Thanks! Great to meet all of you too!!!
let me respond with some comments ... on some of your comments ...
****
"I guess I'm making the mental leap then if someone buys a $700-$1200 audio recorder, they're probably not going to plug a $80-100 mic into it"
It's not how expensive the microphone is, but the microphone sensitivity and how well it can be placed. Not everybody is using a 50mV/Pa sensitivity ME66. The majority of people are using shotgun microphones in the class of something like the NTG2 at 15mV/Pa. Plus, different microphone types have less sensitivity, like the most popular hypercardioids (~10mV/Pa), or lavalier microphones (~5mV/Pa). So, even if your not using a dynamic handheld microphone, you're coming closer to it's sensitivity of ~3mV/Pa.
Plus, if you're not able to get that NTG2 boomed 1.5ft away for a wider shot, and it's more like 4ft away, the inverse square law is just push your microphone output signal down to the range of where the dynamic mic is.
Does that make sense? This is why the front end is so important. Some recorders have a great front end. Others ... not so much.
****
"Cameras like the 5D are recording a 16bit file, but due to the quality of the electronics I doubt it's '16 bit clean'"
I know that we're not focusing on DSLRs here, but looking at them is instructive for understanding what is going on, and you are correct in that it is not 'clean'. The limitation camera's like the 5D is the analog front end. But, a quality low noise preamplifier fixes that. It's what's known as the "cascaded noise figure of amplifiers" principle. Even if you have a poor signal-to-noise performing amplifier, if you stick a low noise preamplifier in front of that, the signal to noise performance of the entire system is set by the first amplifier. So, you can make the signal to noise performance of the 5D 16-bit 'clean' with a good low noise preamplifier.
Similarly, with those recorders that performed poorly in the signal to noise tests (even when recording at 24b/96kHz), you can fix their signal to noise performance by sticking a low noise preamplifier in front of them. Check out this old blog post on how to improve the signal-to-noise performance of a Rode NTG2 recorded by a Zoom H4n:
http://juicedlink.com/blog/2010/10/impr ... se-preamp/So, you can record with excellent signal to noise performance directly in the camera with a quality low noise preamplifier, or in a recorder with a quality front end (or with a good low-noise preamp if the recorder front end is a weak link). But, there still are some differences between using a recorder vs recording direct to camera.
First is bitrate. IMHO, a higher bitrate (beyond 16b/44.1k) is not a big deal. More often than not, the long pole in the tent in terms of signal to noise performance is going to be the front end, not the A/D or bitrate. The AES had a great paper that showed that humans (the vast majority) cannot perceive improvement in audio quality at higher bit rates. I know I can't hear 144dB in dynamic range ...
Next is production flow. A recorder can have the advantage of being located remotely from the camera. Recording direct to camera has the advantage of not having to sync in post, and have multiple devices to remember (or forget) to hit 'record' on.
Third is frequency response. Here's where I would agree that there is an advantage for external recorders. Most cameras will not have as good frequency response as external recorders. They usually don't have enough room to fit the big coupling capacitors in camera's for the low end of the freq range. So, the right tool for recording the Chicago symphony would be an external recorder, not a camera. But for recording dialogue, recording to a camera (with a good low-noise preamp) is more than sufficient.
****
regarding the tests you requested above ...
Regarding a tone generator for signal-to-noise tests, the question always is what signal level to inject. That is why the ProVideoCoalition dynamic microphone challenges such a great way to evaluate, because it's a real world signal level at the low end of sensitivity that allows you to compare the recordings in the most stressed environment. Then, you act like a human HP8903B SignalGenerator/AudioMeter, and compare the noise floor of the different recordings turned quiet periods. So, I would go back and listen again to the recordings in the BMC audio applications video, and there are even more audio clips to compare in a previous blog post video. Once you have the sense on how things compare in the stressed situation, then the NTG2 test is less interesting. With a quality low-noise preamp, the cameras and some of the higher-end recorders performed very well in terms of signal-to-noise.
****
"probably not worth it until the bias issue is resolved"
Let me explain a little bit more about the DC offset. It is problematic in that you could be in a situation where you think you've got a juicy signal on the Ultrascope meter, but in actuality you do not because a significant portion of the meter is sensing the DC offset. The amount of DC offset varies with the gain setting in the camera, and it is higher at high gain settings, and lower at low gain settings. As mentioned in the video, you do not want to use a lot of digital gain (which will result in poor signal to noise performance). You want to set the camera gain to the sweet spot as indicated in the video (where you're getting close to no digital gain or digital attenuation). At that sweet spot gain setting in the camera, you do have some DC offset, but it's not that huge. You will take some hit in signal-to-noise. But, it is not a huge amount, as demonstrated by the signal-to-noise audio tests in the video.
So, if you're going to be using the camera and the current state of firmware, you want to set the camera gain to that sweet spot as indicated in the applications video, and leave it there. Use the low noise preamplifier to adjust the signal level so the Ultrascope meters (in dB) are happy. So, you can record to the camera and the current firmware with decent signal to noise, but it's a big pain. You will need to use Ultrascope for metering, or pre-calibrate an external meter to Ultrascope, or use an external monitor which has it's own meter. Then, you will want to remove the DC offset in post, before you do any manipulation. It's all doable, but a pain.
****
I hope readers find this conversation helpful ...
Best regards,
Robert