iFi iDSD Diablo 2 Measurements

Content and analysis made possible thanks to support from https://headphones.com, and https://patreon.com/goldensound supporters.

This unit was loaned to me by iFi

Note: This post is solely to provide objective data and some of my thoughts on the design. My opinions on the subjective sound will be covered in the video review.

Design:

The iDSD Diablo 2 is the successor to iFi’s original Diablo, and the latest in their flagship portable DAC/Amp lineup. Previous units include the iDSD black label, iDSD signature and iDSD Diablo.

Externally, the unit is quite different from the previous version. The most obvious change being a new chassis design, with numerous ridges that facilitate the use of the included ‘wings’; two pairs of stands, one flat one angled, that can be slotted into the various ridges to have the unit sit at an angle of your choosing.

There are other changes however, gone is the male USB-A socket (thank god for that) and in its place two USB-C ports, one for charging and one for data. The device ships with a 5V iFi iPower PSU so you can leave it plugged in as a desktop DAC/Amp, or you can take it on the go and run it from the internal battery.

On the underside of the device is also an iematch switch, a feature that puts a resistor in series with the output, meaning the signal level of the device is attenuated, but so is the noise. This in theory allows you to use IEMs with a lower noise floor than having them connected directly to the output of the amp.

There is also now a bluetooth receiver inside, with a pairing button on the rear, and this also supports Aptx lossless! Meaning for redbook content, if your phone supports it, you can stream audio without any lossy compression directly to the Diablo.

Lastly the xMEMS support from the Diablo X has now been added. xMEMS devices require a DC offset, meaning unlike normal headphones/speakers that work with signals oscillating between positive and negative, they only work with positive voltages.

So if playing a normal 1khz sine for headphones it’d look like this, crossing at 0V and alternating between positive and negative voltages:

But then when enabling the xMEMS support, the signal remains the same, just offset, here we can see the sine is offset by about +3V:

Internally things seem to be fairly similar to the original Diablo, with some changes to accommodate the new USB C ports for example, but the actual core circuitry seems very similar overall. And as we’ll see shortly, performance/behaviour of the Diablo 2 is for the most part pretty much identical to the Diablo 1.

Measurements:

Test Setup:

– Audio Precision APx555 B-Series Analyzer with 200kOhm input impedance set unless otherwise specified
– USB Source: Intel PC via intona 7055-C isolator
– All measurements shown are with the DAC connected via USB unless otherwise specified
– Measurement setup and device under test are running on regulated 230V power from a Furman SPR-16-Ei
– Audioquest Mackenzie XLR and RCA interconnects
– Intona Reference Impedance Characterized USB Cable
– Audioquest Carbon SPDIF and AES cables (1.5m)
– Exact analyzer/filter configurations for each measurement are detailed in the full reports
– Low Gain used unless otherwise specified
– Measurements shown are from the headphone output with a 32 Ohm load connected unless otherwise specified

Full Measurement Reports:

Reports available here:

Dynamic Range (AES17): 104.9dB

SNR: 105.8dB

IMD SMPTE: -86.0dB

DC Offset: 1.9mV active, 1.9mV idle

Gain: 0dB (Low), 8.5dB (Med), 17dB (High)

Output Impedance: 0.35Ω (Standard), 5.75Ω (IEMatch Enabled)

1khz Sine 0dBfs, 4.4mm Line Out:

THD+N on the line out is near identical to the original Diablo (within 0.5dB), as is the distortion profile/structure itself.

1khz Sine 4V, 4.4mm Headphone Out (300 Ohm):

Performance generally is very similar on the headphone/front output vs using the line out. Ever so slightly cleaner on the line out so if you are using this solely as a DAC use that, but if needing preamp/volume control functionality the front output will work nicely too.

1khz Sine 4V, 4.4mm Headphone Out (32 Ohm):

Performance still holds up fairly well under load. Performance did not seem to change by any meaningful degree when running off the internal battery or when plugged in. Though it’s possible this may not be the case if the device is lower on battery at the time.

This is again within 0.5dB of what I measured the original Diablo, so it seems the analog circuitry on the Diablo 2 is not meaningfully different. If you liked the original, you’ll like this one too, but the upgrade is in the features, not the actual performance.

1khz Sine – 50mV output (IEM level), 4.4mm Headphone Out (12 Ohm):

83dB THD+N for IEMs means that this will work nicely without an audible noise floor on most IEMs. Additionally, channel matching on the volume pot is very good. This was a major issue on the original Diablo and seems to be much better here. Though this will vary from unit to unit, others may not be as close as this one so we will need to wait and see how people find the channel matching once more units are out in the wild. The same pot as the original seems to still be being used though so it’s possible this is just a bit of a ‘golden unit’.

1khz Sine – 50mV output (IEM level), 4.4mm Headphone Out (12 Ohm) – IEMatch enabled:

Using the IEMatch gives an approximately 7dB improvement in THD+N. This is great, though keep in mind that the increase in output impedance from 0.35Ω to 5.75Ω means that enabling this feature may alter the frequency response of some IEMs. Try with/without it to see what works best for your particular IEMs.
Additionally when using the IEMatch whilst the Diablo was plugged in seemed to pick up quite a significant amount of 50hz mains hum, completely negating any benefit and in fact reducing THD+N to 73dB:

I’d therefore suggest that if using IEMatch you keep the Diablo running solely off the internal battery.

Power:

This is something I wanted to discuss early in this post as it’s an important factor in why many people may be considering buying the Diablo 2, and critically:

The iDSD Diablo does NOT have 5W of power as iFi states on the product page and marketing material! It has 1.6W!

First, a bit of info on how amplifier power is measured. The industry standard method of providing a power spec is that unless otherwise explicitly stated, the power spec given is a continuous RMS spec. Meaning that it is the level at which the amplifier can output/run continuously without exceeding 1% THD.
So when you see an amp with ‘6W’ of power, it means it can run continuously outputting 6W, with distortion under 1%. Many amps may be able to output higher levels just with higher distortion (some may even provide a 10% THD power spec as well), and many amps also may be able to output higher than their continuous rating for short periods of time. For example an amp might be rated for 6W, but could supply 15W for very short bursts.

Here are the power specifications for a Zahl HM1 for example:

As you can see, the normal RMS power spec is given, but the peak power measurement is also provided, and is substantially higher. Most of the time amplifiers just give the RMS spec. Peak power is a useful measurement to have certainly, but if a manufacturer attempts to convey a peak measurement as a normal RMS measurement, the consumer will understandably of the assumption that the amp is more powerful in relation to competing products than it actually is.

Furthermore, there is an industry standard for this ‘peak power’ measurement, set out as per CEA-2006 which states that peak power should be measured as the maximum level an amp can output a 20-cycle 1khz sine (20 milliseconds) whilst staying under 1% THD. So the same as the continuous measurement for the most part, except the amp only needs to sustain that output for 20ms.

So what exactly is going on with the Diablo 2? Well, if we try to output 5W continuous into a 32 Ohm load, the device will simply shut off to protect itself. It CANNOT output this level continuously.
Adjusting to the maximum possible sustained output, we get about 1.6W RMS:

Adjusting any higher than this simply causes the device to clip:

This is NOT affected by the gain level, gain is just a measure of how much the input voltage is multiplied, and in current-limited situations like this it will not provide/unlock any extra power. High and Med gain on the Diablo 2 have exactly the same power limit (low gain is technically lower since it only goes up to just over 4V, meaning voltage is the limit not current).
Additionally, when running the Diablo 2 on the internal battery the maximum power was reduced from 1.6W to about 1.1W. This may be lower when the battery is not fully charged too.

This means that when tested the same way as any other amplifier, or if wanting to compare this to the specs given for other amplifiers, the correct power is 1.6W, NOT 5W!!

So what about ‘peak’ power? Well trying to do a 20-cycle 1khz sine as per CEA-2006 standard, at 5W we get this:

As you can see, after the first few cycles the amplifier begins to clip heavily. It cannot provide 5W peak power either…..
The maximum peak power I was able to get was about 2.5W:

Anything higher once again just starts clipping:

So if the Diablo 2 can only output 1.6W RMS, and 2.5W peak, why are they claiming 5W power on the box? Well, unfortunately this was also an issue on the original diablo, though even worse, that unit could only achieve about 0.8W RMS according to both my own and L7 Audiolab’s testing……

I reached out to iFi about this previously expressing frustration at the misleading way in which the Diablo specs were given, and did so again in regards to the Diablo 2. I was informed that the 5W spec was attained using iFi’s own method of testing rather than the industry standard, using not 20, but a SINGLE 1khz cycle, which is a vastly easier test for an amplifier.

The Diablo 2 does indeed do this, but this is a completely non-standard way of testing power that massively inflates the capability of a device. It’s a vastly easier test for an amplifier than either the proper peak power test or the RMS continuous power test. This spec is completely misleading because it’s not comparable to the power specs of any other amplifier. In fact if we were to test other amplifiers the same way, many of those ‘6W’ amplifiers might suddenly be 10W or 15W! It presents the Diablo 2 as being several times more powerful than it is in reality.

I want to be very clear about my stance on this:
Providing additional specs beyond the normal RMS and Peak power measurements is fine, and can be useful. (The Zahl HM1 specs above for example also provide a 1.5ms power measurement in addition to RMS and peak).
However, when a power specification is given, and that figure was NOT attained in an industry standard way / the same way as basically every other amplifier on the market, the manufacturer MUST make this clear to the consumer. Not doing so is incredibly misleading and will lead people to believe that a product is much more powerful than it actually is. You’d be pretty annoyed if you bought a car that had a claimed top speed of 200mph, but later found out it’ll only do that for 2 seconds and then maxes out at 60mph the rest of the time…..

iFi should amend the product page to make clear that this is a ‘transient’ or ‘instantaneous’ power measurement, and that the true power capability as compared to other amps is 1.6W. Not doing so is misleading consumers plain and simple and an unethical business practice.

UPDATE: It turns out that actually the SE output of this device has higher power than the 4.4mm output. You can get up to 2.3W RMS with the SE output.

Jitter:

Jitter performance is fine, though some components visible.

Upsampling Filter:

The Diablo 2 does not have user selectable reconstruction filters. The filter used is a linear phase filter. Fairly typical though some unusual rippling going on out past 23khz or so. This is notably improved over the original Diablo which used a much slower filter.

Low level signal output (Line Out):

No unusual behaviour seen, slightly noisier than ideal.

Idle Noise FFT:

THD+N vs Frequency:

Very flat overall

Intermodulation Distortion (IMD):

IMD is a little high overall. Though no sudden increase at higher output levels as on some devices.

Linearity:

Performance overall remains fairly consistent vs level. Slight increase in distortion as you go above about 2V output.

Crosstalk:

Crosstalk is a bit high on the headphone output when driving more demanding headphones. With 300 Ohm/lower current draw headphones, or when using the line out, crosstalk is lower.

Multitone:

This shows a little more clearly how things change as the load gets more difficult. It’s not that the line out is inherently loads cleaner, the headphone output is actually almost identical in performance. It’s just that distortion increases as the load gets more difficult and draws more current. This is quite normal.

Intersample Overs Test:

When DACs oversample, they can sometimes encounter a situation where the reconstructed/interpolated waveform goes above 0dBfs (the maximum possible digital value).
This is particularly common with poorly mastered music that has been ‘brickwalled’ (To clarify: I mean brick-walled as in the loudness-wars term, not brick-wall nyquist reconstruction filters https://en.wikipedia.org/wiki/Loudness_war ). But is very common in music generally unfortunately!

DACs ideally should have a few dB of digital headroom to accommodate this and reconstruct properly, and many such as Chord, RME, Benchmark etc do, and will output signals with intersample overs without any issue/distortion.
But many do not do this, either out of lack of awareness of the issue or because doing so usually means sacrificing a few dB of dynamic range and/or THD+N performance. And as a result will not be able to properly reconstruct these signals.

We can test this by creating a signal that induces intersample overs.
See in the image below. All the 44.1khz samples (squares) are below the 0dBfs limit, but the actual waveform itself, is above this limit.

The Diablo 2 does not have any internal digital headroom and therefore will clip in the presence of any intersample overs. It’s therefore recommended to set your music player to -3dB when using this device.

Summary:

The iFi Diablo 2 improves upon many aspects of the original, with features such as the IEMatch, USB-C input, xMEMS support and lossless bluetooth.

Actual performance and behaviour is nearly identical to the original, with the main difference I could find being the use of a slightly better upsampling/reconstruction filter. Therefore I wouldn’t call the Diablo 2 ‘performance upgrade’ over the original, but it does bring many welcome feature/usability improvements.

Overall, this is an excellently performing portable device suitable for both IEMs and harder to drive full size headphones such as planars. BUT, the power spec is completely incorrect/misleading and leaves a sour taste for me personally. Manufacturers should be honest with their specifications and the specs/marketing for the Diablo 2 are misleading plain and simple. A shame since otherwise my thoughts on the objective performance of this product are quite good.