NOTE: With the installation of the new 1.11 Firmware, the audio system characteristics of the PMW-EX1 are significantly improved. There is more headroom before the preamp's limiter introduces its effects on the signal and you can now record all the way to the redline on the meters (audio Rec level at "5" or higher) without "suckout" or "ducking" artifacts from the limiter. This document has been updated to take that into account and is highlighted to indicate the changes with firmware 1.11.
Here is an example of the audio fidelity from the PMW-EX1, using the Rode NT4 stereo microphone. Live Audio starts about 1:30 into the video:
I’ve done
some recording out in the field with the PMW-EX1 and have made some
observations about the gain staging setup used internal to the camera and
how to set the attenuators to achieve the best dynamic range out of the
camera.
My stereo mic of choice is the Rode NT4.
Last Saturday, I recorded the Kent Singers, a choir group of 20 singers,
accompanied by piano and guitar, in a small New England church.
The Rode NT4 has a sensitivity of -38dBv.
Given this, I set the EX1’s attenuators for -38dB, as a starting point.
I found that on a lively number, the
choir’s SPL at 12’ was sufficient to not only hit the limiters, but to
achieve some clipping, where the limiters were not fast enough.
I backed off the attenuator settings by
6dB for each channel and that seemed to free the system from dynamic
compression and clipping. However, during applause segments, heavy limiting
was audible.
I discovered with studio testing with a
very loud sound system, that a setting of -11dB on the mic attenuators would
allow recording of the loudest live music/bands in existence without any
clipping or compression.
GAIN STAGING:
There evidently are TWO limiters in the
EX1: One appears just downstream of the attenuators; the other downstream of
the record level controls.
This can be observed by feeding signal to
the camera’s input jacks at gradually-increasing levels, while the record
levels are set to 5. At some point, the sound will start to take on a
“broadcast on the radio” quality, as the dynamics will be flattened out.
Turning down the record level won’t change this compression effect, though
it will reduce the record level going on to the SxS card.
The only way to remove the effects of the
input limiter is to increase the amount of input attenuation. The smaller
the number, the greater the attenuation. The numbers refer to the signal
level that is needed to be present to achieve a reference recording level.
Since they are negative numbers, smaller is bigger. By increasing the
attenuator setting (so it is less sensitive) the signal peaks will be
reduced below the limiter threshold and the sound will pass pretty much
un-altered.
Even if you set the attenuation
correctly, you can still hit a secondary limiter if your record level is too
high. I find that a setting of 5 is optimal. Any higher than 6 and there
will be an increase in the noise floor. If you need more gain to record
quiet content, like nature sounds, it is more effective to reduce input
attenuation than to turn up the record gain.
An observation about the Auto gain
control setting: it completely ignores your input attenuator settings. The
camera will not be able to handle high SPL situations in Auto. Use Manual
gain and set levels conservatively. You can always normalize in post, but
you can’t fix badly-compressed and clipped audio in post.
In conclusion, if you want to get the
most out of the audio system on the EX1, you must set both the attenuators
AND the record gain appropriately for the situation.
Here are my suggestions, vis-à-vis the
Rode NT4 (dB SPL levels are peak, not average):
Quiet nature scenes, SPLs under 70dB. Set attenuator for -38dB (good to 106dB)Record level of 5 will pretty much provide full-scale (full scale without limiting is 3 dots short of the red dot on the EX1’s audio level meter scale for earlier firmware, or AT the red dot with the 1.11 firmware). A level of 6 will push things toward more limiting on the secondary limiter stage and a little more hiss from the preamp. 5 is really pretty optimal, as it will allow full scale record levels with adjustments to attenuators for expected maximum SPL.
Small acoustic or singing ensembles. Set attenuator for -32 or -29dB (good to 115dB)
Jazz ensembles with moderate amplification. Set for -23 or -20dB (good to 124dB)
Loud disco or rock band concerts. Set for -14 or -11dB (good to 133dB SPL)
Fireworks should use the -14dB range, if you’re close enough to need earplugs.
ADDITIONAL NOTES 3/18/2009
For those involved in recording pyrotechnic events, such as fireworks, it is important to bear in mind the psycho-acoustic effects of short-duration, transient sounds. Because of the low average sound levels (lots of near-silence between mortar explosions), it is natural to want to push the recording levels much higher than would be proper for this situation. Fireworks are a tricky thing to reproduce properly. Since the SPLs can exceed 120dB at the spectator location, one needs to set attenuators based on the PEAK levels, not the perceived average loudness. I have heard far too many badly-distorted, compressed fireworks recordings because the camera operator set record levels based on perceived loudness.
Another thing to be aware of with fireworks is that you can't reliably set the levels during the first few explosions. By the time the grand finale happens, levels will increase as much as 8-10dB because of overlapping explosions causing wave crests from one explosion to reinforce in-phase with others that may occur simultaneously. A large grand finale can reach 130dB in peak SPL! Resist the temptation to turn up the recording levels. Trust the meters.
With regard to attenuators, the last fireworks display I shot with the PMW-EX1 had me setting attenuators at -14dB and with using the 10dB attenuator switches on the studio condenser mics that I used. That would be the equivalent of having a -4dB setting on the camera attenuator (which does not exist on the EX1). So bear in mind that peak levels can be VERY high.
The other tricky thing about fireworks is that much of their perceived loudness is due to, well, true loudness, and the rest is due to the reverberation of sound throughout local terrain reflections. I used to record fireworks with an external recording system that samples at 96KHz and when I reviewed the audio in a waveform editor, I found that each explosion looked exactly like a static discharge 'tick', but greatly amplified. The rise times were in the low microsecond range, often just two samples in duration on the rising edge of an explosion. These recordings, when played back at low levels, sounded just like static 'ticks' and not fireworks. To sound like fireworks, the audio must be played back at the original sound pressure level. This requires speaker systems with incredibly-fast transient response and power handling--often several thousand watts of power is required to reproduce the full effect of a properly-made fireworks recording. If you are fortunate enough to have sound equipment capable of reaching 130dB SPLs without audible distortion, then the possibility exists for a realistic playback of fireworks recordings that are properly made.
Cheating, via clipped, compressed and distorted recordings will achieve more of a sense of 'loudness' at lower playback levels, but won't sound like the real thing. Let's face it: it takes serious industrial-strength audio gear to reproduce pyrotechnic sound realistically. So if your recording ends up sounding like static pops, chances are you've recorded it correctly. Now all you need to do is play it at the correct level, which is going to be unattainable for 98% of home hi-fi systems without damage or severe distortion, because you're asking for 20dB more SPL than the average high-end stereo is capable of. That's a factor of 100X more power.
The PMW-EX1 can do a respectable job capturing the transients of fireworks, because it's response extends to beyond 20KHz. But in most cases, you'll need to use the maximum attenuation that is available on the EX1's audio menu, if you're using high output studio condenser mics.
Last Updated 11/27/2023 07:10 PM