Solving Sound problem with AGC

[Chris Barcellos]

Originally Posted by Jon Fairhurst

The 5D2 has a 16 bit A/D. That gives a dynamic range of 96dB (6dB per bit.) At 40dB, the camera is only delivering 7-bit audio(!) The -70dB noise floor is closer to 12-bits of dynamic range - and I recorded at 24 bits! It looks like 16-bit recording is all I need for this combination of mic, pre and recorder. No reason to use any more memory space than needed...

So Jon, I had this happen with my Eng Mixer, where when I ran audible tone, and used the panning switches not disimilar to the ones on the CX231, and I had bleed from the tone channel into the mic channel. I am wondering if there is possiblity of some bleed here raising the floor level with bleeding over of the tone. I am not so sure these pan switches are designed to completely isolate channels. Or should I be complaining to Sign Manufacturing ?

[Dan Chung]

Chris,

FYI in my testing of the 1khz XLR tone generator I've found that I get tone bleed depending on which audio adapter I'm using. With a Beachtek DXA2s it is pretty bad, but with a DXA6 it is non-existant, both fed from my Sound Devices mixer. Its interesting that you are getting issues with the Juicedlink too. I pinning all my hopes on the DXA-5D now.

Dan

[Chris Barcellos]

Dan:

I am not disatisfied with the sound. I was commenting on Jon's numbers, and wondering if the noise level of the floor could be attributed to that. I don't have evidence of it one way or the other. And my question was whether these pan switches allow some bleed.

[Jon Fairhurst]

Originally Posted by Peer Landa

Yes, theoretically 16 bit will give you 96dB, but in real life this will most likely be in the low 80s.

Agreed. Not only is there analog noise, but clock jitter, and errors in the quantization steps. Getting 70dB with the NT1-A, juicedLink and MicrotrackII right out of the box (and with a PC running in the background) is pretty darned clean.

[Jon Fairhurst]

Originally Posted by Chris Barcellos

I am not disatisfied with the sound. I was commenting on Jon's numbers, and wondering if the noise level of the floor could be attributed to that. I don't have evidence of it one way or the other. And my question was whether these pan switches allow some bleed.

Chris, I agree that the sound quality is fine, but the noise floor is high. What frequency signal did you use? I could try some filtering. Unfortunately, the track has very little silence, so it's hard to get a good noiseprint for noise reduction.

One problem with the whole approach is that the connection to the camera is unbalanced stereo in a single cable. No matter what we do, there will be some bleed. In fact, I got lots of bleed even when sending the pilot tone directly into the camera with a short dual-mono to stereo combiner plug. The problem is that we need to crank the signal so high in order to defeat the AGC. I should try a low tone like 200Hz. It will have less crosstalk and might be more effective at reducing gain at lower levels.

Before long the Magic Lantern solution will make this moot. We'll be able to get stereo signals into the camera, will be able to manage the gain of the preamps, and won't have to worry about artifacts from the pilot tone.

[Chris Barcellos]

I see Juiced Link is approaching this for DSLR's, with a product release on the new year.

juicedLink