Audioarts AIR-3 USB recording level low?

[KZME IT]

We recently bought 2 AIR-3 consoles... After initial setup, I ran a few test recordings. At first, I thought nothing was getting to the computer. The app I was using didn't show signals that low. After confirming the USB device was showing up in the devices, I tried a few tones.

After confirming that 0dBu was exiting the program bus (PGM1 and PGM2), I did another recording. The tone I was using to confirm 0dBu, is being recorded at -23dBfs! (A firewire device connected to the PGM1 output, records the same tone at -12dBfs.)

I tried the our 2nd AIR-3 console, same results. I've tried both OS versions I have on site. (Mac OS 10.6 and 10.7)

After doing the recording tests, I tried a few sine waves out of the USB... and found that signals from -3dBu to 0dBu exit the USB output RJ45 at the same signal level?? (-4 and below, output levels with the expected decreases.)

Everything else seems to work as expected. Other USB and Firewire gear using the same tones and apps do not have these signal level issues.

Has anyone seen anything like this? Am I missing something horribly obvious? :

 

[rew]

I had the same problem with my 2 Air-3 consoles. I had sound cards in my computers so I used them for recording. Another computer that isnt used to record is connected to the USB port. I tried it on an XP pro and windows 7 machine with same results.
I hope you get an answer that will help you increase the record level. I will be watching this thread closely.

 

[Lazy J]

Just a shot in the dark... Have you looked at the internal dip switches? There is a set of dip switches inside the console that select which bus feeds the USB input. It could be possible that neither bus is selected.

 

[KZME IT]

We tested the 2 consoles out of the box, and I recently changed one unit to use PGM2 for a confirmation of the results across both consoles. I'm recalling trying both busses early on to confirm they were the same.

I'm fairly certain that the switches are set correctly. Wouldn't that be a forehead slap.
I'll confirm it tomorrow.

We've a few external adapters. All provide a nice signal, where this does not. While I hate to see anyone else go through this, I am relieved to know that someone else is having the same problems. The amount of skepticism coming my way has been a bit much.

Any idea when those 2 were manufactured? I had an odd conversation with support that suggested some sort of circuit change took place in the recent past. Later, this was considered a dead end. But I wonder...

I've been buried in projects, keeping me from being on top of this like I'd prefer to. I'll be more vigilant.

 

[ScottJ]

My initial impression, which our support guys seem to agree with, is that there's not a problem here.

There are a wide array of different "standards" for aligning analog and digital signal levels, but the more prevalent of them (and the one we follow) is sometimes called the "Sony Standard" or "20Ref." Under that rule, +4dBu = -20dBfs. Your posted results (0dBu = -23dBfs) are within 1dB of that relationship. There are no DIP switches or jumpers to vary this standard relationship.

Other standards put +4dBu (0VU) at anywhere from -12 to -24 dBfs. That would explain the varying results with, for example, your FireWire devices.

There has been no circuit change that we're aware of. The same USB circuitry is now used in our new, USB-equipped AIR-1 console.

If the consensus is that we should be providing a selectable reference here, I can certainly pass that along to Engineering.

 

[Jack Griffin]

Here's a really dumb question:

Did you go into the Windows Mixer and set the input level sliders to maximum?

On some computers Windows will set the levels at half scale when it installs new sound cards.

 

[grich]

My initial impression, which our support guys seem to agree with, is that there's not a problem here.

There are a wide array of different "standards" for aligning analog and digital signal levels, but the more prevalent of them (and the one we follow) is sometimes called the "Sony Standard" or "20Ref." Under that rule, +4dBu = -20dBfs. Your posted results (0dBu = -23dBfs) are within 1dB of that relationship. There are no DIP switches or jumpers to vary this standard relationship.

Other standards put +4dBu (0VU) at anywhere from -12 to -24 dBfs. That would explain the varying results with, for example, your FireWire devices.

There has been no circuit change that we're aware of. The same USB circuitry is now used in our new, USB-equipped AIR-1 console.

If the consensus is that we should be providing a selectable reference here, I can certainly pass that along to Engineering.

My day job (TV) has always used a plant reference of +4dBu = -20dBfs, even before we installed a Wheatnet IP system.

I wouldn't want to have only 12dB before hitting the wall. Headroom is good.

 

[KZME IT]

LazyJ: I've confirmed the dip switch settings...
http://print.metroeast.org/IT-temp/Wheatstone/AIR3/AIR-3_USB_DipSw.jpg
These are set to PGM2... I imagine that the surface mount resistors shown nearby, are responsible for padding the signal.


ScottJ: My initial conversations with support, were to ask if this was a standards issues. dBu vs. dBfs, but the response was that it is not. Is this an official reply from Wheatstone support?

We have professional DVcam VTRs which have selectable dBfs standards. I understand the desire for headroom when recording individual tracks, but I'm less convinced when recording a final mix. Also less useful when dealing with only 16bits... (The PCM2600 series is only 16bits/48k... fine for final mix at -12, but lousy for acquisition at -20.)

-6dBfs is half of the resolution of 0dBfs.. in 16bit that's going from 32768 peak to peak... to 16384 steps peak to peak. Drop that to -12dBfs, and you've got 1/4, or only 8192 steps to define the waveforms.

-20dBfs is less than 1/8 of the total scale. I would have less than 4000 steps to define the signal peak to peak.

When I have a nice loud signal, occasionally peaking in the red (+4 to +6 dBu), the signal to the CPU is so low... the resulting recording is terrible, particularly when compared with outboard solutions.


Jack Griffin / ScottJ: I've confirmed in both the System Preferences and Audio MIDI Setup (Mac OS), that there are no gain controls (which is how our other USB and Firewire A2D/D2A gear appears when attached... I wasn't expecting to find any.)

 

[KZME IT]

Wow... time ran out on my ability to edit... lame.

Sorry for the long bit about 16bit resolution... my final edit was less wordy. Didn't want to seem like I'm lecturing... Next time I'll edit offline. Didn't expect the short window for updates.

 

[KZME IT]

There has been no circuit change that we're aware of. The same USB circuitry is now used in our new, USB-equipped AIR-1 console.

If the consensus is that we should be providing a selectable reference here, I can certainly pass that along to Engineering.

The circuit change was a part of early speculation on the part of support... not my own.


I'd be surprised to find a bunch of consensus around recording 16bit at -20dBfs, for a final mix recording. Our TV plant/sister station has Sony DVcam VTRs, and uses -12 where a final mix is intended/expected.


Our TV plant is also aimed at -20, during acquisition... afterwards, everything is normalized/recorded at -12 for broadcast. The USB is aimed at providing a recording of the final mix, not channel discreet acquisition, so I would expect -12.

A switch might help someone expecting -20 at the USB/CPU, but who is that exactly? The USB doesn't provide Wheatstone IP connectivity. The AIR-3 has nothing to do with IP based audio transport. ???

 

[ScottJ]

ScottJ: My initial conversations with support, were to ask if this was a standards issues. dBu vs. dBfs, but the response was that it is not. Is this an official reply from Wheatstone support?

Yes, this is an official response. The standard is SMPTE RP.0155, which defines the alignment level for 0VU as -20dBFS. The corresponding European alignment level, per EBU R.68, is -18dBFS.

-6dBfs is half of the resolution of 0dBfs.. in 16bit that's going from 32768 peak to peak... to 16384 steps peak to peak. Drop that to -12dBfs, and you've got 1/4, or only 8192 steps to define the waveforms.

-20dBfs is less than 1/8 of the total scale. I would have less than 4000 steps to define the signal peak to peak.

Your discussion assumes that the signal level never exceeds 0VU, but unless you're recording an alignment tone, it always does. A signal reading 0VU on a meter with VU ballistics (~165 ms integration time) will often have peaks 6-10dB above that -- even analog circuitry is designed with 20 to 24dB of headroom to accommodate these peaks.

The standards both call for an MPL (Max Permissible Level) 9dB below 0dBFS. The standards organizations took note of the concerns you list and many others when instituting these standards, and we are simply following those established conventions.

 

[KZME IT]

Thank Scott. Much appreciated!

Not sure how to address the disappointingly low signal levels when recording... This wouldn't be an issue if the data was floating point or 24bit. Even with the internal meters 'pegged', I get a very low recording. And pegging the meters isn't the signal I want to send to the tech center/over the air.

Any ideas are appreciated.

Would you please address my question regarding playback, the ceiling I've observed on signals above -3dBfs?
(that is, -3, -2, -1 and 0 dBfs all exit the device at the same volume.)

 

[ScottJ]

Sorry, I missed that. Looking back at your original message, I found this:

After doing the recording tests, I tried a few sine waves out of the USB... and found that signals from -3dBu to 0dBu exit the USB output RJ45 at the same signal level?? (-4 and below, output levels with the expected decreases.)

...but I'm not sure I understand it. When you say "sine waves out of the USB," are you really talking about putting sinewaves into the console via the USB port? And are you describing signals sent to a USB port (digital signals) in terms of dBu? I don't mean to be obtuse, but the language is a little ambiguous, and I want to be absolutely sure I understand what your test setup was, in case I need to try to reproduce the results here.

 

[KZME IT]

Thanks Scott! Sorry to be light on details...

Here's what I did:

I've attached the USB's RJ45 analog output directly to gear that provides a dBu meter, with balanced StudioHub XLR adapters. Alternately, I've fed the output to a FireWire appliance (StudioHub TRS), observing the levels within the utility application that came with it.

In Audacity, using a 16bit linear 44.1kHz timeline... I generated 440kHz sine wave tones.
First at levels that read -24dBfs, -18dBfs, -12dBfs, -9dBfs, -6dBfs, -3dBfs, and then 0dBfs.

Everything looked normal, until the last tone, so I also generated -4, -2, and -1.

When the tones were at levels above -3, the results failed to line up...

-12dBfs tone resulted in -10dBu
-9dBfs shows -7dBu
-6dBfs shows -4dBu
-4dBfs shows -2dBu
-3dBfs shows -1dBu

and here things stop being consistent

-2dBfs shows -1dBu
-1dBfs shows -1dBu
0dBfs shows -1dBu


I've tried the same tones with a FW and a USB appliance, and see no 'ceiling' like I see here.

Not sure what to make of it, to be honest.

 

[KZME IT]

That should say 440Hz (not kHz)... (Doh!)

 

[KZME IT]

I wouldn't want to have only 12dB before hitting the wall. Headroom is good.

I completely agree, when mixing live sound for a band or similar sound support... But for general radio broadcast?

Everything I'm feeding this console has already been massaged by compression, CDs/LPs/automation tracks... nearly all of it is produced within a compressed dynamic range. I don't have huge peaks to contend with.

When we can afford to, our mic treatment will also include compression.

All that headroom makes more sense while recording master tracks for a session. We are not doing that.

I suspect if we were, we would have gone with gear that includes equalization, panning, trimming...
Things this unit does not do... and shouldn't.

An outboard utility mixer is used for live performance support.

 

[ScottJ]

-12dBfs tone resulted in -10dBu
-9dBfs shows -7dBu
-6dBfs shows -4dBu
-4dBfs shows -2dBu
-3dBfs shows -1dBu
-2dBfs shows -1dBu
-1dBfs shows -1dBu
0dBfs shows -1dBu

I'm puzzled, too. All of those levels look far too low to me. The readings near 0dBFS resemble clipping, which should not happen at such low levels. I'm going to try to reproduce your results with an Air-3 here at the factory using an analyzer and scope ... give me a bit of time and I'll let you know what I find out.

 

[KZME IT]

Awesome, much obliged!

 

[KZME IT]

All of those levels look far too low to me.

I'm not sure I agree about the level being too quiet in general. The level was almost exactly 1dB louder than a PreSonus AudioBox USB appliance. (The Abox column, below.)

The Focusrite Saffire FW appliance shows higher levels, but seeing the AudioBox and Saffire differ by 4dB can be explained by the balanced output being +4 vs 0dBu.

I took another stab at it using a meter with more resolution, and I have very similar results. Looks like my previous readings were off by about 1dB. These are all rounded to the nearest 1/2 dB, since a few show up with +/- .1 or .2 fractions. Mainly due to having only 2 significant digits for my tone amplitude.


src Saff AIR3 Abox
file FW USB USB
dBfs dBu dBu dBu
-9 -3 -6 -7
-6 0 -3 -4
-5 +1 -2 -3
-4 +2 -1 -2
-3 +3 0 -1
-2 +4 0 0
-1 +5 0 +1
0 +6 0 +2

The file amplitude equivalents I'm using are:
Amp = dBfs
1.0 = 0
.89 = -1
.79 = -2
.71 = -3
.63 = -4
.56 = -5
.50 = -6
.36 = -9

fairly accurate, but as I said only 2 digits.

 

[ScottJ]

Just an update -- I haven't forgotten about this. I had to wait a few days until the next group of Air-3 consoles came out of final test to snag one to work with. I'll do these tests in the next day or two.

Scott