Should I replace my ageing MOTU converters

[mat6066]

In my studio I use 4 MOTU units for A/D D/A conversion. I have two 1224 units that operate at a maximum of 48k and two 2408 Mk3 units that operate at a maximum of 96k. For recording, I use the inputs on my 1224 unit (its converters are 128x oversampling with a S/N ratio of 116dBfs) and the ADAT inputs on my Mk3 unit, sending audio from a Focusrite 428 (its converters have a S/N ratio of better than 120dBfs 'A weighted'). The audio is sent to summing mixer from the outputs of the units (the 1224 D/A conversion spec is the same as its A/D whilst the the Mk3 D/A converters are 128x oversampling with a S/N ratio of 106dBfs. Finally, I print the mix back into my DAW via the 1224.

I record drums at 96k and dither down for editing/mixing. Everything else I record either at 44.1 or 48.

Given all this. what improvement in sound quality would I get (if any) from replacing this rather ageing set up with something more modern? I'm considering units from RME, Antelope, Focusrite and MOTU.

Thanks in advance for any advice.

Mat

[Hugh Robjohns]

Given all this. what improvement in sound quality would I get (if any) from replacing this rather ageing set up with something more modern?

There is no doubt that converter performance has improved over recent decades. The average dynamic range today is around 117dB (A-WTD), with the very best pushing beyond 120dB (A-WTD) -- and its broadly the same for A-D as D-A. Off the top of my head, I think the lowest dynamic range I've measured in recent years was about 107dB (A-WTD), and that was a Behringer Ada8200 which is still a very usable product...

Having said that, these numbers really tell you more about the calibre of the internal engineering of the products, rather than their audible quality -- the sound differences in ageing converters compared to more modern units is very, very, small and quite subtle, and you'd need very, very good monitoring conditions to be able to hear it in most cases.

Where you might pick up on the weaknesses of older converters is when recording sources with high levels of HF -- like cymbals and snare drums, etc -- and peaking at high levels (ie. tracking with minimal headroom) and at base sample rates (44/48k), since those conditions are more likely to reveal any aliasing issues... But if you're happy to work at 96kHz you avoid all those risks so it won't be a problem anyway (assuming good quality SRC algorithms).

Personally, I'd only consider changing the converters when you know they have become the weakest link in the signal chain. Usually the recording/monitoring room acoustics are the weak-link... and then the monitoring speakers... and then the mics... I still routinely use an old Apogee PSX100, because while I have other, technically better converters, it still sounds fine for the majority of purposes.

And why worry about the difference in dynamic range between 106 and 120dB when the source recordings probably struggle to exceed 60dB anyway, and the CD/mp3 release format only manages 90dB?

Finally -- and intended as positive education for you and other readers rather than nit-picking -- I fear some confusion in the terminology and potential for misunderstandings...

... I use the inputs on my 1224 unit (its converters are 128x oversampling with a S/N ratio of 116dBfs

A ratio, like a signal-to-noise or dynamic range ratio, is expressed simply in decibels or dB. Adding a suffix, like 'u' or 'FS' indicates that the number relates to a defined reference level, and turns the ratio into an absolute value.

So, if you have a signal level of, say -20dBFS, and apply 6dB of gain, you end up with a signal level of -14dBFS, and the system might have a S/N ratio of 100dB...

I record drums at 96k and dither down for editing/mixing.

'Dithering' relates to a reduction in word-length, such as from 24-bits to 16-bits. What you're describing is 'sample-rate conversion' (SRC) from 96k to 44.1k (or whatever).

Hope that helps.

H

[mat6066]

Thanks as ever, Hugh for your thoughtful and detailed reply. Apologies for the sloppy use of terminology!

There does seem to be a wide range of views on the subject (I posted the same query on Gearslutz) so I'm going to audition a few different units next week against my current system (the Orion Studio, an RME Fireface UFX2 with Ferroface A32 and a second hand Apogee Symphony set up with an AD16X and a DA16X) to see if I can hear perceptible differences/improvements. I shall post my conclusions.

Cheers,
Mat

[Hugh Robjohns]

There does seem to be a wide range of views on the subject...

Naturally!

I posted the same query on Gearslutz...

Ah. As a general rule of thumb, I would expect GS to be in favour of the latest, most fashionable and most expensive alternatives, because they are AWESOME and everything else SUCKS...

so I'm going to audition a few different units next week against my current system

Good idea... But best done as a true double-blind comparison to avoid the effect of confirmation bias.

H

[mat6066]

Actually the GS responses were far more measured than I'd expected - https://www.gearslutz.com/board/so-much ... st12579420

If I may enquire, Hugh, what's the best way of testing the units in a double-blind experiment?
I'm presuming sources such as a drum kit with some cymbals and an acoustic guitar would be the best choices to put the converters through their paces. To make it double blind, should I ask a friend to do the recording without them knowing which unit is being used and with me not present?

Cheers,
Mat

[CS70]

Thanks as ever, Hugh for your thoughtful and detailed reply. Apologies for the sloppy use of terminology!

There does seem to be a wide range of views on the subject (I posted the same query on Gearslutz) so I'm going to audition a few different units next week against my current system (the Orion Studio, an RME Fireface UFX2 with Ferroface A32 and a second hand Apogee Symphony set up with an AD16X and a DA16X) to see if I can hear perceptible differences/improvements. I shall post my conclusions.

Cheers,
Mat

There's a wide range of views on most subjects simply because physics does not require to know what you're talking about before talking.

When you test, make sure you don't know what's what. Otherwise you're gonna easily fool yourself.

[Hugh Robjohns]

If I may enquire, Hugh, what's the best way of testing the units in a double-blind experiment?

It's actually quite tricky to get right. In a 'blind' comparison the listener doesn't know what they are listening to, but the person running the test does. In a double-blind comparison neither the person listening, nor the operator know which example is which until after the test. In that way there can be no subconscious direction, no language that subtly suggests which is the new shiny converter and which is the old converter...

I'm presuming sources such as a drum kit with some cymbals and an acoustic guitar would be the best choices to put the converters through their paces. To make it double blind, should I ask a friend to do the recording without them knowing which unit is being used and with me not present?

Yes, pick some typical sources, and get someone else to label the files in such a way that you have no idea what is what. The really critical challenge, though, will be in making sure that the levels are identical through each converter -- in both the digital and analogue domains -- since small level changes can have a big effect psychoacoustically.

H

[mat6066]

Thanks Hugh. Much appreciated.

[mat6066]

So, I've been running tests with an RME UFX2, an Antelope Orion Studio and my current set up (MOTU 2408mk3 and 1224 units). I'll be posting the files up soon but on preliminary listening I've found that I prefer the D/A sound of one of the units but the A/D sound of another.

Given that the specs of the converters of each unit are respectively the same on the way in and the way out this has surprised me. Are my ears deceiving me? Has anyone else found this when comparing converters? I've been very careful to make sure all the files have been gain matched.

[Hugh Robjohns]

Don't assume the A-D and D-A stages in a converter share identical specs just because they are in the same box! In fact it's quite common to find different dynamic range figures, for example. Also, a lot of designs use converter chips from different manufacturers for the A-D and D-A stages, which will almost certainly have different filter performances at the very least.

One of the other things that can make an audible difference to the sound of a converter is the clocking configuration - with external clocking creating a potential quagmire of confusing possibilities! You haven't said how you've hooked up your test devices... But this might offer an explanation for your perceptions, too. Internal clocking is almost always the best option where practical.

I get concerned when people use the word 'prefer' in the context of a converter, because it's not supposed to be a contributor to a sound character, it's supposed to convert signals between the analogue and digital domains transparently and accurately. It seems to me that if you can hear it changing the source sound, however subtly, it's not working properly... Even if you like the way it changes it!

[The Korff]

Tell that to Burl!

[Hugh Robjohns]

Yes... there are a number of converters that claim to 'do nice things' to the sound going in or coming out... and I get why people might think that's a good thing.

With my engineering head on it just worries me. I like to keep the creative elements of the signal path separate from the functional parts...

When it comes to an A-D 'sounding nice', I'm not so worried because what it does is inherently encoded into the recording and remains there for all time as part of the sound everyone hears.

The bigger concern is with the D-A since the only person hearing what that does is the content creator. Everyone else is using a different D-A that won't be 'sounding nice' in the same way... which could potentially be a problem for them and misleading for the content creator.

It's all a lot easier and more reliable if converters are designed to be completely transparent (under all operating conditions). We all know where we stand then, and it's a lot easier to spot the good ones amongst the not-so-good ones!

H