Last week, Blackmagic announced that their v1.9 firmware update for the 4K Production Camera would include audio meters. This was met with great rejoicing by camera purchasers who have been struggling with the amazingly bad audio functionality of this otherwise exemplary device.
Let me preface what follows by mentioning that I have used a lot of cameras and audio equipment over the last 20 years, and the Blackmagic Production Camera is by far the absolute worst. Not only are the onboard preamps noisier than the worst DSLR – meaning that using anything more than 10% gain will result in audio so full of hiss that is almost totally unusable – but the input tolerance seems to have been calibrated for some kind of alien technology. Every type of signal I fed into it was either clipped or inaudible – and often both!
Armed with the new audio meters, I resolved to test the BMPC’s finicky sound circuits more systematically.
After some preliminary testing, I concluded the following:
1) The BMPC seems to apply extremely harsh compression to any input source, such that any signal of desirable volume sounds harshly clipped.
2) The onboard preamps really are useless. You’re far better off amplifying the signal in post-production than in the camera.
3) Even under optimal circumstances, the BMPC audio is far noisier than the files captured by an inexpensive audio recorder.
To illustrate these points, consider the following tests, which were done using the Tascam DR-40 to feed the BMPC.
DR-40 Recorder Tests
Reference files were recorded on the Tascam DR-40, and then everything was normalized to 0db in Premiere Pro in order to determine what a usable final output would sound like.
For the first test, I plugged a shotgun mic into the DR-40, and – as I normally would, when recording audio – adjusted the DR-40’s gain setting so that the audio meters were peaking at about -12db. This was a gain setting of about “30” on the DR-40. I then plugged the “headphone/line” output of the DR-40 into the input of the BMPC. I set the BMPC to “Mic” input, and set the gain to 5% in order to have the meters hitting at about -12db on the camera. As you can hear, the incoming signal was too hot for the camera, and the audio is clipped (although the 4% gain means that there was very little added hiss).
Now, many of you are scoffing at me for setting the BMPC audio input to “Mic” instead of “Line,” when I’m obviously coming out of a jack labeled “Line.” Well, there’s a reason for that, which test #2 will illustrate.
The DR-40 setup for test #2 was the same as in #1, but this time the BMPC was set to “Line.” However, I had to crank the gain up to 100% in order to get a usable level on the camera (around -18db). The hiss, as you can hear, is quite bad. The reason for this is that the BMPC is (seemingly) looking for a +4db line level, rather than the much lower-voltage signal coming from the headphone jack (more about that in a moment).
For the third test, I returned the BMPC input setting to “Mic” at 5%, and turned the gain on the Tascam DR-40 down to “10.” Although the recorded audio level was very low, once the file was normalized, it yielded the cleanest sound.
Here’s the control file: this is what the normalized audio for all three tests looked like, from the files recorded on the DR-40. As you can see, there’s virtually no difference in the waveforms (in fact, you probably can’t tell where each test begin and ends, just by looking at the spectrum view).
Now, here’s what the normalized audio for all three tests looked like, from the files recorded on the BMPC. Here, you can plainly see the clipping in the first test, and all the additional noise (hiss) in the second test.
Here’s what it sounded like. You can hear the unacceptable quality of audio from tests #1 and #2.
FMX-42 Mixer Tests
But wait, there’s more!
Since the DR-40 is not the only audio device in town, I pulled out my Azden FMX-42 mixer, and tried plugging it into the BMPC. To make a long story short, the most successful combination of settings was coming out of the 1/8″ -36db jack on the mixer and setting the BMPC to 4%. This seems to be equivalent to the setup for test #3, which tells me that the headphone jack of the DR-40 is probably calibrated to about -35db as well.
Interestingly, taking the +4db line output of the Azden, and feeding it into the BMPC set to “Line” input, still wasn’t great. Since the signal going to the camera was a higher voltage (compared to the -36db “line” from the DR-40), I only had to turn the gain up to about 30% in order to obtain a decent (-18db) signal, so it yielded an okay (although noticeably noisier) result, as you can see and hear in the spectrum view and audio file below.
At this point, I would still consider any audio recorded on the BMPC to be a backup for dual-system sound (and/or a sync track). The safest approach seems to be to send a low-volume signal to the camera from the headphone/-36db output of an audio device, and to record on both.
I would be very interested to know if anyone else has gotten better results. If you have, let me know how you did it!