is a precise safety choice. Simply put, amplifiers dissipate a lot of heat when very low impedance loads are connected. And of course, such heat is even higher on Balanced output stages vs Single Ended ones. More heat = more potential problems with the circuitry. Excessive heat = guaranteed problems.
That’s why Neo iDSD2 is designed to impose a maximum power limit to Balanced output power when loads under a certain impedance value are connected, and such limit is higher when the same load is connected to the Single Ended output.
And let me add: this is the technical reason why – as I mentioned above – so many amps can power 30 Ω hard planars, but fail on a 14 Ω dynamic driver like E5000…
Official Specifications
| BLUETOOTH AND CODECS | Bluetooth 5.4 (aptX Lossless, aptX Adaptive, aptx, LDAC, HWA/LHDC, AAC and SBC) |
| WIRED INPUTS | USB3.0 B (USB2.0 compatible); S/PDIF (coaxial up to 192kHz, optical up to 96kHz) |
| FORMATS SUPPORTED | PCM 768kHz; DSD 512 (22.6MHz); MQA Full Decoder |
| DNR | Line Section >120dB(A) @ -60dBFS Headphone Section -120dB(A) |
| MAXIMUM OUTPUT POWER | Headphone Section Balanced: >19.5V/650 mW (@ 600Ω); >13.3V/5,551 mW (@ 32Ω); 3.5V/1,021mW (@ 12Ω) UnBAL: >10.5V/184 mW (@ 600Ω); >9.5V/2,832 mW (@ 32Ω); 4.5V/1,685mW (@ 12Ω) Line Section Outputs: Balanced XLR: 19.5V max. (variable) 4.4V fixed UnBAL RCA: 10.5V max. (variable) 2.2V fixed |
| RMS OUTPUT POWER | Headphone Section Balanced: >19.3V/620 mW (@ 600Ω); >12.7V/2530 mW (@ 64Ω) UnBAL: >9.7V/157 mW (@ 600Ω); >9.4V/1390 mW (@ 64Ω) |
| OUTPUT IMPEDANCE | Line Section XLR ≤100Ω; RCA ≤50Ω Headphone Section <1Ω |
| SNR | Line Section <-120dB(A) @ 0dBFS Headphone Section >120dB(A) (3.3V 6.3mm/6.2V 4.4mm) |
| THD+N | Line Section <0.0015% @ 0dBFS Headphone Section <0.0015% (125mW @ 32Ω) |
| DIMENSIONS | 214 x 158 x 41 mm (8.4″ x 6.2″ x 1.6″) |
| NET WEIGHT | 916g (2.0 Ibs) |
| POWER CONSUMPTION | No Signal ~5W; Max Signal ~13.5W |
Considerations & conclusions
I was seriously impressed by the Neo iDSD2.
It’s a very complete output system, accepting multiple digital inputs and also usable as a pure analog amplifier. Its reconstruction capabilities are ace, tonality tuning strikes a wonderful compromise between cleanness and musicality, and the amplification stage delivers equally spectacular quality on a huge range of diversely challenging drivers.
Its extras like Xbass II, Xspace and IEMatch just top the list, and complete the picture of a truly remarkable device, which in my opinion is worth every penny it costs.
Thanks once again to Ifi for the assessment opportunity.
Our generic standard disclaimer.
