Please give me insight on this problem with recording live shows and clipping!

[Kerrie]

Hi all,

Sorry if this is such an elementary question, but it has me scratching my head.

I have a A&H MixWiz that I use for little live sound gigs and recently I decided I wanted to start recording all tracks separately using the direct outs. So I purchased an Echo Audiofire12 and Sonar X1 to run on my laptop.

While the setup works perfectly, no problems actually recording, I'm confused as to the differences in metering on an analog frontend and digital daw.

Here's an example of what's happening:

Say I have a band I'm doing sound for. I have channels for kick, snare, overhead, bass, guitar 1, guitar 2, vocals.

When I'm getting the bands levels set up, I'd take each instrument PFL it, ask the musician to try and play the loudest note/patch/whatever that they would play in the set, and adjust the trim until the body of the signal is hitting zero on the A&H meter, with peak transients obviously going beyond that without clipping. (Except for a'la Dave Rat experiments where I have intentionally clipped the kick drum slightly - but that's for another discussion). Generally, my highest peaks my don't ever exceed +10dB - +14dB on the A&H main meter when I'm PFL'ing an individual track.

So in terms of gain staging for the live sound end of things, from mixer main outs > main eq > crossover > power amps > speakers... no problems there. And everything sounds great.

But now that I'm using the direct outs of the A&H, and recording each track with the audiofire12 into Sonar, every channel is clipping. I'm confused about the metering differences! On the MixWiz, the main meter is -30 to +16dB. In Sonar, 0dB (FS?) is the point of clipping. The recorded tracks are all ultra hot and clipping often.

It seems that the only way for Sonar not to clip when recording from the A&H is to set input levels on the A&H way lower than I normally would, where peaks are hitting around the 0 mark. But that doesn't seem correct to me.

Can someone shed some light on what's going on here? Do I need to somehow pad the inputs going into the audiofire/Sonar?

Thanks for any insight guys!
Kerrie

 

[mixsit]

If the nominal average out put levels of a mixer or mic pre and the input levels of you A/D converters by their design align' to the same (or similar) expected gain staging- your '0vu target on the mixer/pre will come in at around the -18 to -24 dB range on the RMS metering in Sonar. No problem.
But if one sends nice robust with lots of pro level' sig and the interface is built expecting/likes a lower signal, that's what happens - 'nominal on one is 'hot to the other.
Some (for example my RME ADI-8's) have input and output scaling options to expand and meet various situations like that.

Second example; Precision8 pre (-min gain 16 i think), typical kick drum mic, I have to either pad the mic, or the pre's out, or go to 'low gain on the ADI-8.

Hell, third example; Earthworks QTC-1 over a hot kit-- with no preamp what so ever can clip an A/D

 

[RawDepth]

Check that the Nominal Input Level (in the audiofire12 software) is set to +4dBu. If it is set to -10dBV, then it expects the incoming signal to be weaker and it will automatically boost the level.

 

[Bobbsy]

I think Rawdepth is on the right track, with the addition of another detail.

The Direct Outs on a Mixwiz can be set to pre or post fader with internal jumpers--but the default setting is pre fade. This means that the Direct Outs are feeding whatever levels you have set on the gain trims no matter what you do with the faders in the mix.

Since you're setting these to average at 0dB on the A&H meters in PFL, then letting peaks go where they want, if this signal is going into an Audiofire set for -10dBV, the cumulative effect is going to be pretty high.

If this sounds like it could be the issue, the obvious solutions would be to change the Audiofire to expect +4dBu plus setting up the gain staging on the Mixwiz with slightly more conservative levels. Dave Rat aside, I like the PFL meter reading to be PEAKING at 0 on the meters or, at most, only slightly above. This lets me run my channel faders near the nice, linear zero setting without the total mix (which obviously adds all the channels) eating too far into my headroom.

 

[Kerrie]

Thanks so much for the insight guys! I don't have my recording system here in front of me but I'll check on the operating level of the audiofire tonite.

As far as the PFL's being prefade... yes, that's exactly what I'm wanting. I need the raw track information so I can mix things later. Regarding the direct output, by default on the MixWiz3 (I bought it used with no manual), are the direct outs pre-everything? As in pre-EQ and also pre-insert? I'm hoping yes, and if not, is there jumpers to set them pre-everything?

As far as getting PFL signal goes... wow, I must have read some bad information. I never have thought to have my peaks hit 0dB? The resulting signal seems weak. What's the purpose of the metering going to +16dB before clip on the main meters?

Also, I've read that on individual channels, when setting individual gains I should raise the trim until the little channel peak light blinks intermittently, and then roll it back a bit. That wouldn't be the case if I was setting tracks to peak at 0dB.

Now I'm so confused!

 

[bouldersoundguy]

As far as getting PFL signal goes... wow, I must have read some bad information. I never have thought to have my peaks hit 0dB? The resulting signal seems weak. What's the purpose of the metering going to +16dB before clip on the main meters?

Every time you double the channel count you add approximately 3dB. If you have sixteen channels peaking at +6dBVU they will add to about +18dBVU. Pan law probably takes that down 3dB putting everything right at +15dBVU. What I've read (in the manual) is that it will actually put out as much as +26dBu (which would be +22dBVU on the meter if it went that high) without clipping.

The MixWizard provides metering of inputs and outputs.
For best results operate the console with the main meters averaging around ‘0’
allowing the loudest moments to reach ‘+6’. Reduce the channel gain settings if
the red peak indicators start to flash. Note that the peak indicators light 5dB
before actual clipping to warn that you are nearing distortion and should reduce
gain.

 

[mixsit]

...Also, I've read that on individual channels, when setting individual gains I should raise the trim until the little channel peak light blinks intermittently, and then roll it back a bit. That wouldn't be the case if I was setting tracks to peak at 0dB.

Now I'm so confused!

Here's the deal. The mixer's ch. gain set-ups are to maximize their general opperating range (that's where the 'set them for nominal 'zero on their scale comes from- 'Nominal + your ch. head room). Then the mixers internal bus sturcture typically is designed to accommodate all or most of the mixer's channels, and come out with decent safe levels and head room- now at the channels and the main bus.
You have some wiggle room..

I like the PFL meter reading to be PEAKING at 0 on the meters or, at most, only slightly above. This lets me run my channel faders near the nice, linear zero setting without the total mix (which obviously adds all the channels) eating too far into my headroom.

Doing this gives you a bit lower general opperating levels in the mixer, you've built in some more head room- maybe the mixer runs cleaner..(? don't know specifically here), and is a way to accomidate your interface input needs if that's whats needed as well.

 

[Kerrie]

Ok thanks for the additional information. But lets just put my interface clipping issue aside. In general practice, on that MixWiz3, should I be doing this:

I like the PFL meter reading to be PEAKING at 0 on the meters or, at most, only slightly above. This lets me run my channel faders near the nice, linear zero setting without the total mix (which obviously adds all the channels) eating too far into my headroom.

or this:

The MixWizard provides metering of inputs and outputs. For best results operate the console with the main meters averaging around ‘0’ allowing the loudest moments to reach ‘+6’. Reduce the channel gain settings if the red peak indicators start to flash. Note that the peak indicators light 5dB before actual clipping to warn that you are nearing distortion and should reduce gain.

because I'm currently doing this (and everything seems to sound great, and gainstages fine down the chain):

adjust the trim until the body of the signal is hitting zero on the A&H meter, with peak transients obviously going beyond that without clipping. (however some of my peaks can go up pretty high on percussive instruments, but never clip - additionally, I'm doing my PFL with channel inserts like comp/limiter bypassed)

 

[bouldersoundguy]

But lets just put my interface clipping issue aside.

The do what the manual says. I suspect that combined with the right sensitivity setting on your interface will make everything line up properly.

 

[PRHunt]

Seems to me that the "simple" solution might be to use pads between the A&H and the Audiofire.

The AF12 has a specified maximum analog input of +17.5dBu. From your description, you must be approaching or exceeding that with the direct outs from the A&H. If you had a pad of say -14dB, you'd tame the signals to a nicer +3.5dBu. This would give you a nice clean recording with a safe margin of headroom.

If you were using XLR connectors, this would be easy. You'd just plug in an inline pad on each cable. Common and inexpensive.

However, I have only been able to find one source of TRS pads and that is from True Systems, which provide a TRS adapter with a -15dB pad. TRUE Systems No idea on the price of these. You would need one for each channel, and plug it at the A&H end, as I think space is a little cramped on the Audiofire. (At least this is the case on my Audiofire 8, so I have to use narrow TRS plugs).

The alternative would be to build up a plug in box to do the same job. For the AF12, you'd need 24 TRS chassis sockets, some stripboard, resistors to make the pad, and metal box. All you need to know about pads is right here: Uneeda Audio - Build your own attenuator pads. Easy to do, if a little fiddly.

This might let you retain your current workflow while getting better results from the AF12.

Paul

 

[RawDepth]

Kerrie,

Perhaps some schooling is in order here...

As you have found, metering is different from analog to digital. When analog gear is designed, they set the "zero" mark on the meter at the loudest point the gear can be operated without adding measurable distortion to the signal. As the signal level increases above 0dB, (by a few dB) some amount of distortion slowly creeps into the signal. The higher you go above 0dB, the greater (more noticeable) the distortion becomes. At the highest point on the meter, (maybe +18dB,) hard clipping of the waveform occurs and things really start to get ugly. So you see, it is somewhat forgiving at first and the user has a certain amount of headroom to flirt with. If you want your PA system to sound its absolute cleanest, get the signal as close to 0dB as you can without going over.

With digital gear, the signal stays perfectly clean all the way up to 0dB where hard clipping occurs.

As a comparison of signal strength between the two, 0dB in analog is equal to roughly -18dB on a digital meter.

It is up to you how you want to operate your mixer. As long as you understand the headroom issues and the consequences of going too far. Speakers don't like distortion, as it is usually what causes them to get hot and fail. If you have adequate speakers and amps, you shouldn't need to push the mixer above 0.

Hope this helps.

 

[bouldersoundguy]

Kerrie,

Perhaps some schooling is in order here...

I trust Carey Davies' word when it comes to setting gain on a MixWizard.

Distortion is often associated with speaker damage but it isn't the cause. Power = heat, regardless of the shape of the wave.

 

[mixsit]

Ok thanks for the additional information. But lets just put my interface clipping issue aside. In general practice, on that MixWiz3, should I be doing this:



or this:...):

What's the difference between this..
"The MixWizard provides metering of inputs and outputs. For best results operate the console with the main meters averaging around ‘0’ allowing the loudest moments to reach ‘+6’. Reduce the channel gain settings if the red peak indicators start to flash. Note that the peak indicators light 5dB before actual clipping to warn that you are nearing distortion and should reduce gain."

and this..?

"..adjust the trim until the body of the signal is hitting zero on the A&H meter, with peak transients obviously going beyond that without clipping. (however some of my peaks can go up pretty high on percussive instruments, but never clip "

Does 'MixWiz leave out the part about setting (each) ch. 'body/average' around '0, a bit lower if they peak at the channel? Given just what you quoted implies you wouldn't know what's going to happen at the main bus untill all the ch's are up and running.

 

[mixsit]

Ok, what I wrote above is open to misunderstanding.
They both say the same methodology; 'body of the signal is hitting zero, peaks going beyond that without clipping', just one is speaking of the mains, the other back at the channels.

 

[RawDepth]

I trust Carey Davies' word when it comes to setting gain on a MixWizard.

Distortion is often associated with speaker damage but it isn't the cause. Power = heat, regardless of the shape of the wave.

A square wave indeed causes more heat buildup in a speaker than a correct wave. (But that discussion is for another thread. Let's not hijack this one.)

 

[Bobbsy]

I wonder if musical styles might have something to do with the set up advice. When I set up the channel gain trims, I want the "normal" signal to be at 0dBu but allow very occasional peaks to go higher--and on the style of stuff I'm mixing there are indeed very occasional peaks. On some other music though, the "peaks" might be almost constant--and that's the situation where it can get you in trouble because "constant peaks" become the "average level". Am I making sense?

Far be it from me to say A&H have it slightly wrong but that's exactly what I'm going to do. Their set up advice doesn't make sense to me. If every channel is operating normally around +6 then, once they're all summed you'll just have to pull down the masters or the output of the mixer will be into clipping (as several of us have pointed out). I prefer to have my mixer set up so, as a starting point for the mix, every fader can be near it zero point. Why? Because the faders are at their most linear there meaning I can do nice subtle adjustments. If you have to run with every fader down lower, each movement makes a much more dramatic difference to the sound.

In any case, Mix Wizards are lovely quiet mixers so I see no reason to compromise my headroom to get 90dB S/N instead of 85dB.

Anyhow, the first thing is to check the input setting on the Audiofire (which is switchable -10/+4. But I still think you have room to run your mixer at a slightly lower level and preserve a bit more headroom.

 

[ecc83]

I had essentially the same problem with my ZED10 and a 2496.

The soundcard has a maximum input of +2dBu (about 1V rms) and to keep the average DAW level at the required -18dBFS I had to keep the ZED's meters at a pretty useless -20 -30 "vu".

I followed Paul's suggestion and fitted attenuators at the soundcard end of the line but I used multiturn presets so that I could fine trim the system. This gave me back the Z10 meters with 0vu now corresponding to -18dBFS.....However! Son was not happy! He said he was finding it difficult to get a big enough "picture" of the wave so the attenuators were bypassed and he ran "seat of pants" on levels.

Fortunately this was possible in our situation with just one part recorded at a time and where he could watch the DAW levels. He was STILL tracking to hot for my liking, but what can you do??!!!

Dave.

 

[PRHunt]

Still think that some attenuation should be tried between mixer and Audiofire. The Audiofire requires peak inputs of no more than +5.5dBu to provide safe digital headroom of 12dB. 0dBFS (ie full scale analog input before clipping) on the Audiofire is specified at 17.5dBu.

I'm not sure which Mixwiz is being used, but I've had a glance through the pdf manual for the 16:2. 0dB on the meter on the A&H corresponds to a signal level of +4dBu. If peaks on PFL are reaching +14dB, then you should expect clipping at the Audiofire as you would be hitting it with +18dBu. Furthermore, the direct outs from the A&H are specified for a maximum of 21dBu output, so there is definitely potential to clip the Audiofire inputs.

The best solution would probably be to run quality splitters from each source, and to operate the live console and the recording hardware as completely independent systems. The advantage would be that the live console and recording systems could each be optimally gain-staged. The cost and increased complexity, however, will be limiting factors for home/hobby recordists.

So, for practical solutions: either 1) tame the gain levels on the A&H, or 2) buy or build pads to insert between the A&H and the Audiofire.

Paul

 

[ecc83]

There should be no compromise of noise or headroom if using attenutators.
The actual signal presented to the input of the A/D converter is not likely to exceed 5V peak to peak (1.76V rms) so considerable signal voltage scaling is already being employed in any AI.

Dave.

 

[Bobbsy]

The Audiofire is software configurable to use a nominal input level of either -10dBV or +4dBu. With the nominal level at +4 and a maximum level of +17.5dBu, it should be a good match for the Mix Wizard even without attenuators (unless the mixer is being run at much hotter levels than I'd recommend anyway.

Just because the Mix Wiz has extremely good headroom, there's no need to use every bit of it as a matter of course. The signals all have to go somewhere (be it the Audiofire, to an effects unit or just to an amp so you always have to be aware of the levels, both on individual channels and on the final mix.

...unless you're a DJ of course, then just shove everything into the red and go for a beer!