USB Audio limits? [Archive]

Boardin' Fool

October 21st 03, 12:30 AM

I had an M-Audio Quattro (4 in & 4 out) for awhile and had serious problems
with it. If I wanted to record at 96 khz, I could only record two tracks at
once. If I pushed it down to 44.1 khz, I could theoretically record four
tracks. However, whenever I tried to record two tracks, I'd get about one
minute in and get a dropout. Windows would then pop up a window saying I
used all of the USB bandwidth and it was shutting down the hardware.

"Jack A. Zucker" > wrote in message
. net...
> > For 16 bit, 44k audio:
> >
> > (((16 x 44100) / 1024) / 8) = 86KB / Second (mono)
> >
> > For USB (12mBit/Sec)
> >
> > ((12 * 1024) / 8) = 1536KB / Second
> >
> > Maximum audio mono:
> >
> > 1536KB / 86KB = 17 tracks mono
> >
> > Maximum audio stereo
> >
> > 17 / 2 = 8 stereo tracks
>
>
> I just thought about this some more (duh...)
>
> My M-Audio USB Duo handles 2 mono inputs and 2 mono outputs. Therefore,
the
> maximum I/O that would be going over USB would be:
>
> 4*86KB/Second
>
> Or...
>
> (((16 x 44100) / 1024) / 8) * 4
>
> or
>
> 345k
>
> which is less than 1/4th the total USB bandwidth. The real limitation will
> be the processing power of cakewalk which has to combine the audio and
midi
> tracks, apply it's virtual effects algorithms and apply them into a stereo
> mix to send to the USB audio device for playback.
>
> Jaz
>
>

Arny Krueger

October 21st 03, 01:15 AM

"Boardin' Fool" > wrote in message

> I had an M-Audio Quattro (4 in & 4 out) for awhile and had serious
> problems with it. If I wanted to record at 96 khz, I could only
> record two tracks at once. If I pushed it down to 44.1 khz, I could
> theoretically record four tracks. However, whenever I tried to
> record two tracks, I'd get about one minute in and get a dropout.
> Windows would then pop up a window saying I used all of the USB
> bandwidth and it was shutting down the hardware.
>

The point is that few computer interfaces transfer data at anything like
their maximum bitrate. In the case of USB, we're talking a highly complex
protocol with lots of overhead.

2 * 96000 * 8 = 1,536,000 bps which is only a tiny fraction of USB's
maximum bitrate of 12 megabits per second.

So much for USB 1.1 and multichannel audio.

There seems to be a lot more *life* in USB 2.0. I know that for CDs and hard
drives, it rocks!

There still aren't a lot of USB 2.0 interfaces. AFAIK the Roland UA-1000 is
the only really interesting one that seems to be available for immediate
delivery.

Jack A. Zucker

October 21st 03, 02:36 AM

No,

The point is that USB 1 has plenty of bandwidth for 44k audio with 2
channels in and 2 channels out which is what my posting was about.

"Arny Krueger" > wrote in message
...
> "Boardin' Fool" > wrote in message
>
>
> > I had an M-Audio Quattro (4 in & 4 out) for awhile and had serious
> > problems with it. If I wanted to record at 96 khz, I could only
> > record two tracks at once. If I pushed it down to 44.1 khz, I could
> > theoretically record four tracks. However, whenever I tried to
> > record two tracks, I'd get about one minute in and get a dropout.
> > Windows would then pop up a window saying I used all of the USB
> > bandwidth and it was shutting down the hardware.
> >
>
> The point is that few computer interfaces transfer data at anything like
> their maximum bitrate. In the case of USB, we're talking a highly complex
> protocol with lots of overhead.
>
> 2 * 96000 * 8 = 1,536,000 bps which is only a tiny fraction of USB's
> maximum bitrate of 12 megabits per second.
>
> So much for USB 1.1 and multichannel audio.
>
> There seems to be a lot more *life* in USB 2.0. I know that for CDs and
hard
> drives, it rocks!
>
> There still aren't a lot of USB 2.0 interfaces. AFAIK the Roland UA-1000
is
> the only really interesting one that seems to be available for immediate
> delivery.
>
>

Romeo Rondeau

October 21st 03, 02:57 AM

On modern CPU's this is not where the bottleneck is. Mixing tracks is no
sweat for a CPU, even modest one's (like my 866 p3 can handle well over 32
tracks and still mix them and apply effects with no problem. Get a newer
machine and you can add a good amount of virtual synths to those tracks. USB
sucks, really bad. It was never intended for serious audio use. It's a
miracle that they could even get it to be somewhat practical at all.

>
> which is less than 1/4th the total USB bandwidth. The real limitation will
> be the processing power of cakewalk which has to combine the audio and
midi
> tracks, apply it's virtual effects algorithms and apply them into a stereo
> mix to send to the USB audio device for playback.
>
> Jaz
>
>

Jack A. Zucker

October 21st 03, 03:01 AM

"Romeo Rondeau" > wrote in message
...
> USB
> sucks, really bad.

Is that a technical term?!? Everyone keeps saying that, yet 4 tracks of 44k
audio is less than 1/4th the total bandwidth of USB1. Why will that not
work. Please explain.

Jaz

reddred

October 21st 03, 04:19 AM

"Jack A. Zucker" > wrote in message
. net...
> No,
>
> The point is that USB 1 has plenty of bandwidth for 44k audio with 2
> channels in and 2 channels out which is what my posting was about.
>

It's pretty good for stereo audio - but not much else, media wise. This
mostly has to do with the way usb transfers information - only half the time
is data actually transfered. The other half is spent recieving and issuing
requests. Also, a huge chunk of the bandwidth is consumed by data buffers.

You don't have as much problem with accurate file transfers because you can
put a great big cache in software on the host side. If you try that with
audio, there goes your 'low-latency' out the windows.

USB only gets lamer when you start to add devices - say a 2x2 audio device
and a USB midi controller. Each device uses half the bandwidth, even though
the controller would need next to nothing.

Well, that's why it's a low-speed 'serial bus'. It's not supposed to do
anything fast, really - it was desinged with dial-up modems in mind. That's
why the 'usb vs. firewire' thing was always ridiculous, 'apples and oranges'
as they say. It's good that we don't have to choose.

Like I said, it's fine for stereo overdubs.

jb

Randall Hyde

October 21st 03, 04:54 AM

The *very* best USB host controller stacks (device driver on the host side)
seem to be capable of supporting only about 50% of the theoretical USB
bandwidth. A faster machine helps, but not as much as you would think.
ISO transfers *should* do better, but...
Cheers,
Randy Hyde

"Jack A. Zucker" > wrote in message . net...
> Does anyone have anything other than anecdotal or theoretical evidence of
> the limits of USB Audio?
>
> Here's my analysis but I've yet to do a full test.
>
> For 16 bit, 44k audio:
>
> (((16 x 44100) / 1024) / 8) x = 86KB / Second (mono)
>
> For USB (12mBit/Sec)
>
> ((12 * 1024) / 8) = 1536KB / Second
>
> Maximum audio mono:
>
> 1536KB / 86KB = 17 tracks mono
>
> Maximum audio stereo
>
> 17 / 2 = 8 stereo tracks
>
>
>
> --
> http://www.jackzucker.com
>
>

Randall Hyde

October 21st 03, 04:57 AM

"Jack A. Zucker" > wrote in message . net...
> "Romeo Rondeau" > wrote in message
> ...
> > USB
> > sucks, really bad.
>
> Is that a technical term?!? Everyone keeps saying that, yet 4 tracks of 44k
> audio is less than 1/4th the total bandwidth of USB1. Why will that not
> work. Please explain.

Because the host controller software, that moves those packets
around in memory and puts them on/off the bus, stinks.

Microsoft's stacks are especially guilty of this.
The best I've seen, on the average, is about 50% bus utilitization.
You can put together some special tests that do a little bit better,
but in the real world you don't get much better than about 50%
bandwidth.

As memory speeds increase, you should do better.

Now USB 2.0 High-Speed should do better.
Cheers,
Randy Hyde

Geoff Wood

October 21st 03, 05:12 AM

PalmeGårIgen

October 21st 03, 09:18 AM

I use the quattro on an Toshiba laptop, and since i upgraded my XP home
edition it has done exactly what it is supposed to.
Works great for me

/Per
"Boardin' Fool" > skrev i meddelandet
...
> I had an M-Audio Quattro (4 in & 4 out) for awhile and had serious
problems
> with it. If I wanted to record at 96 khz, I could only record two tracks
at
> once. If I pushed it down to 44.1 khz, I could theoretically record four
> tracks. However, whenever I tried to record two tracks, I'd get about one
> minute in and get a dropout. Windows would then pop up a window saying I
> used all of the USB bandwidth and it was shutting down the hardware.
>
>
> "Jack A. Zucker" > wrote in message
> . net...
> > > For 16 bit, 44k audio:
> > >
> > > (((16 x 44100) / 1024) / 8) = 86KB / Second (mono)
> > >
> > > For USB (12mBit/Sec)
> > >
> > > ((12 * 1024) / 8) = 1536KB / Second
> > >
> > > Maximum audio mono:
> > >
> > > 1536KB / 86KB = 17 tracks mono
> > >
> > > Maximum audio stereo
> > >
> > > 17 / 2 = 8 stereo tracks
> >
> >
> > I just thought about this some more (duh...)
> >
> > My M-Audio USB Duo handles 2 mono inputs and 2 mono outputs. Therefore,
> the
> > maximum I/O that would be going over USB would be:
> >
> > 4*86KB/Second
> >
> > Or...
> >
> > (((16 x 44100) / 1024) / 8) * 4
> >
> > or
> >
> > 345k
> >
> > which is less than 1/4th the total USB bandwidth. The real limitation
will
> > be the processing power of cakewalk which has to combine the audio and
> midi
> > tracks, apply it's virtual effects algorithms and apply them into a
stereo
> > mix to send to the USB audio device for playback.
> >
> > Jaz
> >
> >
>
>

Jack A. Zucker

October 21st 03, 12:07 PM

"Geoff Wood" -nospam> wrote in message
...
> Because a bunch of extra bandwidth is burned up with all the other stuff
> that needs to squirt down the wire to keep USB talking.

Come on....You're telling me protocol and header packet information take up
over 75% of the bandwidth?

Arny Krueger

October 21st 03, 12:27 PM

"Jack A. Zucker" > wrote in message
. net

> "Geoff Wood" -nospam> wrote in message
> ...

>> Because a bunch of extra bandwidth is burned up with all the other
>> stuff that needs to squirt down the wire to keep USB talking.

> Come on....You're telling me protocol and header packet information
> take up over 75% of the bandwidth?

The proof of the theory lies in the lab work, right?

How many tracks can a USB 1.1 interface handle with accuracy and reliability
equal to that of a PCI, Firewire, or USB 2.0 interface?

In another post you claimed:

"The real limitation will
be the processing power of cakewalk which has to combine the audio and midi
tracks, apply it's virtual effects algorithms and apply them into a stereo
mix to send to the USB audio device for playback."

So, do the experiment in a program with low CPU overhead on a system with
CPU power to burn.

Hint: the hidden agenda with audio interfaces is that not only do they
require bandwidth, they require bandwidth when they need it, not a few
seconds later. The obvious end run is buffering, but that can increase
latency.

Les Cargill

October 21st 03, 12:54 PM

"Jack A. Zucker" wrote:
>
> "Geoff Wood" -nospam> wrote in message
> ...
> > Because a bunch of extra bandwidth is burned up with all the other stuff
> > that needs to squirt down the wire to keep USB talking.
>
> Come on....You're telling me protocol and header packet information take up
> over 75% of the bandwidth?

Half, anyway. It's a bus, and there have to be delays between packets.
The drivers on the host computer probabaly suck, too. USB was
designed for keyboards and mice, and to replace serial ports.

For (shared) Ethernet, the recommendation is to start thinking about upgrading
when it gets to 40% utilization. A similar dynamic is in effect here.

--
Les Cargill

Jack A. Zucker

October 21st 03, 01:17 PM

"Arny Krueger" > wrote in message
...
> The proof of the theory lies in the lab work, right?

Right - Certainly not in technical terms such as "USB Sucks!". :-)

> How many tracks can a USB 1.1 interface handle with accuracy and
reliability
> equal to that of a PCI, Firewire, or USB 2.0 interface?

I don't know because I don't have a PCI, Firewire or USB 2 audio interface
to compare it with. Furthermore in theory, it looks like the USB 1 solution
will have more than enough bandwidth to handle 8-12 tracks but I'm planning
on doing a test this week to find out.

With absolutely no optimization of my notebook or win xp pro, I recorded 5
simultaneous audio tracks and had absolutely no glitches.

http://www.jackzucker.com/JazGuitar/audio/mp3/ex1blues.mp3

Unfortunately, the recording software that came with the USB Duo is horrible
so I'm awaiting the arrival of Sonar to do a more thorough test.

>
> In another post you claimed:
>
> "The real limitation will
> be the processing power of cakewalk which has to combine the audio and
midi
> tracks, apply it's virtual effects algorithms and apply them into a stereo
> mix to send to the USB audio device for playback."
>
> So, do the experiment in a program with low CPU overhead on a system with
> CPU power to burn.
>
> Hint: the hidden agenda with audio interfaces is that not only do they
> require bandwidth, they require bandwidth when they need it, not a few
> seconds later. The obvious end run is buffering, but that can increase
> latency.
>
>
>

Jack A. Zucker

October 21st 03, 01:20 PM

"Les Cargill" > wrote in message
...
> "Jack A. Zucker" wrote:
> >
> > "Geoff Wood" -nospam> wrote in message
> > ...
> > > Because a bunch of extra bandwidth is burned up with all the other
stuff
> > > that needs to squirt down the wire to keep USB talking.
> >
> > Come on....You're telling me protocol and header packet information take
up
> > over 75% of the bandwidth?
>
> Half, anyway. It's a bus, and there have to be delays between packets.
> The drivers on the host computer probabaly suck, too. USB was
> designed for keyboards and mice, and to replace serial ports.
>
> For (shared) Ethernet, the recommendation is to start thinking about
upgrading
> when it gets to 40% utilization. A similar dynamic is in effect here.

That's still well under what this audio interface will be reading or writing
to the USB. More like 22%.

Remember, it'll never be seeing more than 2 input and 2 output channels.

Les Cargill

October 21st 03, 01:42 PM

"Jack A. Zucker" wrote:
>
> "Les Cargill" > wrote in message
> ...
> > "Jack A. Zucker" wrote:
> > >
> > > "Geoff Wood" -nospam> wrote in message
> > > ...
> > > > Because a bunch of extra bandwidth is burned up with all the other
> stuff
> > > > that needs to squirt down the wire to keep USB talking.
> > >
> > > Come on....You're telling me protocol and header packet information take
> up
> > > over 75% of the bandwidth?
> >
> > Half, anyway. It's a bus, and there have to be delays between packets.
> > The drivers on the host computer probabaly suck, too. USB was
> > designed for keyboards and mice, and to replace serial ports.
> >
> > For (shared) Ethernet, the recommendation is to start thinking about
> upgrading
> > when it gets to 40% utilization. A similar dynamic is in effect here.
>
> That's still well under what this audio interface will be reading or writing
> to the USB. More like 22%.
>
> Remember, it'll never be seeing more than 2 input and 2 output channels.

I don't have specific models for specific traffic loads for
specific devices, so I dunno. I was more addressing the "Come
on" part of your post.

I vaguely remember discussions where 4 in, 4 out was pretty
much living dangerously over USB. I'd be guessing, but
40% is probably about where it starts to get unreliable. 4 in
4 out seems to be about that limit, if we all did the
math right.

So far as I know, lots of people use 2-in/2-out with USB all
the time. You get an Arkansas Guarantee with that - if it
breaks in half, you get to keep both halves.

--
Les Cargill

Richard

October 21st 03, 02:11 PM

wrote...
> "Geoff Wood" -nospam> wrote in message
> > Because a bunch of extra bandwidth is burned up with all the other stuff
> > that needs to squirt down the wire to keep USB talking.
>
> Come on....You're telling me protocol and header packet information take up
> over 75% of the bandwidth?

<cough>Appletalk<cough>

--
Exporting jobs is treason.

Clark '04

Arny Krueger

October 21st 03, 03:16 PM

"Jack A. Zucker" > wrote in message
. net
> "Arny Krueger" > wrote in message
> ...

>> The proof of the theory lies in the lab work, right?

> Right - Certainly not in technical terms such as "USB Sucks!". :-)

Put yourself in my shoes - the order of the day is 8-16 channels of 24/44.
USB 2.0 or Firewire sure looks good!

>> How many tracks can a USB 1.1 interface handle with accuracy and
>> reliability equal to that of a PCI, Firewire, or USB 2.0 interface?

> I don't know because I don't have a PCI, Firewire or USB 2 audio
> interface to compare it with.

AFAIK, any of them are cool with 8-16 channels of 24/44, or more.

> Furthermore in theory, it looks like
> the USB 1 solution will have more than enough bandwidth to handle
> 8-12 tracks but I'm planning on doing a test this week to find out.

> With absolutely no optimization of my notebook or win xp pro, I
> recorded 5 simultaneous audio tracks and had absolutely no glitches.

> http://www.jackzucker.com/JazGuitar/audio/mp3/ex1blues.mp3

That's all fine and good, but if you really want the best shot at hearing an
audio interface fall apart, 100 Hz sine waves are IME hard to beat. The
problem is that a lot of really nice music has plenty of tics and pops of
its own.

> Unfortunately, the recording software that came with the USB Duo is
> horrible so I'm awaiting the arrival of Sonar to do a more thorough
> test.

In a previous post you said that you could record and play 2 tracks at the
same time. That's 4 concurrent tracks or about 25% of the maximum data rate
of USB 1.1.

Now, you've upped that to 5 tracks recording (no playback claimed) which is
31% or so of the maximum data rate of USB 1.1. Again, that's all fine and
good, but its not exactly leaping tall buildings with a single bound.

I'm genuinely interested in how far you can go, and wish you the best of
luck in getting there as easily and happily as possible.

I'm not sure that Sonar is the best choice, but maybe it will work out just
fine.

But, if things don't go your way with Sonar, remember that N-Track is as
close as a download and the plastic equivalent of a bit more than a couple
of $20 bills.

Keep us informed!

Boardin' Fool

October 21st 03, 05:17 PM

I thought the Duo had only two ins and two outs. If that's correct, then
your five tracks were all from the same two inputs? If that is the case,
then you aren't actually increasing the amount of bandwidth used by the USB
device. Your software would be the one doubling the inputs to different
tracks in the software.

"Jack A. Zucker" > wrote in message
. net...
> "Arny Krueger" > wrote in message
> ...
> > The proof of the theory lies in the lab work, right?
>
> Right - Certainly not in technical terms such as "USB Sucks!". :-)
>
> > How many tracks can a USB 1.1 interface handle with accuracy and
> reliability
> > equal to that of a PCI, Firewire, or USB 2.0 interface?
>
> I don't know because I don't have a PCI, Firewire or USB 2 audio interface
> to compare it with. Furthermore in theory, it looks like the USB 1
solution
> will have more than enough bandwidth to handle 8-12 tracks but I'm
planning
> on doing a test this week to find out.
>
> With absolutely no optimization of my notebook or win xp pro, I recorded 5
> simultaneous audio tracks and had absolutely no glitches.
>
> http://www.jackzucker.com/JazGuitar/audio/mp3/ex1blues.mp3
>
> Unfortunately, the recording software that came with the USB Duo is
horrible
> so I'm awaiting the arrival of Sonar to do a more thorough test.
>
> >
> > In another post you claimed:
> >
> > "The real limitation will
> > be the processing power of cakewalk which has to combine the audio and
> midi
> > tracks, apply it's virtual effects algorithms and apply them into a
stereo
> > mix to send to the USB audio device for playback."
> >
> > So, do the experiment in a program with low CPU overhead on a system
with
> > CPU power to burn.
> >
> > Hint: the hidden agenda with audio interfaces is that not only do they
> > require bandwidth, they require bandwidth when they need it, not a few
> > seconds later. The obvious end run is buffering, but that can increase
> > latency.
> >
> >
> >
>
>

reddred

October 21st 03, 05:32 PM

"Richard" > wrote in message
.. .
> wrote...
> > "Geoff Wood" -nospam> wrote in message
> > > Because a bunch of extra bandwidth is burned up with all the other
stuff
> > > that needs to squirt down the wire to keep USB talking.
> >
> > Come on....You're telling me protocol and header packet information take
up
> > over 75% of the bandwidth?
>
> <cough>Appletalk<cough>
>

Rows of five inch screens waiting... and waiting...

jb

Romeo Rondeau

October 21st 03, 05:36 PM

What USB audio interface has 12 channels? None that I know of. Why don't you
buy an RME and a small digital board and record and playback 24 tracks with
24 outputs all day long with no problems instead of speculating on what USB
can and can't do?

> I don't know because I don't have a PCI, Firewire or USB 2 audio interface
> to compare it with. Furthermore in theory, it looks like the USB 1
solution
> will have more than enough bandwidth to handle 8-12 tracks but I'm
planning
> on doing a test this week to find out.

Romeo Rondeau

October 21st 03, 05:41 PM

Then your track count has nothing to do with your interface. What is to be
taken into account is the physical number of inputs and outputs and the
bandwidth requirements of that, not the track count. The computer doesn't
send every track to the interface, just the buss outputs.

> Remember, it'll never be seeing more than 2 input and 2 output channels.
>
>

Romeo Rondeau

October 21st 03, 05:42 PM

<cough>Not relevent<cough>

> <cough>Appletalk<cough>
>
> --
> Exporting jobs is treason.
>
> Clark '04

Richard

October 21st 03, 06:13 PM

wrote...
> <cough>Not relevent<cough>

<cough>Obtain a sense of humor<cough>

--
Clark '04

MikeK

October 21st 03, 06:39 PM

"Richard" > wrote in message
.. .
>
>
> wrote...
> > <cough>Not relevent<cough>
>
> <cough>Obtain a sense of humor<cough>
>

DAMN, somebody get these guys a drink of water!

Jack A. Zucker

October 21st 03, 06:40 PM

"Boardin' Fool" > wrote in message
...
> I thought the Duo had only two ins and two outs. If that's correct, then
> your five tracks were all from the same two inputs? If that is the case,
> then you aren't actually increasing the amount of bandwidth used by the
USB
> device. Your software would be the one doubling the inputs to different
> tracks in the software.

That's correct and that's what I said initially when I said the max
bandwidth I'd ever consume on the bus would be 345k...

Go back and read my other posts and you'll see that I said exactly that.

Jack A. Zucker

October 21st 03, 06:42 PM

"Arny Krueger" > wrote in message
...
> In a previous post you said that you could record and play 2 tracks at the
> same time. That's 4 concurrent tracks or about 25% of the maximum data
rate
> of USB 1.1.
>
> Now, you've upped that to 5 tracks recording (no playback claimed) which
is
> 31% or so of the maximum data rate of USB 1.1. Again, that's all fine and
> good, but its not exactly leaping tall buildings with a single bound.

I think I said several times that with 2 ins and 2 outs the max I can ever
have on the USB bus is 345k (4 tracks). The 5 tracks I was referring to were
in the software, not on the USB bus. I realize those are 2 different
bandwidth paths.

> I'm not sure that Sonar is the best choice, but maybe it will work out
just
> fine.

Just got it today. So far so good. Truthfully, I'd have opted for Cubase if
I was starting from scratch but I already had Cakewalk Pro Audio 9 and it
was "only" $150 to upgrade.

> But, if things don't go your way with Sonar, remember that N-Track is as
> close as a download and the plastic equivalent of a bit more than a
couple
> of $20 bills.
>

Really?!? Thanks, I'll check it out.

Boardin' Fool

October 21st 03, 07:26 PM

You're right (of course), I just got a little mixed up in the posts. I
ended up getting rid of my USB devices and going with a Echo Layla. A
friend of mine was upgrading from that to a Pro Tools 002R system, so he
sold it to me cheap ($225). If it works for you, that's great. It just
didn't work for me. As for the actual specs of USB, I've never really
looked them up.

"Jack A. Zucker" > wrote in message
. net...
> "Boardin' Fool" > wrote in message
> ...
> > I thought the Duo had only two ins and two outs. If that's correct,
then
> > your five tracks were all from the same two inputs? If that is the
case,
> > then you aren't actually increasing the amount of bandwidth used by the
> USB
> > device. Your software would be the one doubling the inputs to different
> > tracks in the software.
>
> That's correct and that's what I said initially when I said the max
> bandwidth I'd ever consume on the bus would be 345k...
>
> Go back and read my other posts and you'll see that I said exactly that.
>
>

Geoff Wood

October 21st 03, 07:45 PM

"Jack A. Zucker" > wrote in message
. net...
> "Geoff Wood" -nospam> wrote in message
> ...
> > Because a bunch of extra bandwidth is burned up with all the other stuff
> > that needs to squirt down the wire to keep USB talking.
>
> Come on....You're telling me protocol and header packet information take
up
> over 75% of the bandwidth?

That plus other inefficiencies, yes.

M-Audio clearly state that my Transit USB will do 2 in /2 out at 24/48K, but
is limited to one-way only at 24/96, totally due to USB capbilies. Do the
math yourself....

geoff

Jack A. Zucker

October 21st 03, 07:51 PM

"Geoff Wood" -nospam> wrote in message
...
> M-Audio clearly state that my Transit USB will do 2 in /2 out at 24/48K,
but
> is limited to one-way only at 24/96, totally due to USB capbilies. Do the
> math yourself....

I already did. Sounds too much like Reiki and aromatherapy to me. Lots of
conjecture and guess-work and faith. I'll try it and let you guys know how I
fare. Fortunately, I have 25 days of my 30 day return policy left. I hate
spending another $175 to get the M-Audio Firewire 410 if I don't really need
it. If I can get my laptop to reliably do 2 in/2 out at 44k, that'll be
fine.

Jack A. Zucker

October 21st 03, 07:52 PM

"Boardin' Fool" > wrote in message
...
> You're right (of course), I just got a little mixed up in the posts. I
> ended up getting rid of my USB devices and going with a Echo Layla. A

I'd love to have that but it's more than my budget will allow - Particularly
after paying for my recent spinal surgery!

> friend of mine was upgrading from that to a Pro Tools 002R system, so he
> sold it to me cheap ($225). If it works for you, that's great. It just
> didn't work for me.

Great deal! :-)

> As for the actual specs of USB, I've never really
> looked them up.

Thanks for the info. I appreciate the time.

Arny Krueger

October 21st 03, 10:27 PM

"Richard" > wrote in message

> Exporting jobs is treason.

It's the intersection of capitalism and technological progress.

Ben Bradley

October 22nd 03, 03:22 AM

Oh, so now I see it's crossposted amomg not just two, but THREE
groups...

In alt.music.4-track,rec.audio.pro,rec.music.makers.guitar, "MikeK"
> wrote:

>
>"Richard" > wrote in message
.. .
>>
>>
>> wrote...
>> > <cough>Not relevent<cough>
>>
>> <cough>Obtain a sense of humor<cough>
>>
>
>
>DAMN, somebody get these guys a drink of water!

They remind me how glad I am I quit smoking.

Jack A. Zucker

October 24th 03, 03:53 AM

Just a follow up.

So far in 3-4 days of working with the M-Audio USB Duo, I'm not getting any
USB glitches. Maybe USB doesn't suck as bad as some folks thing?!?

OTOH, maybe I'll be bitchin' and moanin' in a couple weeks about how awful
it is. However, for the moment it appears to do what it says it'll do.

Jaz

"Geoff Wood" -nospam> wrote in message
...
>
> "Jack A. Zucker" > wrote in message
> . net...
> > "Geoff Wood" -nospam> wrote in message
> > ...
> > > Because a bunch of extra bandwidth is burned up with all the other
stuff
> > > that needs to squirt down the wire to keep USB talking.
> >
> > Come on....You're telling me protocol and header packet information take
> up
> > over 75% of the bandwidth?
>
>
> That plus other inefficiencies, yes.
>
> M-Audio clearly state that my Transit USB will do 2 in /2 out at 24/48K,
but
> is limited to one-way only at 24/96, totally due to USB capbilies. Do the
> math yourself....
>
> geoff
>
>

Romeo Rondeau

October 24th 03, 05:51 AM

When your needs exceed the bandwidth, you will :-) Nobody said that USB
interfaces won't do what they claim, just that they don't claim to do much
because they don't do much. When your need for inputs grows (and it will),
you'll have to buy a non-USB interface and get rid of your old one. If
that's not a problem with you, knock yourself out :-)

> So far in 3-4 days of working with the M-Audio USB Duo, I'm not getting
any
> USB glitches. Maybe USB doesn't suck as bad as some folks thing?!?
>
> OTOH, maybe I'll be bitchin' and moanin' in a couple weeks about how awful
> it is. However, for the moment it appears to do what it says it'll do.

Jack A. Zucker

October 24th 03, 11:54 AM

"Romeo Rondeau" > wrote in message
...
> Nobody said that USB
> interfaces won't do what they claim

Actually, most folks said exactly that.

>, just that they don't claim to do much
> because they don't do much. When your need for inputs grows (and it will),
> you'll have to buy a non-USB interface and get rid of your old one. If

Not sure I'll ever need more than 2 inputs though. For live recording, I
prefer a simple stereo mic setup and for studio stuff and being a jazz
musician, I prefer to just do a complete take through the stereo inputs
using a mixer if necessary to mix the band down to stereo before it hits the
interface.

If I ever need more than that, I'll give the current setup to my kid. At
$229, it's not like I had to mortgage the house to get it.

Jaz

Ben Bradley

October 24th 03, 04:25 PM

In alt.music.4-track,rec.audio.pro,rec.music.makers.guitar, "Jack A.
Zucker" > wrote:

>"Romeo Rondeau" > wrote in message
...
>> Nobody said that USB
>> interfaces won't do what they claim
>
>Actually, most folks said exactly that.
>
>>, just that they don't claim to do much
>> because they don't do much. When your need for inputs grows (and it will),
>> you'll have to buy a non-USB interface and get rid of your old one. If
>
>Not sure I'll ever need more than 2 inputs though. For live recording, I
>prefer a simple stereo mic setup and for studio stuff and being a jazz
>musician, I prefer to just do a complete take through the stereo inputs
>using a mixer if necessary to mix the band down to stereo before it hits the
>interface.

Apparently the USB interface doesn't have much 'bandwidth headroom'
- if you install more software on this machine, it may 'break' your
interface. An audio editor is probably okay, but anything that
installs a bootup program (such as 'hplamp' to turn off a scanner lamp
after a few minutes, or especially any virus scanner), will take up
'background' processing time, and the processor may not get back to
the USB driver before a buffer overruns.
Problems occurr because audio recording is one of the few real-time
tasks on personal computers. When a computer runs more background
tasks, it doesn't 'break' most programs, they just run noticably
slower, but such tasks can easily break audio recording.
A friend said he was recording LP's (using the built-in soundcard),
but was having problems (clicks apparently from buffer overruns) on
his 350MHz system. I've successfully recorded on a 200MHz machine, so
I knew it was possible for his system to run fine. I told him to look
at what's going on at bootup and what background tasks might be
running that would use up processor resources. He later said his virus
scanner and several other tasks had been running, and when he removed
them he had no problem.
I suspect this is the main problem people have with USB audio
interfaces, and not everyone knows enough to be able to diagnose and
fix it. No one needs a virus checker running while recording sound,
and if you happen to capture some infectious music, the virus checker
won't get it anyway.

>If I ever need more than that, I'll give the current setup to my kid. At
>$229, it's not like I had to mortgage the house to get it.
>
>Jaz

Jack A. Zucker

October 24th 03, 05:26 PM

"Ben Bradley" > wrote in message
...
> Apparently the USB interface doesn't have much 'bandwidth headroom'
> - if you install more software on this machine, it may 'break' your
> interface. An audio editor is probably okay, but anything that
> installs a bootup program (such as 'hplamp' to turn off a scanner lamp
> after a few minutes, or especially any virus scanner), will take up
> 'background' processing time, and the processor may not get back to
> the USB driver before a buffer overruns.

But that has nothing to do with USB bandwidth. In those cases, firewire or
USB2 would have the same issues. When you're talking about the operating
system buffering up queues and latency of serially processing background
tasks, the speed of the bus is no longer the bottleneck.

> running that would use up processor resources. He later said his virus
> scanner and several other tasks had been running, and when he removed
> them he had no problem.

Of course. Virus scanners are a waste of time (and processing power) anyway.

> I suspect this is the main problem people have with USB audio
> interfaces, and not everyone knows enough to be able to diagnose and
> fix it. No one needs a virus checker running while recording sound,
> and if you happen to capture some infectious music, the virus checker
> won't get it anyway.

You referring to Marilyn Manson? :-)

EggHd

October 24th 03, 05:39 PM

I'll tell you what (and I have said it before) the $400 or MBox has been the
best by far price for return piece of gear I have ever bought 30 fold.

---------------------------------------
"I know enough to know I don't know enough"

Goldtop71

October 25th 03, 06:35 AM

I picked up a used "Layla" 8-track PCI set-up for $255.00 so you might not
be as far away from the whole shebang if you want to go that route.
Personally, I am used to Fostex self-contained systems, i.e. D90, D160
because they are essentially just tape recorders and I treat them as such.
Just upgraded to Sonar 3.0 so we'll see how bad I choke.

"Jack A. Zucker" > wrote in message
. net...
> "Romeo Rondeau" > wrote in message
> ...
> > Nobody said that USB
> > interfaces won't do what they claim
>
> Actually, most folks said exactly that.
>
> >, just that they don't claim to do much
> > because they don't do much. When your need for inputs grows (and it
will),
> > you'll have to buy a non-USB interface and get rid of your old one. If
>
> Not sure I'll ever need more than 2 inputs though. For live recording, I
> prefer a simple stereo mic setup and for studio stuff and being a jazz
> musician, I prefer to just do a complete take through the stereo inputs
> using a mixer if necessary to mix the band down to stereo before it hits
the
> interface.
>
> If I ever need more than that, I'll give the current setup to my kid. At
> $229, it's not like I had to mortgage the house to get it.
>
> Jaz
>
>

Laurence Payne

October 26th 03, 02:27 PM

On Mon, 20 Oct 2003 16:30:02 -0700, "Boardin' Fool"
> wrote:

>I had an M-Audio Quattro (4 in & 4 out) for awhile and had serious problems
>with it. If I wanted to record at 96 khz, I could only record two tracks at
>once. If I pushed it down to 44.1 khz, I could theoretically record four
>tracks. However, whenever I tried to record two tracks, I'd get about one
>minute in and get a dropout. Windows would then pop up a window saying I
>used all of the USB bandwidth and it was shutting down the hardware.

I use a M-Audio Duo for location recording to a laptop.

I don't get the error message, but, using the supplied ASIO drivers,
recording fails after a time. It works well recording into Cubase
using the program's Duplex ASIO drivers or the MME drivers. As
latency isn't an issue for what I'm doing, this is OK.

CubaseFAQ page www.laurencepayne.co.uk/CubaseFAQ.htm

David Morton

October 26th 03, 04:33 PM

In article >,
(Jack A. Zucker) wrote:

> "Romeo Rondeau" > wrote in message
> ...
> > Nobody said that USB
> > interfaces won't do what they claim
>
> Actually, most folks said exactly that.

I've lost count of the number of times I've been told that USB audio
interfaces don't work, cannot work, or that even if I do get them working
they'll have dropouts and unacceptable latencies.

And yet my Edirol UA-5 and MBox just worked perfectly right out of the
box(es).

But then my (old model, straight-across bridge) Godin ACS has perfect
intonation too. :-)

Ben

December 13th 03, 02:17 AM

Let's look at some figures...

USB2.0 has a max bandwidth of 480Mbs (bits)

A modern hard drive typically does 100MBS (bytes) = 800Mbs

That isn't bad.

The size of your RAM and processor speed is far more critical than the
storage bandwidth.

Getting recorded data to the CPU and memory whilst recording is far more
important getting stored on disk, this can occur in the nanosecond gaps
where system resources can be spared.

"Romeo Rondeau" > wrote in message
...
> When your needs exceed the bandwidth, you will :-) Nobody said that USB
> interfaces won't do what they claim, just that they don't claim to do much
> because they don't do much. When your need for inputs grows (and it will),
> you'll have to buy a non-USB interface and get rid of your old one. If
> that's not a problem with you, knock yourself out :-)
>
> > So far in 3-4 days of working with the M-Audio USB Duo, I'm not getting
> any
> > USB glitches. Maybe USB doesn't suck as bad as some folks thing?!?
> >
> > OTOH, maybe I'll be bitchin' and moanin' in a couple weeks about how
awful
> > it is. However, for the moment it appears to do what it says it'll do.
>
>
>

Jukka Andersson

December 13th 03, 06:54 AM

for me I use E-Magic emi 2/6 usb audiodevice for main purposes also
i have tweaked SB Live! for softsynths (like analog wannabes)

before this setup I had ST-Audio DSP2000 10in/10out. Nice card but
I am happy I changed for emagic.

Only negative thing with usb devices is that those does eat a little more
CPU
that PCI cards does.

..jukka

"Ben" > wrote in message
...
> Let's look at some figures...
>
> USB2.0 has a max bandwidth of 480Mbs (bits)
>
> A modern hard drive typically does 100MBS (bytes) = 800Mbs
>
> That isn't bad.
>
> The size of your RAM and processor speed is far more critical than the
> storage bandwidth.
>
> Getting recorded data to the CPU and memory whilst recording is far more
> important getting stored on disk, this can occur in the nanosecond gaps
> where system resources can be spared.
>
>
>
> "Romeo Rondeau" > wrote in message
> ...
> > When your needs exceed the bandwidth, you will :-) Nobody said that USB
> > interfaces won't do what they claim, just that they don't claim to do
much
> > because they don't do much. When your need for inputs grows (and it
will),
> > you'll have to buy a non-USB interface and get rid of your old one. If
> > that's not a problem with you, knock yourself out :-)
> >
> > > So far in 3-4 days of working with the M-Audio USB Duo, I'm not
getting
> > any
> > > USB glitches. Maybe USB doesn't suck as bad as some folks thing?!?
> > >
> > > OTOH, maybe I'll be bitchin' and moanin' in a couple weeks about how
> awful
> > > it is. However, for the moment it appears to do what it says it'll do.
> >
> >
> >
>
>

Kurt Albershardt

December 13th 03, 08:20 PM

Ben wrote:
> Let's look at some figures...
>
> USB2.0 has a max bandwidth of 480Mbs (bits)
>
> A modern hard drive typically does 100MBS (bytes) = 800Mbs

Only while emptying or filling its ram buffer.

Current 7200 RPM ATA drives (like those used in external drive
enclosures) peak around 60 mbytes/sec (at the outer edge of the platter)
and drop to half that or less at the inside cylinders.

Laurence Payne

December 13th 03, 11:21 PM

On Sat, 13 Dec 2003 02:17:01 GMT, "Ben" >
wrote:

>USB2.0 has a max bandwidth of 480Mbs (bits)
>
>A modern hard drive typically does 100MBS (bytes) = 800Mbs

Isn't that the burst transfer speed? Not applicable to sustained
transfer of large files, e.g. audio.

CubaseFAQ page www.laurencepayne.co.uk/CubaseFAQ.htm

Tony

December 15th 03, 02:09 AM

You are incorrect. It is 480Mbs (megabits per second) and not 480 MBS (megabytes per second).

Tony

On Sat, 13 Dec 2003 12:20:04 -0800, Kurt Albershardt > wrote:

>Ben wrote:
>> Let's look at some figures...
>>
>> USB2.0 has a max bandwidth of 480Mbs (bits)
>>
>> A modern hard drive typically does 100MBS (bytes) = 800Mbs
>
>Only while emptying or filling its ram buffer.
>
>Current 7200 RPM ATA drives (like those used in external drive
>enclosures) peak around 60 mbytes/sec (at the outer edge of the platter)
>and drop to half that or less at the inside cylinders.
>
>

Cornelius J Rat

December 15th 03, 11:13 PM

"Tony" > wrote
> You are incorrect. It is 480Mbs (megabits per second) and not 480 MBS
(megabytes per second).
>
> Tony
>
>
> On Sat, 13 Dec 2003 12:20:04 -0800, Kurt Albershardt > wrote:
>
> >Ben wrote:
> >> Let's look at some figures...
> >>
> >> USB2.0 has a max bandwidth of 480Mbs (bits)
> >>
> >> A modern hard drive typically does 100MBS (bytes) = 800Mbs
> >
> >Only while emptying or filling its ram buffer.
> >
> >Current 7200 RPM ATA drives (like those used in external drive
> >enclosures) peak around 60 mbytes/sec (at the outer edge of the platter)
> >and drop to half that or less at the inside cylinders.
> >
> >
>
Hmm. I bet you felt silly when you re-read that, Tony.

Tony

December 16th 03, 05:08 AM

X On Mon, 15 Dec 2003 23:13:12 -0000, "Cornelius J Rat" > wrote:

>
>"Tony" > wrote
>> You are incorrect. It is 480Mbs (megabits per second) and not 480 MBS
>(megabytes per second).
>>
>> Tony
>>
>>
>> On Sat, 13 Dec 2003 12:20:04 -0800, Kurt Albershardt > wrote:
>>
>> >Ben wrote:
>> >> Let's look at some figures...
>> >>
>> >> USB2.0 has a max bandwidth of 480Mbs (bits)
>> >>
>> >> A modern hard drive typically does 100MBS (bytes) = 800Mbs
>> >
>> >Only while emptying or filling its ram buffer.
>> >
>> >Current 7200 RPM ATA drives (like those used in external drive
>> >enclosures) peak around 60 mbytes/sec (at the outer edge of the platter)
>> >and drop to half that or less at the inside cylinders.
>> >
>> >
>>
>Hmm. I bet you felt silly when you re-read that, Tony.
>
Why? I posted the correct information (megabits, not megabytes!)

Tony

Mike McKernan

December 16th 03, 12:59 PM

On Tue, 16 Dec 2003 05:08:20 GMT, Tony > wrote:

>X On Mon, 15 Dec 2003 23:13:12 -0000, "Cornelius J Rat" > wrote:
>
>>
>>"Tony" > wrote
>>Hmm. I bet you felt silly when you re-read that, Tony.
>>
>Why? I posted the correct information (megabits, not megabytes!)
>
Yes, but so did the original poster.

salvarsan

December 17th 03, 02:28 AM

Laurence Payne wrote:
> Ben wrote:

> >USB2.0 has a max bandwidth of 480Mbs (bits)
> >
> >A modern hard drive typically does 100MBS (bytes) = 800Mbs
>
> Isn't that the burst transfer speed? Not applicable to sustained
> transfer of large files, e.g. audio.

The same argument applies to USB 2.0.

As a rule of thumb, I estimate the worst-case bandwidth
as 25% of the maximum. 480Mbs burst -> 120Mbs -> 15MBs.
15 Megabytes/sec worst case isn't too shabby.

Sixteen 16-bit channels at 44kHz is around 1.5MBs.
The hardware has plenty of bandwidth for streaming
data like that. Once you do mixing and equalization
on top of the GUI, you're working the computer hard.

-drh
--

Jack A. Zucker

December 17th 03, 05:27 PM

dan lavry

December 17th 03, 09:38 PM

salvarsan > wrote in message >...
> Laurence Payne wrote:
> > Ben wrote:
>
> > >USB2.0 has a max bandwidth of 480Mbs (bits)
> > >
> > >A modern hard drive typically does 100MBS (bytes) = 800Mbs
> >
> > Isn't that the burst transfer speed? Not applicable to sustained
> > transfer of large files, e.g. audio.
>
> The same argument applies to USB 2.0.
>
> As a rule of thumb, I estimate the worst-case bandwidth
> as 25% of the maximum. 480Mbs burst -> 120Mbs -> 15MBs.
> 15 Megabytes/sec worst case isn't too shabby.
>
> Sixteen 16-bit channels at 44kHz is around 1.5MBs.
> The hardware has plenty of bandwidth for streaming
> data like that. Once you do mixing and equalization
> on top of the GUI, you're working the computer hard.
>
> -drh

I am amazed at how easy it is to B.S. so many people.

The first thing: in order to run say 480MBs, one needs to be able to
at least handel a 480MHz sine wave transmission. Now, in
instrumentation, we use coaxial cables with carfully terminated
impeadance. Folks, 480MHz is half way to microwaves... That 10$ USB
cable is anorder of magnitude away from being a flimzy way of
transmition of such rates.

The second thing: We are talking audio, not just data transfer. Say
you have an AD or a DA at the end of the USB cable. Where do you need
the stable LOW JITTER clock? Inside the computer? Near the converter?
Of course near the converter. So we are now dealing with the need to
synchronize the computer data rate to an external device. That does
alter the picture a lot.

The third thing is the main one: Yes, you can write your own drivers
and come up with all sorts of hickups and problems. Why not use the
generic drivers? They are very stable!!! The answer is: Generic
drivers are 16 bits. If we want to DECIVE the customer with the magic
word 24 bits, we need to write our own drivers. Yes, there are 24 bits
there, but the imlication that bore bits yield better accuracy is
often false. You need wide words for computations, but not so for real
life converters.

24 bits amounts to 144dB dynamic range. Take a +/-15V analog supply
and the lowest noise amplifier I can buy at any cost! With BALANCED
12.28V peak to peak signal (Full scale AES/EBU) I have 34.72V peak to
peak. That is a huge signal. If I want the noise to be low enough to
have 144dB, it needs to be about 5 nano volts/square root Hz, for a
20KHz audio!

This is all a theoretical excersize, because no microphone is going to
yield me a 34.72V peak to peak signal! I need to amplify the small
signal, and if I use say only 24dB gain, than my signal to noise ratio
is down by 24dB which is 3 bits degredation. I am now in 21 bit world
(126dB range)... That is extreamly good! Take a look at the best Mic
pre-amplifiers around. What do you see? 127dBu range? In other words,
24 bits is not just about the digits we use. Some of those digits are
"just there doing nothing". It is like counting real money with 4
digits for cents. Instead of saying $1.27, we say $1.2700. The right 2
digits are always 00.

Now, lets go into USB and computers world: Say I have a 5V analog
supply enviornment, and I use a 5V power supply enviornment instaed of
the +/-15V above (which is equivalent to 30VDC). I just reduced my
signal range by about 6:1 and lost about 15.5dB in signal to noise...
Another 2.5 bits gone...

That too is a terribly optimistic picture. That 5V from the USB bus is
so dirty, that by the time you clean it, you are probably starting at
3.3VDC, not 5V... and you are entering the 16 bit performance world.
Do not feel bad. Most CD's are less than 16 bits in terms of signal to
noise, and some of them are great sounding!

But, don't spend your money on BS hype. 24bit USB device for $200? You
can not get 24 bits for $200000. For reasonably low cost, you will do
very well to really achive 16 bits. That is why generic USB drivers
are a better choice. They are robust, and you do not need to forward
the extra 8 "nothing bits".

Now lets go and compute a real good stable USB audio device...

BR

Dan Lavry
www.lavryengineering.com

I use generic USB drivers, and avoide that whole set of issues
regarding having to write my own drivers.