Question about different bit depths and spdif

R

Richard Curran

New member
Hoping someone here has some in-depth knowledge of the spdif protocol.
I have a very old Pro Tools system which I use on a daily basis. The AD converter is a Digidesign 882/20 io, which is specific to the DAW. The inputs are legendary for being a bit rubbish so I'd like to upgrade the converter (I know, it should have been junked 20 years ago but it works and for what I use it for it's more than adequate.). The soundcards for the Pro Tools system will only accept data from the digidesign converter though.
I'd like a Rosetta but even the used ones are bonkers money. So I'm interested in a Focusrite PRO40 Saffire. It's got an RCA type spdif output and the 882 has an RCA type spdif input so I'm assuming I'll be able to put audio into the saffire, transfer the digital signal to the 882 and on to my 21 year old G4 mac.
Now, the spdif signal from the Saffire will be 24bit, but the 882 is only 20bit. I'm aware that in the transfer the extra 4 bits will just be junked but will there be any other potential problems apart from a slight loss of resolution?
I've read plenty about going from low bit depth to high bit depth but can't find anything about the consequences of going from high to low.
Any help would be very much appreciated.
 
I have no idea about the bit-depth compatibility, but I'd be a little concerned about whether you can set up the routing, i.e., if the default mode does not work, since that requires Saffire Mix Control software. I'd test that it works on your old Mac OS.

I'm guessing you have an "in" to a great buy (or free) Saffire 40, because any of the Saffire (firewire) units would work (or not!) in the same way. I gave my 26 away for the price of shipping to someone that could use it - there's really very little market for these things, even though the preamps are supposedly very similar to the Scarlett (USB) models that followed. (You'll only have 2 inputs on any preamp going to S/PDIF, so the number of inputs shouldn't be much concern, I'd think.)

Good luck!
 
Generally speaking, it's optimal to add dither when reducing bit depth. I suspect there isn't a good way to get that done with the setup you're looking at. That said, it might not be that much of a quality loss unless you're doing stuff with ridiculous dynamic range (classical) or raising the level poorly recorded audio by a large amount.

Maybe you should just keep the Pro Tools rig the way it is to access old projects and use the Saffire via FireWire with Reaper. If the recent version of Reaper doesn't work, there are older versions you can get.
 
keith.rogers said:
I have no idea about the bit-depth compatibility, but I'd be a little concerned about whether you can set up the routing, i.e., if the default mode does not work, since that requires Saffire Mix Control software. I'd test that it works on your old Mac OS.

I'm guessing you have an "in" to a great buy (or free) Saffire 40, because any of the Saffire (firewire) units would work (or not!) in the same way. I gave my 26 away for the price of shipping to someone that could use it - there's really very little market for these things, even though the preamps are supposedly very similar to the Scarlett (USB) models that followed. (You'll only have 2 inputs on any preamp going to S/PDIF, so the number of inputs shouldn't be much concern, I'd think.)

Good luck!
Click to expand...
Thanks for the tip, I was aware that the Saffire came with software and I would be absolutely astounded if it worked on os9 :). I was kind of hoping that was just for "extras" but it's a good shout, I should check out whether it functions without it.
 
bouldersoundguy said:
Generally speaking, it's optimal to add dither when reducing bit depth. I suspect there isn't a good way to get that done with the setup you're looking at. That said, it might not be that much of a quality loss unless you're doing stuff with ridiculous dynamic range (classical) or raising the level poorly recorded audio by a large amount.

Maybe you should just keep the Pro Tools rig the way it is to access old projects and use the Saffire via FireWire with Reaper. If the recent version of Reaper doesn't work, there are older versions you can get.
Click to expand...
The dither might be an option, I think there's something in the 882 settings that allows for this but I'll check.
I work full time as a remote recording string player, so cello, viola and violin get recorded through this and frequently layered to get full strings. I have much more up to date DAWs that I use for arranging and scoring from midi files but I've yet to come across anything quite so clean and functional as Pro Tools 5.1 if all you ever do is record and some basic editing. Also I've been using it for ages. Parts and spares are cheap, I never ever have to bother with upgrades or compatibility issues and with the DSP cards there is zero latency. I really want to stick with it if I possibly can but I have a U87 and an Avalon valve pre and I'm aware that the inputs on the 882 are the weak link in the signal path.
Thanks for taking the time, much appreciated.
 
Have you tried contacting Digidesign with this question? Maybe they could answer definitively how to best improve your setup, i.e., without buying all new stuff!

I'm curious about what, exactly, other than "legend" about the inputs, that is making you do this. The inputs might be "rubbish" but all they're doing is taking a line input from your preamp, right? At least spec-wise, any noise you're getting likely are coming from before it gets to the A/D. If you're hearing something, it should be easy to test any current interface, like the Saffire, with your mic/pre setup and just record something to see if it fixes any capture/signal issues you've got.
 
Richard Curran said:
Thanks for the tip, I was aware that the Saffire came with software and I would be absolutely astounded if it worked on os9 :). I was kind of hoping that was just for "extras" but it's a good shout, I should check out whether it functions without it.
Click to expand...
I think that generation of Saffire interfaces needs the control software to be run to set up standalone mode before you can use it without a computer. The earlier Pro26io didn't need this as it had a couple of default modes which were automatically set up if it didn't detect a computer was connected. Once set up, you shouldn't need to set it up again so maybe you could borrow a more recent computer with Firewire and a supported OS in order to run the software to set it up.
 
keith.rogers said:
Have you tried contacting Digidesign with this question? Maybe they could answer definitively how to best improve your setup, i.e., without buying all new stuff!

I'm curious about what, exactly, other than "legend" about the inputs, that is making you do this. The inputs might be "rubbish" but all they're doing is taking a line input from your preamp, right? At least spec-wise, any noise you're getting likely are coming from before it gets to the A/D. If you're hearing something, it should be easy to test any current interface, like the Saffire, with your mic/pre setup and just record something to see if it fixes any capture/signal issues you've got.
Click to expand...
Hi Keith,
Thanks for the tips. Funnily enough I hadn't thought about Digidesign!
A couple of trusted and knowledgeable producers, one of who'm used the same system as me and helped me set it up told me that the inputs were the weak link. I've since had a couple of clients comment and that's not good. To be clear, there's no noise as such, in fact the signal is remarkably clear. Not so remarkable maybe as there's just an Avalon vst 737 sp in the signal path before the 882 and that's all. What I'm hearing is a certain brittleness, particularly when I'm doubling high strings, which I'm told is characteristic of these inputs.
Now, I know it's subjective and it could be the mic or the room or even, dare I say it, me! I'm also fully aware of the law of diminishing returns, but I'm slowly, slowly trying to improve each element of my set-up and having eliminated each of these potential problems as far as I'm able to, I come to the A/D. It might make no difference but I'm willing to give it a go.
 
jamesperrett said:
I think that generation of Saffire interfaces needs the control software to be run to set up standalone mode before you can use it without a computer. The earlier Pro26io didn't need this as it had a couple of default modes which were automatically set up if it didn't detect a computer was connected. Once set up, you shouldn't need to set it up again so maybe you could borrow a more recent computer with Firewire and a supported OS in order to run the software to set it up.
Click to expand...
Thanks James,
I'd literally just looked that up! I've an old macbook pro lying around that I could do the set up from firewire no problem.
 
Richard Curran said:
Ah, I had wondered, thanks for the knowledge. What would be the effect of not applying dither?
Click to expand...
My take (and I haven't really read anything lately)... When you reduce the bit depth, you can't change 0dBFS, the result is lifting the bottom, and the noise that was "hidden" down there in the larger bit depth is more noticeable, and its variance in it may make it more obvious. Dither actually adds noise to make that not noticeable. It's like spreading sawdust and peanut shells around the dance hall to cover up the chewing-tobacco-spit stains in the floor.

The problem I see is that you'd have to round-trip the digital signal from your 24-bit A/D (e.g., the F'rite) to a ditherer, and how would you do that? Another bit of gear?

If it's not noise that's the problem, then maybe not worry about this until you hear the 24->20-bit truncation.
 
Richard Curran said:
Ah, I had wondered, thanks for the knowledge. What would be the effect of not applying dither?
Click to expand...
Reverb tails and fades would not be as smooth.

I do understand your reason for wanting to stick with what you are used to, but saddling yourself with hardware from the dark days of digital might not be worth it.

Just about any interface on the market now is an order of magnitude better that those digidesign cards. If you were doing rock music, the truncation wouldn't really matter, but you are working in one of the genres where truncating at 20 bit will actually make a difference.
 
Richard Curran said:
...but I've yet to come across anything quite so clean and functional as Pro Tools 5.1 if all you ever do is record and some basic editing.
Click to expand...
I started on 7.4. I found it to be brutally primitive compared to the other DAW I was using at the time. I believe it was 40 bit fixed-point processing at that time. Floating point processing is much cleaner, and it might actually be an audible improvement for the kinds of things you say you're recording, even for basic editing.
 
Richard Curran said:
Thanks James,
I'd literally just looked that up! I've an old macbook pro lying around that I could do the set up from firewire no problem.
Click to expand...

In which case you would probably be better off installing Reaper and using a Protools theme for Reaper if you want to make it look like Protools. You can then use the Pro 40 directly.

Old versions of Protools have issues with dithering - it isn't always handled correctly so you need to be very careful and know exactly what you are doing to avoid your audio being truncated. This could be another reason for the sound quality issues that you are encountering.
 
bouldersoundguy said:
I started on 7.4. I found it to be brutally primitive compared to the other DAW I was using at the time. I believe it was 40 bit fixed-point processing at that time. Floating point processing is much cleaner, and it might actually be an audible improvement for the kinds of things you say you're recording, even for basic editing.
Click to expand...
I'm learning as I go! As I understand it, floating point is most relevant when it comes to plug-ins, summing and high volume signals (do correct me if I'm wrong). Apart from auditioning arrangements for clients (rather than chucking 30 odd tracks of strings at them straight off), I never mix and I never use plug-ins. My signal source is always from the same 6 instruments and there are rarely calls for producing transients that come anywhere near what could be described as high volume.
All my stuff goes to clients as straight and unadorned exported individual audio tracks. Is the floating point thing so important in this context? I'm guessing SNR is a factor but is it enough of a factor?
 
keith.rogers said:
My take (and I haven't really read anything lately)... When you reduce the bit depth, you can't change 0dBFS, the result is lifting the bottom, and the noise that was "hidden" down there in the larger bit depth is more noticeable, and its variance in it may make it more obvious. Dither actually adds noise to make that not noticeable. It's like spreading sawdust and peanut shells around the dance hall to cover up the chewing-tobacco-spit stains in the floor.

The problem I see is that you'd have to round-trip the digital signal from your 24-bit A/D (e.g., the F'rite) to a ditherer, and how would you do that? Another bit of gear?

If it's not noise that's the problem, then maybe not worry about this until you hear the 24->20-bit truncation.
Click to expand...
Thanks for the tip. I certainly will listen to that.
One of the reasons I'm asking is so I can decide whether I should swap the 882/20 for an 888/24, then (hopefully) no truncation would be needed processing the digital signal from a better A/D. These dusty old boxes are still out there and are cheap enough.
I'm aware that this all sounds like pointless penny pinching, I actually have a much more up to date version of Logic on a mac pro and Studio One on a reasonably capable Windows machine which I use for arranging. I just really like the Pro Tools version I have for actually recording takes. Mainly because of its simplicity and negligeable latency. I might be flogging a dead horse but if I can get better out of it without changing the surface there are practical advantages for me. And that's why I'm here!
 
Richard Curran said:
I'm learning as I go! As I understand it, floating point is most relevant when it comes to plug-ins, summing and high volume signals (do correct me if I'm wrong). Apart from auditioning arrangements for clients (rather than chucking 30 odd tracks of strings at them straight off), I never mix and I never use plug-ins. My signal source is always from the same 6 instruments and there are rarely calls for producing transients that come anywhere near what could be described as high volume.
All my stuff goes to clients as straight and unadorned exported individual audio tracks. Is the floating point thing so important in this context? I'm guessing SNR is a factor but is it enough of a factor?
Click to expand...
I think it will matter when you do edits (e.g. crossfading the edges of a punch) or adjust the volume, pretty much anything other than record, dither/truncate and bounce with no other processing. But I don't know how much it will matter. In any case, going from fixed to floating point processing was one of the few improvements to the actual sound of DAWs.
 
You must log in or register to reply here.
Back
Top