Zoom F8 Noise Performance: One Simple Test

Jim Braly asked a good question about the Zoom F8 field recorder:

A question about how quiet the preamps are -- I'm impressed that the Zoom has 75db of gain (compared to 66 on the Sound Devices 7-series recorders). I found that the last 6db on a 722 SD recorder was not very usable, as it got pretty noisy above +60db of gain. And that last bit of db helps for quiet dialogue, or on wider shots when the boom mic cannot get as close to the actors.

How about the F8? What is the noise floor like if you crank it to +75 with the mic in a quiet room? (No recording of dialogue, just the room tone and the self-noise.) Do you hear a lot of hiss? At what level would this hiss be? Louder than, say, -50db if you are measuring the audio file in iZotope RX? If the F8 does get noisy at high gain settings, at what point does it become unusable? Maybe above 70db of gain? But would, say, 66db of gain sound good?

Of course, I'm wondering how the F8 specifically compares to Sound Devices in the noise floor department, as I think of their recorders as the gold standard.

I ran a Shure SM58 into the Zoom F8 via a 25' ProCo XLR cable with Neutrik connectors, set the gain at +75dB and recorded my basement studio. This room is not perfectly silent and not perfectly treated in an acoustical sense. But I do have a sound blanket hanging to one side, exposed batt insulation in the ceiling and an old duvet on the floor.

Looking at the silent section, I measured the noise floor (this is the sound of the room + the self noise of the mic and the F8 recorder) at -63dB RMS. In my final produced audio, I usually aim for the "silent passages" to sit at -60dB or lower. This seems acceptable and is a rather impressive result for the Zoom F8.

Looking at the spectral chart, it appears that of the noise that is present, the majority of it is in the 100Hz and lower range which should be reasonably straightforward to reduce with a simple high-pass filter. There is some spread throughout the entire spectrum, but very little. (in the screenshot above, I measured the section from 35 to 39 seconds. Just prior to that, the mic was able to pick up a few notes from my wife's violin upstairs.)