Sample video (NYC subway performers):
http://www.palsomedia.com/canon5d2/n...ground_5d2.wmv
(download and save)

The first part part is with the built-in mic.
The rest are with the MKE 400.

It's very UNlikely these cameras have AGC. There will be no pumping. This is a worry that has been an audio myth for several decades.

The cameras most likely have a limiter that prevents clipping. Which means no signal can exceed 0dB. But, most likely peaks are kept below -12dB. This is exactly what you MUST do with digital -- except a limiter can do it faster than you can.

With digital audio that's all one needs as long as you use an external mic that is matched. Matching is something you do BEFORE you going out shooting. If you know your mic will not drive the recording into distortion, you are set.

YOU are where the audio IS. If you can hear it at the camera, you will be recording it. If you can't hear it -- forget about it.

If you want to hear audio QUALITY -- do the XLR boxes will have an earphone jack? If not, if the audio is output via the AV jack during recording -- you can use that signal with a tiny amp. (One IC and a battery.)

There is no Confidence playback for video. If you think audio might not make it from the audio jack or internal mic to the record heads -- then you should be equally worried about video. Yet today, no one worries about this. Plus, you certainly want to make a test recording.

PS: If you notice -- most of the posted movies don't use on-location audio. Or, very little. For ENG, the audio should be more than adequate.

You can have a bad connection between your mixer and the mic input jack. I think that's what might worry most people, especially since it is an 1/8" connector - not exactly robust. In addition to cable failure, you might forget to connect it or it might inadvertently pull out. If there were visual audio level indicators on the lcd screen of the camera I would feel more comfortable not being able to monitor directly off the camera. Are there visual audio level indicators?

Huh? A true limiter is only preventing clipping at the very top of the scale. How's the gain set then? How does the camera "keep levels below -12?" Because somewhere, somehow it sets the gain -- the point being, you don't set gain, it does. Now sure, maybe it won't obviously pump, but surely in a loud environment it lowers the gain (doesn't seem like it could just fully depend on a limiter without sounding like crap in an extremely loud environment) and in a quiet environment it raises the gain? Without manual gain control, something is setting levels, and I find it hard to imagine there's a constant pre-set gain with all overs handled by a limiter. But hey, what do I know?

The engineers set a gain value. They set it based on the sensitivity of the internal mic.

That's why mics have a sensitivity spec. like -60dbu. And, why pro camcorders have a sensitivity spec. And, why pro camcorders have settings like -50dbu, -60dbu, and -70dbu. You set your camera to be equal to or higher (smaller negative value) than the mic rating.

The same can be done with any of the converter boxes. except YOU need to set the box output level yourself.

the reason you need to match is because you don't want your input to exceed the analog stage input rating because if you do so it will create analog distortion that can't be eliminated. Yet, you want the signal high enough to get the average level to be mid-scale.

the 16-bit range takes care of everything else -- although these cameras may have only 12-bit a/d's because IMHO they were designed for ENG. A reporter talking to someone. but, as the subway sample shows, on-location music sounds fine for ENG. (a bit bass-shy but that is likely the mics.)

PS: the real problem is the way too much notion blur. For 24p, the shutter should be 1/50th or 1/60th.

look at this clip:

Canon 5DmkII with Nikon 85mm lens at f1.4 in broad daylight on Vimeo

you can just barely hear the hammer on the bricks, I assume it was a good distance from the camera. There is however a ton of hiss, which to me indicates an AGC system cranking the gain as high as it can go in the absence of any noise. Whether it pumps aggressively or not is something we won't know until someone can specifically test it, but it does appear to have AGC and not just a fixed gain+limiter.

Evan,

Erm, the reason the sound was so bad on my clip was that the MKE400D on the top had run out of battery without me noticing. Obviously if I had been able to monitor it then I would have known! So yes the AGC then kicked in and generated the background hiss.

Its for this very reason that I'm keen on monitoring and a backup audio device in case of emergency.

Dan

If someone is feeling like gambling on a solution you could look here http://www.dvinfo.net/conf/private-c...c-pre-amp.html

The Sound Devices mic pre amp looks perfect for 5dmkII use, only single channel though.

Dan

Certainly it's possible the camcorder "created" the hiss, but it's equally possible the hiss was generated within the mic because the battery was failing.

The simple test is go in a closet and record. If a camera has dumb AGC it will raise the gain to its max. You'll hear hiss.

However, a smart AGC will not do this. So you need a second test. In a room with a low level of noise, turn-on your MKE400. A smart AGC will not raise gain on the mic when it is turned-off. Once turned-on, when it senses "some" signal it will raise the gain.

Now the question is how much does the gain change and how fast? Ideally it should adapt itself to the average input and change very slowly. This should be acceptable for journalism.

Trying to turn these cameras into "movie cameras" is inviting frustration.

As far as I'm concerned, that's the DEFINITION of AGC: the camera is changing gain.

So, to be honest, I'm confused about the whole popular myth thing. In my experience, which I admit might be much less than yours, I haven't yet met a consumer camcorder (that couldn't be set into manual levels) that didn't ride the gain in some form or another. I find it hard to believe that the 5d has a set gain plus limiter. I simply don't believe it.

Feel free to prove me wrong though...

That's the problem with "definitions." Technically a 911 has an Automatic Transmission and so does an F1 car. Yet, these AT's don't really behave very much like the AT in a Ford Pinto.

I've reviewed camcorders for 15 years and from the around the time DV was introduced while the spec sheets still claimed AGC, in fact, the behavior had none of the negatives of an AGC. (The reason, of course, is the switch to PCM audio recording.)

Analog-based AGCs had certain negative aspects. When digital electronics came along it became possible to have an AGC that behaved as would a human. When you "ride the gain" (for a digital recording) you follow a VERY simple set of rules based upon the incoming signal level. It is trivial to build these rules into a DSP chip. The difference is that the DSP is far faster and more reliable than you can ever be. Can it make an error? Yes. Can you make an error? Yes.

So the myth really is that all AGC's are the same. And, therefore, should all be avoided. The problem is in the word AUTOMATIC. AUTOMATIC in the digital world is very different in behavior than it was in the analog world.

PS: That doesn't mean the DSLRs have smart AGCs. It all depends what IC they buy for audio I/O. Or, what they designed into their I/O chip.

That depends. If I don't want the level changed, then it's completely un-reliable to me. Personally, as long as my levels are up in the decent resolution part of dbfs (s/n), I prefer to do my compression and other tweaks in post. But again, it depends on what you're doing of course.

But that's not what you said, and that's not what I implied either. You said no AGC, when clearly these cams ride the levels, which was my point.

I'm not saying they can't do it in an intelligent way, or that they can't do it without pumping, or that they have long release times, I was just saying that they ride the levels. For some, that might be a good thing, and for others it might not, but the fact there's no gain control whatsoever on the 5dm2 is pertinent fact -- not that I personally expected anything less, I was actually shocked there was even an input to begin with.

If you want to get super geeky record an hours worth of time signal and covert to mp3

Then feed the time signal from an iPod to your 5D2 and to your sound recorder.

Here's a short (6MB) clip which demonstrates the auto gain working (using the MKE 400):

http://www.palsomedia.com/canon5d2/a...o_gain_5d2.wmv
(download and save)

What you hear is a very good job of preventing audio clipping. When the overall signal gets near clipping, the gain is reduced which naturally lowers the background noise. It's hardly objectionable. Clipping, which would ruin your audio would be far worse.

Of course, a human might think far enough ahead to set the gain based on the loudest noise. If they knew ahead of time what that would be at a race track. And, where exactly on such a tiny camera would you put the knob? Or, do you want someone to stop and use a menu. I really doubt anyone shooting in gun & run mode at a race track would have time to do this.

Most operators shooting with a $2500 camera would be very happy to get the shot for the nightly news or a webcast.