Multitrack Reality - A question and a report back

[ullanta]

INPUT clipping - which (I was surprised to find) is specified as -10dBu on the MixPre (some documents say 0dBu with Phantom on, some don't - I think that's the pre-revised model). That's pretty low! There's no clear spec on input clipping level for the OctoPre that I can find - what I've seen is not clearly input or output headroom, but I think it's output. But, certainly a lot higher than the MixPre! (and, for comparison, definitely lower than my Millennia's +23dBu input clipping - but I don't think there's much out there that tops or matches the Millennia in that regard).

I was micing a grand piano (outside the lid, about 5 feet from the sounboard) with THE KR-1F omnis, which are more sensitive than Oktavas - but have never overloaded anything before (I've used 'em in similar configurations with everything from an MBox to my Millennias, most frequently with the OctoPre).

I was pretty surprised too, then went back for a look at the specs. What's interesting is that the MixPre is spec'd for such low input headroom but such high output headroom. Very low levels on the meters, but clip city. And, of course, the limiter actually handles the high levels better than I would expect, which may mask the effect for other loud sources where dynamics are not so clear and essential. But with the limiter off... nasty!

I'm not used to carrying pads around - I guess I'll have to start!

 

[Noiz2]

-10 is VERY hot for a mic in, thats consumer line level. Well not exactly since that is -10 dBv I belive not dBu. Output clipping is at +22 dBu and that's what I had been looking at. The OctoPres limiter is fixed at +20 and and it clips at +22 (out put not input). There is no input level spec that I can find so???
I'm not saying your wrong I'm just very surprised because of how I and a number of other folks use them and our different experiences.

 

[ullanta]

Hmmm... I'm looking around for specs on different pieces of gear..

Millennia HV-3: Maximum Mic Input Level: +23 dBu
Mackie Onyx: Maximum Input Levels Mic Input: +21 dBu
Lexicon Omega: Microphone Inputs Maximum Input Level: -2 dBu (150 Ohm source impedance)
M-Audio Firewire 1814: mic inputs maximum input level -3.8 dBu, typical
Fostex FR2LE Mic inputs: Maximum Input Level +2dBu
Presonus ADL600 Maximum Input Level: Microphone (1500Ohm, +20dB Pad out) +5dBu
Zaxcom DEVA: Input Clipping Level: +28 dBu
SD MM-1: Input Clipping Level (Mic) +4 dBu minimum at the 0 or +18 dB gain setting

... so, the MixPre seems to fall low on the low end...?

I'd love to be wrong!

The OctoPres limiter is fixed at +20 and and it clips at +22 (out put not input). There is no input level spec that I can find so???

Right - what I meant was that I empirically place the OctoPre's input clip level as much closer to the Millennia than the MixPre.

 

[venuestudios]

I would like to chime in on this. First I know your feelings on the Onyx, and there should be no reason to get away from it. You are correct that a limiter on the insert will not stop the pre's from clipping. But you can keep the gain at the board lower and use the make up gain from the limiter to add as you need. But why a limiter? I would reach for a compressor first. You can run at 6:1 ratio, higher if needed, and should be able to have great control. It sounds like you are most concerned with comics mic, which is a single channel. Are you only bringing in 3 channels? Did I miss something in the channel count? FMR makes a nice little comp the FMR RNC, two channels and very clean. For location work it is small (1/2 rack).

 

[Noiz2]

Well, I'll chime in on why I would choose a limiter. When I'm recording I like having that safety valve for when I get surprised. I don't want any "gain riding" which is essentially what the compressor is doing, I don't want to reduce my dynamic range. I just want to keep from blowing out when the actor gets excited and slams the door when he was supposed to close it or more often in my case when some FX is much louder in reality than in the "test run". On run 2 I'll have my levels lower but at least #1 is usable in a pinch. It's really hard to predict how loud an F-15 going supersonic a couple of hundred feet overhead is going to be other than LOUD. It's a warm and fuzzy feeling to know that unless the mic blew the recording will be usable. At 6:1 the compressor would have to effect just about everything I record to be "clip proof".

 

[timapter]

thanks noiz and ullanta and venuestudios...

venue, as noiz said - i want to keep my dynamic range as high as possible going to the recorder so that post has more options with cleanup/compression etc - i dont want to be doing that stuff on location, i just want to be sure that i dont get any clips.

Channel wise, im just as concerned about the judges actually - they are comics themselves so they are quite dynamic too! track count is 1 for the comic, 3 judges, 2 presenters.

my sd302 is providing some output limiting on the 2tk mix that is going to camera (which Im going to attempt to make the master mix so they dont need the iso's in post), but if I screw up the multitrack is there to catch me....

 

[venuestudios]

F15's and loud voices are certainly at different ends of the spectrum! I can understand the limiter on the jet. Chances are the dynamic range between coming at you and overhead is a pretty scary swing for any mic. As far as vocals go I fall back on my live sound work and can't think of many times I used a limiter to tame a vocal. As you probably know the only difference between the comp and lim is the ratio (all things being equal). I personally would rather gently compress a vocal than have it slam into a brick wall.
I don't know of a right answer for every situation. Just offering up options. With the comp you have anything from a gentle reduction to a hard limit in the same box.
I have just started working in video so it is interesting to see how the film industry handles sound as compared to the live concert folks like myself. I run up to 48 channels to a multi-track for recording and don't put anything on the inputs. I set levels ahead of the concert and go from there. Any dynamics are handled in post. I split everything at the stage so I don't have to worry about front of house too much. Film seems to have a whole new set of obsticles to jump over.

 

[Noiz2]

Ah, there is also a difference with live VS recorded. For live I would not be very likely to use a limiter, but if there is a little distortion then you pull back on the channel, maybe a few people notice - maybe, and it's soft (ish) analog distortion. Production recording people are going to scrutinize at there leisure and your recording is digital so your overs are REALLY nasty. Also a limiter "slamming" the track is not nearly the issue with dialog as it is with music. You don't have the continuity that you have with music so it's much less noticeable.

 

[Ted Spencer]

Hmmm... I'm looking around for specs on different pieces of gear..

Millennia HV-3: Maximum Mic Input Level: +23 dBu
Mackie Onyx: Maximum Input Levels Mic Input: +21 dBu
Lexicon Omega: Microphone Inputs Maximum Input Level: -2 dBu (150 Ohm source impedance)
M-Audio Firewire 1814: mic inputs maximum input level -3.8 dBu, typical
Fostex FR2LE Mic inputs: Maximum Input Level +2dBu
Presonus ADL600 Maximum Input Level: Microphone (1500Ohm, +20dB Pad out) +5dBu
Zaxcom DEVA: Input Clipping Level: +28 dBu
SD MM-1: Input Clipping Level (Mic) +4 dBu minimum at the 0 or +18 dB gain setting

... so, the MixPre seems to fall low on the low end...?

I'd love to be wrong!

Specs like those of the Millenia (+23) and Onyx (+21) are maximum *line* levels, not mic levels, even if the mic input socket is being used for it. The preamp circuitry in such devices would obviously have massive amounts of gain reduction built in (in effect, a pad) so that line level signals could be handled.

In use on average sources like semi-close miked voices for example, most microphones put out a tiny fraction of line level, around -45 or so typically. They rely on a preamp to add the 40 or 50 db of gain to bring it up so it can be fed to a line level device like a recorder.

A mic pre that lists -10 as a clip point probably just has less gain reduction capability built into its circuitry, and would be closer to a simple mic-only device. The only time a mic might get that hot is in the case of something like a close drum mic, like right on a rock snare drum for example. For your scenario it shouldn't be an issue.

Almost any modern mic preamp equipped with a gain control "trim pot" (and ideally, but not necessarily, a pad) is capable of handling almost anything any mic can put out. Many condenser mics also have built-in pads that can make the preamp's job easier.

Recording set-sound or club sound (except for bands, which are usually pretty consistently loud), involves a VERY wide dynamic range. I personally advocate simply keeping record levels low enough to be safe. Limiters are a safety net at best, and can mess thing up if they get hit too hard/often.

 

[ullanta]

Sorry, Ted, the Milennia spec, at least, is native mic input headroom. It doesn't do line, has a simple, clean signal path without any of what you're suggesting. I record live opera, and on rare occasions where a soprano ends up singing inches from the mic, you have an SPL that'll drive most mics to overload most preamps. That's why I use the Millennias.

And indeed, the specs above that I listed and you quote are all listed by manufacturers specifically as mic level, and without pads engaged. PEAK output levels for condenser mics are typically much higher than the nominal levels (-50/-60), and often rise well above 0 dBu in instrument or close voice micing.

Anyway, the empirical evidence remains: the MixPre displays input clipping at a much lower level than anything else I've used, which is something people should be wary of if they plan to use any of the SD mixers for musical purposes. Typical piano setups, with sensitive mics, will require external pads; I'd imagine the same to be true certainly of percussion, and many other instruments as well. This is especially important to live classical recordists, 'cause if it surprises you in a long piece, it may be 40 minutes before you have a chance to stop recording and add pads to your signal chain. For music, I'd suggest ALWAYS keeping switchable pads in the signal chain unless you have a really thorough souind check, or are distant-micing.

 

[Noiz2]

Not dismissing your experience but that has not at all been my experience. Poss there is something wrong with the MixPre you were using? I have used the MixPre to record Extremely loud sounds and classical music, though that was ~~mid miced. Not right in the belly of the beast. And percussion at close range. Now maybe it's the limiter that is saving things but I'm rarely seeing a hit and it's set all the way out so?

As to input levels I think we may all be a victim of spec wars. For instance on the SD 302 it lists 0dBu as input clipping min (trim down) but they also list the input limiter as having a threshold of +18dBu. Those can't both be right I don't believe. So???
I looked up Schoeps max out put and it's listed as 1V max. That should translate to a bit over 0dBu. But of course the problem with these things is that they tend to all measure things a bit differently or name them differently from each other so comparing specs is often useless. SD also lists one spec as being 20dBu (would that be + or -? probably + but ...). So unless your MixPre is broken we can probably say that the Octopre is going to be even more robust and sound great, which is probably more to the point in this thread. If it had a FW out option I would be shopping right now. When I get back home (I'm off recording the Dali Lama :~)), I'll try to crap out the MixPre with the limiters off and see if they are making the difference in our experiences.

 

[Ted Spencer]

(in response to Ullanta)

Well I suppose we might go round and round on this forever, but the reason I referred to the +23 Millenia spec as line level, is that for most pro audio devices that's pretty close to the *overall* headroom at the output stage. An SSL console's headroom for example, is spec'd at +26 throughout. It's been right around that number on line level pro audio devices for decades.

Analog audio gear is typically spec'd for hard, continuous clipping as well, by the way...like with a 1 KHz sine wave...not the instantaneous peak-style clipping that digital devices normally deal in. We're talking about clipping as hard, full-on breakup.

Nominal pro audio line level is +4dbm, or about 1.23 volts. I'm not sure exactly, but I believe +23 would be roughly 10 times that amount, or about 12 volts! There aren't any mics I know of that can cleanly put out a volt continuously, let alone 12.

Given that the Millenia has a gain knob (as it must), if it can take +23 at a mic input, and a mic pre in normal use might be providing ~25-50db of *gain*, what it tells me is that with the gain control all the way down, eliminating almost *all* the amplification, it would then have a clip point of +23. In that circumstance you really don't have a mic pre. You have a (near) unity gain line amp.

I noticed on the Millenia HV3R (8 channel version) that the gain control has 65 1db steps, and that "minimum gain" is 8 db. So, at -65, you've eliminated all but 8db of the mic pre's gain, effectively taking it all but out of the picture. If you factor out the 8db minimum gain, which in effect reduces the input headroom by the same amount, you'd get a total output headroom of +31, which is outstanding, but Millenia products do have exceptionally low distortion at very high gain levels. I had a Millenia "Origin" on loan for a while - those puppies are *cleeeaan*.

So what that means, is that they designed the HV3's input gain/attenuation knob with enough range that it can handle line inputs when necessary. And that the line level circuitry, which is everything in the chain after the mic preamp section itself, most particularly the output amps, have ~+31db headroom. This would be on the high end of industry standards, as would befit a $2000 stereo mic pre.

But it doesn't mean that any mic, ever, is expected to put out +23, or anything remotely close to that.

The MixPre's seemingly "low" clip point of -10 is probably just a reflection of its gain structure - its input gain control at minimum would result in clipping when fed a -10 signal. As others have pointed out (again, in analog audio terms) that is *way* hot for a mic - much hotter (again, in analog audio spec terms) than you should ever have to worry about. The MixPre is presumably a film/video mic pre *only*, and isn't set up to handle any line sources, so higher headroom/gain reduction isn't needed.

The last point to make is regarding the fact that, as you said, "the MixPre displays input clipping at a much lower level than anything else I've used". This can simply be a byproduct of conservative clip indicator design. It may be showing the kind of momentary clips that are not terribly important in reality.

On SSL consoles equipped with LCD meters, setting levels using VU mode and then switching to peak mode can make levels look insanely hot when going to analog tape, but they're not. VU is the better of the two, and is what should be relied on in that situation.

 

[ullanta]

Noiz - Yes, specs are specs, and I hope I'm clear that my main point here is the empirical result (I mic pianos almost daily, and use a variety of mics and pres, so I have a wealth of comparative experience - I've probably used the same THE mics and similar placement 300 times with various preamps).

And yes, the limiter on the MixPre is great as limiters go, and for most non-musical (and some musical) purposes solves any issue of the input level nicely - with the limiter on, the main issue is just, of course, flattened dynamics (specifically, with the trim at close to 0, the line of peaks was flattened at about -15db by the limiter - the METERS barely brushing the 1st red LED even though the limit LEDs were on frequently, and close to constantly in some sections - the final waveform looked like a pop song (pretty constant level) rather than a classical piece).

Also - I've used the MixPre/THE combo on piano CONCERTO (~10 feet from the soundboard IIRC) and piano/violin (5' from violin, 8' from soundboard) without this being a problem. But my typical setup for nicest piano solo sound (3-5' off the soundboard on a line bisecting the lid angle) gets into the problem range.

So yes, please check it out with your MixPre, especially with the limiter off! Do you have a piano around?

 

[Ted Spencer]

..."with the trim at close to 0"...

You mean the MixPre's mic gain pot? If so, just turn the sucker down. If the limiters are flattening by 15db you have *way* too much level hitting them. Is there some reason why you don't just back off the mic gain on the MixPre?

 

[ullanta]

(in response to Ullanta)
I noticed on the Millenia HV3R (8 channel version) that the gain control has 65 1db steps, and that "minimum gain" is 8 db. So, at -65, you've eliminated all but 8db of the mic pre's gain, effectively taking it all but out of the picture. If you factor out the 8db minimum gain, which in effect reduces the input headroom by the same amount, you'd get a total output headroom of +31, which is outstanding, but Millenia products do have exceptionally low distortion at very high gain levels. I had a Millenia "Origin" on loan for a while - those puppies are *cleeeaan*.

Yes... Millennia specs output headroom at +32dBu.

So what that means, is that they designed the HV3's input gain/attenuation knob with enough range that it can handle line inputs when necessary....

But it doesn't mean that any mic, ever, is expected to put out +23, or anything remotely close to that.

And no... as explained at http://www.mil-media.com/design.html and other papers, the Millennia is designed to handle hot mic levels, which commonly reach the levels you claim they don't. As you note, many preamps and mics have pads/attenuators to be able to handle high levels... because most mics, in some close-micing situations, can indeed put out a level that will overload most preamps.

The MixPre's seemingly "low" clip point of -10 is probably just a reflection of its gain structure - its input gain control at minimum would result in clipping when fed a -10 signal. As others have pointed out (again, in analog audio terms) that is *way* hot for a mic - much hotter (again, in analog audio spec terms) than you should ever have to worry about.

This simply isn't true. -10dBu is not particularly hot for the peak output of a condenser mic. Again, the commonality of pads on mics and preamps shows the expectation that mic levels can often get too hot for a preamp to handle.

The last point to make is regarding the fact that, as you said, "the MixPre displays input clipping at a much lower level than anything else I've used". This can simply be a byproduct of conservative clip indicator design. It may be showing the kind of momentary clips that are not terribly important in reality.

Perhaps my choice of words is wrong - by "displays" I meant "exhibits the behavior of", not "lights up an LED". I mean the sound sounds squashed, the resultant waveform is flattened. The dynamics are clearly limited. Without the limiter, the sound was unlistenable, clearly clipping quite nastily even when the meters were nowhere near the output clipping level.

Ted... we can talk specs all we want, but I've been in the exact same micing situation hundreds of times. The MixPre (without pads) couldn't handle the situation. Every other preamp I've tried (without pads) has handled the situation.

 

[ullanta]

..."with the trim at close to 0"...

You mean the MixPre's mic gain pot? If so, just turn the sucker down. If the limiters are flattening by 15db you have *way* too much level hitting them. Is there some reason why you don't just back off the mic gain on the MixPre?

The mix pre has 2-stage limiting. The signal was being squashed by the input limiter, which acts BEFORE the gain pot. I did try with the gain fully off (no difference except the output/metered level was lower - that is, quieter overall sound with quiter nasty clipping noise or similarly squashed dynamics), and probably could have recorded with the gain up another 15dB if I wanted the hottest output signal. But the problem was the input stage which, short of adding pads before the MixPre, I could do nothing about.

 

[Ted Spencer]

The mix pre has 2-stage limiting. The signal was being squashed by the input limiter, which acts BEFORE the gain pot. I did try with the gain fully off (no difference except the output/metered level was lower - that is, quieter overall sound with quiter nasty clipping noise or similarly squashed dynamics), and probably could have recorded with the gain up another 15dB if I wanted the hottest output signal. But the problem was the input stage which, short of adding pads before the MixPre, I could do nothing about.

The product literature's a little ambiguous, but if the adjustable limiter's threshold can't be raised pre-preamp, leaving you with a choice of clipping or excess limiting, then it would seem you have only the following options:

Use pads (on the mics if available, or inline)
Use different mics (which will change the sound a lot)
Back off the mics (which will change the sound a lot)
Use a different pre (which will change the sound somewhat)

Of the list, given all you've reported, it seems the last may be the best. This device just may not be set up for handling the mics you've got. Presumably, those KR-1Fs are pretty darn hot.

I used a Sound Devices mixer on a shoot with a pair of lavs and a shotgun and had no level issues. We were recording a couple of guys singing in a bar over a CD music track played back on a guitar amp. It was fairly loud, but I had no problem getting clean signals through the box. It was the 302, a 3 channel model. We were outputting line level to an HVX200.

By the way, one of the only mics I could find maximum output level specs on was a Neumann U87. It was 390 mv, or about -8 DbV, which, OK, is a hair above -10. I believe this would be its clipping point, so...+23? Don't think so, not in real life, Millenia's website notwithstanding...

 

[ullanta]

http://www.rane.com/note148.html has a nice chart relating sensitivity (more commonly available than max mic output) to output in dBu. For loud sources, even fairly low-sensitivity mics will break the 0dBu barrier...

When I get a chance, I'll bring my meter along and try to measure SPL around a grand piano... DPA estimates "> 130 dB SPL" at 20cm; I'll try to chack at my typical distances.

 

[Ted Spencer]

http://www.rane.com/note148.html has a nice chart relating sensitivity (more commonly available than max mic output) to output in dBu. For loud sources, even fairly low-sensitivity mics will break the 0dBu barrier...

I don't think so...

I just did a little test here. I connected a Shure SM57, which is admittedly not a very hot mic, but is one that will take enormous SPL levels cleanly, and it's a dynamic, so no clipping electronics would artificially limit its output, to a fixed, nominal -10 dbv (about -8 dbM) line input on a mixing console. Holdng the mic only about 1" from my mouth, I shouted "hey" into it at the top of my voice (which still hurts a bit from it) and read the level on the input meter, which is a semi-peak reading type (about halfway between peak and VU). The level peaked at -15, which means -25 when the operating level is taken into account, or about -23 dbm. I then did the same into my SPL meter, and got a reading of about 120db. That is VERY VERY LOUD, as expected (I was pretty concerned the neighbors might call the cops!).

But it only resulted in -23 dbm output from the 57. Kinda backs my side of the argument...er...discussion : ) ...pretty well wouldn't you say?

When I get a chance, I'll bring my meter along and try to measure SPL around a grand piano... DPA estimates "> 130 dB SPL" at 20cm; I'll try to chack at my typical distances.

I just did it for you. I played a nice loud C chord triad with two c octaves in the left hand on my Yamaha grand with the SPL meter sitting on the piano bench next to me, which is of course, fairly close. It read about 94 db. From about 20 cm above the strings (lid open) a C triad only (I needed a hand to hold the meter!) read about 102. I suppose if you used a super-fast peak reading meter you might see instantaneous readings substantially higher, but remember that sustained 130 db SPL is basically deafening, literally, in a very short time.

Most SPL charts show 130 db as equivalent to a jet plane takeoff from fairly close by, loud rock concerts at 120 and so on. A piano at 20 cm making 130? Uh...nah...

 

[ullanta]

But it only resulted in -23 dbm output from the 57. Kinda backs my side of the argument...er...discussion : ) ...pretty well wouldn't you say?

Umm... dude... not at all. Seriously, an SM57? Try the same with a C414... An SM57 has a sensitivity of about 1.6mV/Pa; my KR-1Fs are about 12 mV/Pa; a large diaphragm mic like a C414 puts out around 23 mV/Pa. A low-output dynamic is NOT a useful tidbit of information here!

I just did it for you. I played a nice loud C chord triad with two c octaves in the left hand on my Yamaha grand with the SPL meter sitting on the piano bench next to me, which is of course, fairly close. It read about 94 db. From about 20 cm above the strings (lid open) a C triad only (I needed a hand to hold the meter!) read about 102. I suppose if you used a super-fast peak reading meter you might see instantaneous readings substantially higher, but remember that sustained 130 db SPL is basically deafening, literally, in a very short time.

Most SPL charts show 130 db as equivalent to a jet plane takeoff from fairly close by, loud rock concerts at 120 and so on. A piano at 20 cm making 130? Uh...nah...

Sorry again, but this is not useful. Can you play some Schumann at concert level? Especially if you're holding a meter with one hand, you can be sure you're not generating anywhere near the max level of the instrument. Distance matters... a lot. The DPA 130 dB measurement should be pretty reliable, they're not idiots.