FWIW... Microphone self noise and mic preamp noise make the 32-bit v 24-bit debate sort of irrelevant. 24-bit is more than big enough a container to handle the dynamic range of any real world sound and provides all the resolution and accuracy you need even when capturing extremely low level analog signals.
This pretty much sums up how I am looking at this, although I am no expert. I would love to hear a well conducted comp or see some numbers. If there is a real advantage, it should be hearable or measureable.
I believe Paul's comment was in regards to 32-bit
fixed; not
floating. They are not at all the same thing. Please see the links I posted in
Reply #25.
I do agree with you on "hearable and measureable", but I believe the links in that post explain what the advantages of float-point recording are.
That review by Curtis Judd is very misleading, I think. All of the gain changes were made after conversion. It isn't like he actually recorded a really quiet source and then showed it to be noise free on amplification. I guess he showed that 32-bit float is useful for processing, but not much else.
It is not misleading at all. He did one recording at an extremely low level, and another at extremely high level. He showed that increasing the level of the very low-level recording in post did not increase the preamp noise, and then he demonstrated the other recording that had levels way above 0 dBFS could be lowered to within normal digital signal levels, and there was no clipping.
This demonstrates that you never need to worry about your gain being set too high or too low in such a system. In fact,
the F6 preamp gain is fixed when in 32-bit float recording mode. This means the preamp gain was identical for both the high-level and low-level recordings. The difference in levels for the recorded tracks were made by the F6 post-ADC (the channel level knobs are faders; not gain trims in 32/float).
I'm assuming your use the of the word "gain" here doesn't really mean "gain" in the analog sense but digital signal "level". Either way: changing the levels after conversion is
exactly the point. With this kind of system, you can do that without compromising the original quality.
Try either of those extreme recordings with similar post manipulation with a 24-bit (or 32-bit) fixed-point recording system. If you have very low-noise mics and preamps, you might get similar results on the quiet recording, but definitely not on the loud one - it will be clipped all over the place.
I also find the reactions to this kind of ironic; when the MixPre-3/6 came out, SD claimed something pretty similar, due to the wide dynamic range. "Now, precisely setting the Gain is much less critical (in most cases unnecessary) due to the extremely quiet (lack of hiss) nature of their design. This allows you to use the MixPre-6 with the confidence that even if optimal Gain levels haven’t been set, your audio will still be handled and recorded with grace. The extraordinarily low noise floor of the Kashmir mic preamplifiers allow recorded tracks to be 'normalized' (the process of adding level to a whole recorded track as to align the loudest peak with near-maximum) in post-processing without fear of ruining the track due to the added noise that would have been present in other preamplifier designs. Finally, you’re free to capture audio on the fly without worry!" People here were not at all receptive to that argument (it's a dealbreaker and whatnot), but with this Zoom it's a game-changer...
You cant directly compare the two. MixPre 3/6/10 only records in 16 or 24 bit, but uses 32-bit for internal processing. The criticism (at least from me) was that they were being a bit misleading in their specs and promo materials by touting "32-bit precision" of the ADC, when the recorder doesn't actually produce 32-bit files, be they fixed or float.
Sound Devices' selling point for the MixPre is the extremely low noise floor of their preamps, which allows you to set levels lower than you otherwise would normally do, because you don't want your low-level signals to be anywhere near the self-noise of your preamp.
Zoom's selling point for the F6 is the dual ADCs and 32-bit float point recording, which means that the dynamic range available is so wide, with the full resolution available even at extreme levels, that the gain setting is entirely irrelevant.
Two different approaches to tackling the same issue.
So, yes, I absolutely believe the F6 is a game-changer, especially for what I record. The only other recording devices I am aware of that incorporate a similar dual-ADC system are Zaxcom recorders, and those are many times more expensive.