Digital I/O for Edirol UA-101?

[BillB]

Hi folks

I have just got an Edirol UA-101, 10 in/out USB interface, for which Roland kindly provide W10 drivers, which auto-install when it is plugged in - very good of them for a 2005 piece of hardware:
https://www.soundonsound.com/reviews/edirol-ua101

I'm happy enough with it being 8 analogue in/outs, but of course would like to explore options for using the optical digital I/O. I would mostly be interested in an ADC for the input. I don't want to spend £100's - more like £10's!

I wondered for a brief moment whether the UA-25EX, which the UA-101 is replacing, could act as and ADC (and DAC) but it would appear that:
a) it needs to be USB bus-powered from a computer (ie it doesn't turn on when powered from a 5v USB adaptor) therefore it would register as an additional USB audio device, which would conflict with the UA-101 - or at least make using its digital I/O pointless.
b) not sure that the UA-25EX would route audio in to digital out, except 'thru' a computer...

Anyway, it did get me wondering whether there were any other low cost (S/H) interfaces with stand-alone power, which could act as an ADC to the UA-101 optical digital in.

Any thoughts on possible interfaces, or suggestions for any other low-cost line-level ADC(/DAC) ?

[BillB]

Hmmm, just spotted that there is a range of discontinued Edirol mixers from whenever ('noughties'?) with optical out - presumably pre-USB. Might look out for one of those. Other suggestions welcome

[BillB]

One other thought is that an FX unit like the TC Electronic M300 has SPDIF I/O and the manual claims that it would make a good AD/DA converter. Would need a co-ax to optical converter though. Any 'gotchas' in pursuing this route? I am guessing that one limitation would be a fixed sample rate determined by the M300 (or whatever) of 48 kHz.

[ef37a]

https://www.lindy.co.uk/audio-video-c2/ ... rter-p2730

Cannot see if that ^ is 16 or 24 bits? There are lots of converters around but read the small print! Many turn out to be D to A or coax to optical (or V-V).

If you could find one the old M-A Fast track Pro would serve as an A/D, stand alone converter. It is possible many others of the same era would, Tascam 144?

Dave.

[BillB]

Thanks Dave. Looking at the manual for the Fast Track Pro, I got the impression that it is only possible to activate the spdif outputs from the software control panel, that is, a separate USB audio connection. Ideally I would like a completely stand-alone (non- computer linked) ADC. There are few on eBay / amazon etc ,but as you say, they may be of dubious spec/quality.

I decided to order a couple of toslink optical connectors (£8 investment!) and to test the UA-101 with an old Sony minidisk deck which has optical in/out. At least that way, I should be able to see how the UA-101 behaves with digital I/O before thinking about any other kit.

[ef37a]

Thanks Dave. Looking at the manual for the Fast Track Pro, I got the impression that it is only possible to activate the spdif outputs from the software control panel, that is, a separate USB audio connection. Ideally I would like a completely stand-alone (non- computer linked) ADC. There are few on eBay / amazon etc ,but as you say, they may be of dubious spec/quality.

I decided to order a couple of toslink optical connectors (£8 investment!) and to test the UA-101 with an old Sony minidisk deck which has optical in/out. At least that way, I should be able to see how the UA-101 behaves with digital I/O before thinking about any other kit.

Yes, the FTPro does need to be configured from a PC initially but once done it remains in that state indefinitely. Had one, done it.

Dave.

[BillB]

Aah, that's very useful - thank you!

[ef37a]

Aah, that's very useful - thank you!

Should have said...This applies to every AI that I have ever read about that can run stand alone. How would it know?!!

Dave.

[James Perrett]

I decided to order a couple of tolsink optical connectors (£8 investment!) and to test the UA-101 with an old Sony minidisk deck which has optical in/out. At least that way, I should be able to see how the UA-101 behaves with digital I/O before thinking about any other kit.

I was going to suggest using an old DAT or Mini Disk machine - there are plenty of worn out DAT machines whose convertors are still fine.

[ef37a]

I decided to order a couple of tolsink optical connectors (£8 investment!) and to test the UA-101 with an old Sony minidisk deck which has optical in/out. At least that way, I should be able to see how the UA-101 behaves with digital I/O before thinking about any other kit.

I was going to suggest using an old DAT or Mini Disk machine - there are plenty of worn out DAT machines whose convertors are still fine.

Indeed James! I have a Sony and a Grundig, both in perfect working order, a shedload of tapes and not a SCOOOOBY what to do with them! I also have a working Philips DCC and tapes.

I always reckoned it was a plot that MD lost out to CD for ICE!

Dave.

[BillB]

So, 'out of the box', I have minidisc (Sony JE530) audio input being output to optical out and synced and monitored by the UA-101 optical input, no problem. So that, more or less, solves the Analog to Digital to provide an extra pair of audio ins (UA-101 channels 9-10). It seems to require 44.1kh input sample rate on the UA-101, to match the MD.

Thus far, no audio signal input to the UA-101 has been output to the optical out, as far as I can tell. The Minidisc appears to sync to the optical cable input, as the MD's display reads 'DA', meaning that the incoming digital signal is being synced and converted to analog (D>A). But there is no sound at all from the MD headphone out (yes I did turn it up!). My guess is that the UA-101's software mixer needs to be switched to pass analogue audio to digital out. A factory reset didn't fix it...

I have been playing with the UA101 without any USB/PC connection, to see how far it would go, but it looks like it's time to hook up to USB and try the config software. Hopefully it will be something like Dave said for the Fast Track Pro - set and forget....

[James Perrett]

One thing to check - your interface needs to be clocked from the digital input when you are using the MD as an A/D convertor. If you don't, you'll hear the occasional click as a sample is missed or doubled up.

[BillB]

Thanks James. The UA-101 has a nice green sync LED which flashes if its sample rate is not set correctly. You set the incoming sample rate with a knob on the UA-101 (in this case 44.1khz) which, as you say, then becomes the master clock for the interface. Once that is set, the LEd turns solid green, so all is well

[Hugh Robjohns]

Hmmm... not entirely convinced there's complete understanding here, and the Edirol's operation is slightly misleading... so, for the avoidance of confusion for anyone else following this thread, allow me to explain the processes involved.

In any digital (sub-) system there can be only one master clock device. Everything else has to be a clock slave.

In this specific situation, if you want to connect a digital input to the Edirol interface, that digital source becomes the master clock for the entire system, and the interface has to become a clock slave to that input. That's what pushing the 'Digital On' button does.

In most digital devices the sample rate of the slave is automatically configured to match the selected master device... but not in the Edirol, and that's why you need to select the appropriate sample rate.

The important point, as you've already observed, is that green LED, since it only lights solidly once the interface has achieved a stable slave clock synchronisation to the incoming digital signal.

Another important point to consider is that the technical performance of the A-D and D-A converters in the Edirol interface may well be affected (by which I mean degraded) by the clock (in)stability of the MiniDisc player.

Whether that's the case or not, and the extent to which the converter performance is degraded, depends on the design of the slave clock circuitry in the Edirol. Hopefully you won't notice any difference... but you might, and if you do the reason is the inability of the Edirol's clock recovery circuitry to remove jitter and other clocking artefacts from the incoming digital signal!

As for getting an output signal back to the MiniDisc player... I don't think this will ever work, for the simple reason that to do so will create a 'clock loop'.

In the same way that the Edirol has to become a clock slave to the MD player's digital output, the MiniDisc player has to become a clock slave to it's own digital input in order to be able to receive and decode the digital signal.

So by connecting the equipment in the way you have the Edirol is chasing the clock from the MD player, and the MD player is chasing the clock from the Edirol. It is the digital equivalent of a howl-round loop, and just as destructive! In such situations sensibly designed devices typically mute their inputs to avoid nasty noises, and I suspect that's what has happened in your MiniDisc.

Having said that, some MiniDisc recorders feature a sample-rate converter on the input, so that they can accept a digital input from sources with different sample rates and still record the 44.1kHz standard disc format. If that is the situation in this case, then the SRC essentially breaks the clock loop and the MD player remains the master clock for the entire system... and the reason for a lack of sound will probably be that the signal paths are not configured appropriately in the interface control software.

If it is a clock-loop problem, and you still want to make use of the digital output to feed an external D-A, that D-A will need to be an entirely separate device (as far as clocking is concerned) from whatever you're using to feed the digital input. In other words, a standalone D-A converter.

In that way, the MD acts as clock master, the Edirol interface slaves to the the MD player clock, and the external D-A slaves to Editrol interface. One master, and everything else is a slave.

I hope that helps.

H

[James Perrett]

The MD will only work as a D/A OR A/D. It won't do both at the same time as it normally has to be in record mode to pass a signal through it. The input selector switch on the MD should control its clocking (if it works in a sensible way). With an analogue input the MD should be master while, when it is set to digital input, the Edirol should be the master.

To avoid confusion it would be best to disconnect the SPDIF input on the Edirol when you are trying to use the MD as a D/A convertor (just in case the Edirol tries to sync to it).

[BillB]

Thank you both. I think I already had most of that but, looking back, I phrased it quite badly! Whereas you have stated it all very clearly.

I certainly wasn't aware that jitter from the MD could affect the performance of the UA's AD and DA circuits, but I can see why that makes sense if it becomes the master clock for the entire system.

Is it likely that a device like a fast track pro would have better (less) jitter than a Sony Minidisc and, as I would need a coax to optical converter, could that introduce jitter, or is it just a clock-less device which converts bits represented by light into bits represented by voltage?

Also I had wondered whether there was the possibility of a digital feedback loop, but nothing bad appeared to happen with the MD out UA in, UA out MD in setup, except for a lack of signal from UA to MD. Perhaps that was the UA refusing to close the loop. In any case, when I connected *only* UA out to MD in, although the MD seemed to acknowledge the UA as master, there was no audio data transferred, as far as I can tell.

Will get around to playing with the internal switch matrix at some point, see if that results in audio passing from UA to MD.

Thanks both for your thoughts - although this seems a bit 'niche' it may well be a route that others follow, so as you say, Hugh, your clear explanations may potentially help more than just me

[Hugh Robjohns]

Is it likely that a device like a fast track pro would have better (less) jitter than a Sony Minidisc and, as I would need a coax to optical converter, could that introduce jitter

It is likely that more modern devices will have better clocks than older devices, simply because the technology, and the engineering understanding to implement it properly, has improved dramatically over the last 20 years.

But it is a two-sided issue: you not only need a good clock in the digital source, you also need a good clock recovery circuit in the receiving device...

Moreover, when working with AES3 and S/PDIF signals, the connecting cable itself will inherently introduce jitter, even with the most blamelessly perfect clock source! The longer the cable, and the higher its capacitance, the worse the cable-induced jitter will be. So again, the emphasis really has to be on the clock-recovery circuits in the receiving device: that is really what determines the overall quality.

Optical connections don't escape either -- light scattering inside a low-quality light-fibre introduces jitter too.

Without expensive test gear to hand, the only way to find out if there is a clocking/jitter problem worth worrying about is to listen. Listen to an analogue input into the interface when running on its internal clock. Then connect the external digital device and set the interface to clock to that while still listening to the analogue source. If you can hear any quality change at all (even if you like the change!) when the interface is clocked to an external digital source then its clock recovery system is not working properly.

Whether the audible effects of inadequate clocking bother you or not is a subjective decision -- and some people like the effect, of course (although it is entirely unpredictable)!

H

[BillB]

Thanks Hugh. Complicated, isn't it?
There was me thinking that fibre would be incapable of corrupting the signal, but no!

You suggest listening to an input to determine any change in quality. Would you select any particular type of signal: a favourite well-produced CD, a pure tone, a rhythm track - or does it make no difference?

If there is a danger of corrupting the entire A-D process by trying to add another pair of inputs, which I don't desperately need, I think I may end up being happy with just the 8 built-in analogue inputs.

Thank you, it has been an education - I wasn't really paying attention in the 90's 'digital but pre-USB' era!

Bill

[BillB]

Reporting back with a little success. Finally got the studio PC up and running again (it's been 9 months, so there were a few updates!), and installed the UA-101 - plug and wait about 10 minutes while Windows downloads and installs the necessary. Automatic and successful, which is nice, but slow.

Then went to the UA-101 control panel software, and patched the Monitor channels ( i.e. All inputs) to outputs 9-10 (digital Out). Instant signal on connected minidisc, success. Changed the UA-101 from 44.1 to 48khz sample rate, still a good signal on the MD - presumably it has built in sample rate conversion? Then upped it to 88khz on the UA-101 and the MD lost sync. So that's the choice with the MD connected digitally, 44.1 or 48khz.

I saved the patch setting to UA memory 1 (it has 5, only accessible from the software control panel). As Dave predicted, the UA remembers the setting, whether or not the PC is turned on and after the UA is powered off/on. So there is now a fixed routing of Monitor to digital out. I think I will probably hook up the MD permanently, as a way to quickly capture any interesting noodlings, without the need to have the PC or anything else turned on - its very quick to hit power, record+play on the MD if inspiration strikes.

One specific question, is there any particular advantage to using 44.1 or 48khz? I seem to recall, many years ago, that there were arguments about avoiding resampling from 48 to 44.1 when burning CDs. Has that argument been consigned to the great format archive in the sky?


Here is the Frequency Response info from the manual
96.0 kHz: 20 Hz to 40 kHz (+1/-1 dB)
88.2 kHz: 20 Hz to 40 kHz (+1/-1 dB)
48.0 kHz: 20 Hz to 22 kHz (+1/-1 dB)
44.1 kHz: 20 Hz to 20 kHz (+1/-1 dB)

I think I would generally be happy with the lower overheads of the two lower sample rates, but professional knowledge would be welcome!

[Martin Walker]

One specific question, is there any particular advantage to using 44.1 or 48khz? I seem to recall, many years ago, that there were arguments about avoiding resampling from 48 to 44.1 when burning CDs. Has that argument been consigned to the great format archive in the sky?


Here is the Frequency Response info from the manual
96.0 kHz: 20 Hz to 40 kHz (+1/-1 dB)
88.2 kHz: 20 Hz to 40 kHz (+1/-1 dB)
48.0 kHz: 20 Hz to 22 kHz (+1/-1 dB)
44.1 kHz: 20 Hz to 20 kHz (+1/-1 dB)

I think I would generally be happy with the lower overheads of the two lower sample rates, but professional knowledge would be welcome!

Hi Bill!

I'm still quite happy using 44.1kHz for all my projects, for its significantly lower overheads.

First, the difference in audio quality between 44.1kHz and 96kHz recordings made with my audio interface is vanishingly small (it may not be for yours, but do try it and see before committing yourself to more than double the overheads). Yes, I can hear improvements at 96kHz when running some plug-in effects such as compressors and high frequency EQs for instance, but many of these now offer internal 'oversampling' options that enable them to reap the benefits of higher sample rates where they may be audible, without having to run the entire project at 96kHz. Moreover, many of the higher-end softsynths offer similar 'high quality' options so you can once again trade off CPU against audio quality.

Hope this helps!

Martin

[BillB]

Thank you, Martin. Yes, that's very helpful Indeed. If it's good enough for you...

A bit of rummaging on the internet "audio interface optical I/o" eventually showed that the Emu 0404 USB (as reviewed by your own fair hand in 2007) would make a good standalone AD/DA interface, as it has optical built in, so one less converter box to worry about. Need to go off and scout for a cheap one. I'll probably hook up a little Sony digital recorder I have, for capturing noodlings. (Would have used an old minidisc deck, but they are a bit too big to fit into my box room studio.)

The 0404 could also be pushed into service as an extra pair of inputs, if I really want 10 simultaneous tracks - with Hugh's jittery warning above, noted!

Incidentally, I tracked down one of Hugh's articles from April 2003 on digital clocking, in case it is of interest or use to others passing this way:
https://web.archive.org/web/20150609004 ... ocking.asp

[ef37a]

Hugh, ref "clock loop"? A couple of years ago I had two desktop PCs equipped each with a 2496 card. There was a S/PFIF coax cable (75R PTFE insulated cable, super stuff!) to and from each RCA connector.

We, son and I, could exchange digital signals across each computer each way, that is, one could be source and one destination as required.

I am pretty sure the 2496 card selected the S/PDIF input as and when 'told to' by the software?

Bill: I have quite a few MDiscs, PM if you want some.

Dave.

[Hugh Robjohns]

Changed the UA-101 from 44.1 to 48khz sample rate, still a good signal on the MD - presumably it has built in sample rate conversion?

Seems likely -- as far as I can remember, a lot of MD recorders did have SRCs.

One specific question, is there any particular advantage to using 44.1 or 48khz? I seem to recall, many years ago, that there were arguments about avoiding resampling from 48 to 44.1 when burning CDs. Has that argument been consigned to the great format archive in the sky?

Sample rate conversion is generally very good these days, and so converting between different sample rates isn't as big a deal anymore -- although it must be said that not all software SRCs are as good as they should be. The http://src.infinitewave.ca/ website makes interesting reading. You can compare different software converters -- I'd suggest using the iZotope RX plots as a reference as they offer pretty much a perfect example of how it should be done!

As to choice of sample rate, 48kHz is mandatory if you're working with video, and 44.1kHz for CD and most consumer downloads. Sound quality-wise, there really isn't anything to choose between them in practical terms, and modern converters running at 44.1 generally deliver excellent results.

There are a few specific situations where, for ultra critical applications, 96kHz capture can offer some small sonic benefits, but often the extra processing and storage overhead is a more important consideration. Non-linear signal processing (like dynamic control) also benefits from higher sample rate working, but most DAW plugins (and many hardware digital mixers) employ transparent internal oversampling/decimation for that anyway.

For most things, I record at 44.1k. Video-related tracks at 48k, and specialist stuff at 96k.

H

[BillB]

Dave, thanks for the kind offer of Minidiscs (I assume you mean discs, not players!). I think I have enough thanks, but I'll check and let you know if not.

Hugh, thanks for the further clarification on sample rates. The 44.1 general, 48 video and 96 special, is especially helpful, I might even manage to remember that.

[Hugh Robjohns]

All converters these days are delta-sigma types, which sample at stupidly high rates (with a low wordlength) and then internally decimate the signal to the required sample rate and wordlength. This arrangement is massively more accurate and consistent than the crude (by today's standards) multi-bit converters of a couple of decades ago.

However, the vast majority of delta-sigma converter chips are 'half-band' types -- for computational simplicity in the digital filtering -- which means that instead of providing >100dB attenuation at the Nyquist frequency, they only manage 6dB of attenuation.

Clearly, this fails the Nyquist requirements (let nothing through above half the sampling rate), but in practice it's a bodge that can usually be tolerated because there is rarely any significant sound going into the converters above 20kHz. Most microphones have a plummeting sensitivity at that frequency, and most musical sources exhibit a falling energy spectrum anyway.

So, in practice, there isn't usually much energy at the upper bandwidth limit, and so any aliasing that does occur will be very low level, and probably lost in the acoustic noise floor -- and in the majority of cases, this 'bodge' actually works out perfectly acceptably.

However, there are a couple of situations where it can cause problems, and these are when recording sources with strong ultrasonic outputs and peaking the levels very high. Such sources will include close-miked brass instruments, close-miked stringed instruments, close-miked cymbals and snare drums... and things like that.

Cymbals and snares drums aren't actually much of an issue as they generate an essentially noise-like signal. So any aliasing that does occur just produces more noise. Of course, the original and aliased noise will have small tonal differences, but most wouldn't notice any significant change and many might even like the effect!

In the case of strings and brass, though, the aliased components won't have the correct harmonic structure and so inherently sound wrong -- and even very low levels of aliasing can be recognised as a form of distortion.

So, in those pretty specialist situations -- I very rarely find myself sticking mics down the bells of brass instruments, or close-miking strings -- I find it better to work at 96kHz and then employ a decent software sample rate converter that applies the correct anti-alias filtering (I use iZotope's RX SRC) as post-production process after completing the mix.

Hope that helps understand the reasoning!

H