Universal Audio Apollo Twin X Review (Audio Interface)

[Koeitje]

This times 100.

There is a few places doing measurements on consumer audio. You would have pro audio all to yourself. I get the Pro side isn't of main interest to many here, but maybe Audio Science could embrace the recording side more than it does.

Even though its not the full set of measurements you can do for an interface like this, seeing a high quality DAC output does provide a good indication that the manufacturer knows what they are doing. Sure some of the other functionality might not be at the same level, but lets just say I have more faith in it if the DAC hits 112dB SINAD instead of 85dB SINAD.

 

[hyperplanar]

I hope I can provide some context for people that may be confused about what the purpose/market positioning of this device is compared to other pro-level interfaces:

UAD’s devices have always had competent measurements, but for all intents and purposes, the main reason people buy these over other interfaces is to use UAD plugins, which is a whole proprietary ecosystem in itself. UAD plugins only run on the DSP cores in their devices. Although this isn’t as much of a concern anymore with the vast increase in available CPU power compared to when the system was first introduced, offloading these plugins to the hardware DSP was a great benefit in the past to conserve CPU power for the DAW and other plugins. Running one good amp simulation on a Pentium 4 and barely having leftover resources for everything else is a whole different ballgame compared to simultaneously running dozens on the processors we have now. There are also quite a few UAD-exclusive plugins that people still seem to prefer sonically over purely software-based plugins, but I do feel that software plugins have made enormous progress in recent years with some surpassing UAD offerings.

For these reasons, they’re commonly found in pro studios to this day. However, due to the added DSP cores, these devices don’t provide the best value-for-performance as an audio interface, and are only interesting if one wants to get into that whole ecosystem of UAD plugins. I loved the ergonomics and beautiful design, but personally, I prefer to stick to all software plugins to keep my projects as portable as possible, so that ruled out UAD’s offerings for me, and I went with an RME Babyface Pro instead.

 

[IVX]

what's a beautiful device! Doesn't matter good or so-so inside but outlook I like already

 

[digitalfrost]

It was kingly sent to me by the company for testing.

As it should be, your majesty.

 

[phoenixsong]

I have been SO looking forward to this!
Thanks for the awesome review as always

 

[phoenixsong]

The DAC and cleanliness of the headamp (not power) in the Motu M2 can compete, but the ADC of this is looking good! I could drive the Motu M2 to clipping on many of my headphones at louder than listening but bearable levels

 

[PeteL]

I do this normally. As I explained however, there is no way for me to test the ADC on this unit because it only exposes a 64-bit ASIO interface which my Audio Precision 32-bit app cannot see. UA products are the only ones so far that I have seen this limitation. For all others I have a dedicated section for ADC.

I expect we should see more and more tough, any Insight if it's something Audio Precision is working on and could be available eventually or if it's hardware based and we'll need a fully new generation of analysers?

 

[Rja4000]

I expect we should see more and more tough, any Insight if it's something Audio Precision is working on and could be available eventually or if it's hardware based and we'll need a fully new generation of analysers?

It's software. Nothing prevents them to provide 64 bits drivers.
But the question is: what will they gain from it ?
Of course, sooner or later, 32 bits drivers will become the exception.
But we're not there yet.

 

[PeteL]

It's software. Nothing prevents them to provide 64 bits drivers.
But the question is: what will they gain from it ?
Of course, sooner or later, 32 bits drivers will become the exception.
But we're not there yet.

I am not sure I understand what you mean. Audio Precision is the reference, we're in front of a piece of equipment that cannot be fully measured by it. I don't think it's reasonable to ask audio gear manufacturers to make 32 bits version of their audio drivers when they will be connected to 64 bits operating systems, just so they can be measured by Audio Precision Gear?

 

[TheTalbotHound]

How do the ADCs in this compare to say a Motu M4 or RME babyface pro FS/ADI2 pro FS?

 

[SmackDaddies]

it is borderline criminal on their part.

that is a little harsh

 

[liu]

the HP output is not great because it is using tpa6120a2. worth measuring the output impedance.

 

[peng]

I can answer these questions.
1, It's noise modulation issue on 901x series DACs. This is different from the actual imd increase at that range.
2, Of course it's different this is much much cheaper machine.
3, Can't be solved.
4, Likely not nor important. This is very down in level. For years tons of music was created on machines perform way worse and people are trying to emulate them. I don't think it will be audible even less so than the actual imd increase in the 9038pro/q2m.

1) What do you mean by different from the actual imd increase at that range?

I would think that the -50 to 0 dBFS is the important range and with the hump, IMD ration did not fall below -90 until about -13 dBFS.

4) How come Oppo's didn't seem to have that hump, or at least not as secure.

I have 3 Oppo devices that used the 9018 and 9038pro, that's why I am interested in yours, or Amir's comments.

Thank you.

 

[JohnYang1997]

1) What do you mean by different from the actual imd increase at that range?

I would think that the -50 to 0 dBFS is the important range and with the hump, IMD ration did not fall below -90 until about -13 dBFS.

4) How come Oppo's didn't seem to have that hump, or at least not as secure.

I have 3 Oppo devices that used the 9018 and 9038pro, that's why I am interested in yours, or Amir's comments.

Thank you.

View attachment 113132

It's the noise that's increased not noise. You'll see the difference in FFT.
When noise in the implementation is higher than the chip itself it won't show that.
Implementation would probably reduce it to someone extend but won't eliminate.
All the above is for 901x.

On to the IMD increase for 9038pro/q2m that's actually IMD increase in the range. And implementation would almost completely eliminate it. And there's still a little noise modulation especially for q2m. And there are different ways to eliminate the IMD increase, or different things in implementation can cause the IMD increase. Clock, Vref, I/V circuit just to name a few.

 

[Alexander Gödde]

Even though its not the full set of measurements you can do for an interface like this, seeing a high quality DAC output does provide a good indication that the manufacturer knows what they are doing. Sure some of the other functionality might not be at the same level, but lets just say I have more faith in it if the DAC hits 112dB SINAD instead of 85dB SINAD.

This exactly. The pro audio companies are not implementing any esoteric concepts but are generally engineering-driven. If their engineers know how to get things right in what Amir measures then there is a high likelihood of them getting other things right as well (though of course no assurance). The reviews here can server to call out those companies that do not have the knowledge or the motivation to generally engineer things right, which is already a large help.

 

[lc155]

I love the look of this device for desktop mic usage (no XLRs on front), but I hate how wonky this thing is with Windows, particularly in getting it to recognise stuff like Discord, etc.

Doesn't want to play nice and so it's a no-go for me over something like the MOTU M2/M4.

 

[jtwrace]

I should send my Rodecaster Pro in too. Would be interesting to see what it does.

 

[Dave Tremblay]

Hi @Dave Tremblay

Some questions if you could kindly assist:

1. Can you kindly comment on the 'ESS IMD hump' here?

2. Is this interface using different ESS DAC chip to your X16, which has no signs of the ESS hump.

3. And have Universal Audio engineers discussed with ESS Labs themselves - do they offer any advice to solve this? Are they aware of it? Do they even care or are they indifferent?

4. Does Universal Audio think there is any audibility impact with this hump? Or Universal Audio don't think it's an issue at all, either audibility or engineering issue?

Thanks!

Happy to answer these questions, but I have to say @JohnYang1997 kind of nailed it.

This is indeed an ES9016S part in our TwinX products. It performs quite well for the price, power draw, and physical size. It is a different part than the X16, largely due to different goals in those two products. While this Quad DSP unit is up there a bit in price, it has the same analog/digital subsystem as the Duo, which is $899. Very different price point, very different thermal requirements, but still excellent performance.

We absolutely work with ESS while designing our products and they are a valued partner of ours. As to audibility of some of these measurements, I think it depends. While I appreciate the use of SINAD here as a singular benchmark, when it comes to audibility, it makes sense to separate noise from distortion. Especially in pro studio recording market. As others have mentioned, some amounts of tasteful distortion are often added to recordings. These levels of distortion are WAY above what you'd see measured here. Orders of magnitude above, in fact. Add to that there isn't a transducer (headphone, speaker, etc) that is within an order of magnitude of these distortion levels, and it could be hard to argue audibility at this level. That said, it is certainly possible you could find someone who can hear it, under specific circumstances. And if we can prevent uncontrollable distortion, we'll do it, even if only so you can add your distortion of choice.

Noise, on the other hand, is a very audible problem, especially in studio settings. When we listen to mastered music, we are listening right near full scale signals, let's say within 10dB even for light touch albums. During the recording process, it's common to track with high levels of overhead so you don't clip, maybe as much as 20dB for dynamic signals. You lose 20dB of noise floor right there. Then you apply gain, compression, and EQ during the mixing phase, and you may bring up the noise floor another 20-30dB. This becomes audible quick. Worst case scenario is something like a guitar signal going through a modeled guitar amp, which can see 80dB of gain. And all of that noise stacks when you mix your tracks together.

But the goal of a product like this is find the right compromises to enable a customer to create a great sounding record. That requires balancing many different things, while keeping the cost down. I'll probably write a separate post on that process.

 

[Dave Tremblay]

For pure raw performance the RME ADI-2 PRO is the better choice. However the internal DSPs are probably why one would buy it if one has an application on hand. And if one uses distortion based sound "enhancement" plugins the ESS hump probably is swamped in the plugins distortion anyway.

Yep. Product design is really about tradeoffs and where you spend money to get to a price point. The ADI-2 PRO is definitely a different target market, at nearly 2x the price of the TwinX Duo, and vastly different feature set.

 

[Music1969]

We absolutely work with ESS while designing our products and they are a valued partner of ours.

Thanks Dave but this didn't answer my question: when you work with ESS Labs, are they aware of this "ESS IMD hump" ?

And do they care and offer solutions?

Or they don't care?

Or they aren't aware?