Personal View site logo
Make sure to join PV Telegram channel! Perfect to keep up with community on your smartphone.
Sound Devices MixPre II 32bit float format recorders
  • image

    Specs

    sa9883.jpg
    800 x 449 - 50K
  • 14 Replies sorted by
  • according to the manual only the mixpre 10 has timecode in-out, right?

  • 32 bit float for a field recorder is perhaps one of the stupidest things I have ever seen in a half century of recording.

  • @DrDave

    Can you be more specific on what is stupid?

    Idea with two ADC channels and making 32bit float files for easy and unclipabble workflow is superb.

    Thing that we see in audio industry very closely resemble with video industry, as here it is also patent ( Zaxcom ) that prevented to do this, but it seems like it comes to an end (but consider that Zoom recorder had been delayed!).

    In real field recording lot of levels issues happen, as well as lot of situation where clipping can happen due to audio person responsible or lack of such.

  • It doesn't give you any more headroom as the noise floor intrudes before you hit 32 bits. Actually, it isn't the stupidest, come to think of it I have seen some pretty ridiculous things.
    I would rather see some improvements in the preamps and converters than more bits. Especially real Class A, etc.
    Here is the solution (idea copyright): a parallel circuit in isolation that records a bitmatched duplicate at a true lower level (before the signal gets gooped up), then switches over seamlessly if a clip is detected so that the recorded master has an edit in place for the clipped replacement, ready to roll, at the proper gain level, with a backup, of course.
    This is what we really need, not more bits. This would actually be pretty simple to engineer, especially if you are using surface mount stuff anyway.

  • @DrDave

    It doesn't give you any more headroom as the noise floor intrudes before you hit 32 bits. Actually, it isn't the stupidest, come to think of it I have seen some pretty ridiculous things.

    You got it all wrong.

    32 float is mainly used also due to handy editing and wide support and simple recording logic, not only for optimal size initial audio storage.

    Also last generation 32bit ADC chips are used in 2 channels here, and it is only microphone own noise that is limit.

    I would rather see some improvements in the preamps and converters than more bits. Especially real Class A, etc.

    Improvements with preamps in ADC come with it also.

    Here is the solution (idea copyright): a parallel circuit in isolation that records a bitmatched duplicate at a true lower level (before the signal gets gooped up), then switches over seamlessly if a clip is detected so that the recorded master has an edit in place for the clipped replacement, ready to roll, at the proper gain level, with a backup, of course.

    Sounds fuzzy.

    In reality it is two different gain preamps doing to two ADC channels. Whole thing with 32 bit float is that you don't do ANY switching, it just get part of info from one channel (lower bits of higher gain channel are being thrown out due to ADC properties) and another part from another.

  • I could have it wrong, I suppose :)
    So here's the thing--let's say you make a zillion recordings, over decades, and you keep track of which ones are ruined.
    Some are ruined because of overs.
    But half of those overs occur before the signal hits the recording box. Most of these overload the mic, some overload the preamp that is in the mic (not the "preamp" as we know it, but the stuff inside the mic), some are in fact capsule or shell resonance, where the mic itself vibrates sympathetically. And buzzes. Or worse.
    And here is the solution: You buy quality mics that don't overload. You take the mics apart and apply acid free dampening silicon to the shells and test for ringing.
    You buy or make high quality Y cables.
    Then you apply basic redundancy at all levels. So, for example, instead of this 32 bit thing, which I'm sure is fine, you buy two Zoom F8 and connect them with a sync cable. They are totally locked. You set one at the standard -12dB, which gives you an effective 22 bits. You set the other one at -32 to -24dB, which gives you around 20 bits. Since it's a Zoom, you can add the limiters which gives you another 10dB or so, depending. You split the mic cables with the Y and hook up your two zooms. The mics go in before anything. Before the pres. Before the AD. Before the plastic knobs which may in fact have a cheap, wonky pot.
    This is the kind of redundancy we use for recording major symphony orchestras, where a mistake can be astronomically expensive.
    This will cover 99.99 percent of all issues.
    But what, you ask, can you do for the other 0.01 percent?
    You double up the main and spot mics with ones that have slightly different (and more forgiving) overload characteristics. And then you are set. Most people don't do the mic doubling. I have had one mic fail 20 years ago. I had it doubled. You can buy the CD.
    My parallel circuit idea is better, of course, but we still use two recorders. How you use the two recorders depends on how much insurance you want. If you don't use the Y cables, you add a degree of risk. It's small, but it's there. You can daisy chain the two recorders. A lot of ppl go that way. It's very reliable. It can definitely fail once in a blue moon.
    The fact is, it is so much cheaper than it used to be. Two Zooms cost less than a decent mic. Decent times.

  • My parallel circuit idea is better, of course, but we still use two recorders. How you use the two recorders depends on how much insurance you want. If you don't use the Y cables, you add a degree of risk. It's small, but it's there. You can daisy chain the two recorders. A lot of ppl go that way. It's very reliable. It can definitely fail once in a blue moon.

    Well, it is called "kolhoz" here. I mean not good DIY like thing prone to issues, lot of things can go wrong if people are not professionals and not have lot of experience like you.

    Thing that you actually want is to have second SD card and some backup power.

    All else already present inside this recorders, as they do thing you described, use two preamps with gain offset and combine resulting signal throwing out some data from high gain channel.

  • Nice timing and nice to see a lower price on top model!

  • according to the manual only the mixpre 10 has timecode in-out, right?

    Only one with a DEDICATED timecode I/O

    All the others reuse a 3.5mm stereo input for this purpose.

    Thing that we see in audio industry very closely resemble with video industry, as here it is also patent ( Zaxcom ) that prevented to do this, but it seems like it comes to an end (but consider that Zoom recorder had been delayed!).

    Not to the end, nope. Rather Sound Devices has filed their own patent, which does it a little different. (3 vs 2)

  • Interesting article with samples

    Low Level Signal Resolution: 32-bit Float versus 24-bit - https://www.sounddevices.com/low-signal-32-bit-float/