Topic: New Sound Devices MixPre II-series coming?

[aaronji]

^ With respect to the analog limiters, I was specifically referring to 24-bit mode (limiters are disabled in 32-bit). The way I look at it, there are three possible scenarios:

1. The signal exceeds 14 dBu and overloads the input, irrespective of  the limiter. No need for an analog limiter.
2. The signal is less than 14 dBu and there is no analog gain. As the input and ADC are configured to avoid issues, that signal wouldn't overload the ADC. No need for an analog limiter.
3. The signal is less than 14 dBu and analog gain is available. Gain applied in the analog realm could then create a signal that would overload the ADC. Analog limiter comes in handy!

So, basically, the analog limiter only matters if there is analog gain. Maybe I am missing something, though. It certainly wouldn't be the first time... 

[Gutbucket]

Possibly something of a hedge on user acceptance of the novel 3 ADC recombination scheme?

Could approach it two ways:
1) Signal follows a traditional ADC scheme in 24 bit mode, making analog limiter advantageous given that dynamic range limitation.  (Hedging bet, evolving the tech)
2) Signal follows multi-ADC scheme in 24 bit mode as it does in 32 bit mode, with 24 bit derived from that.  In which case fitting signal to 24 via ranging, limiting, or whatever can be all done digitally (Requiring user acceptance of multi-ADC scheme even if recording 24 bit)

[Gutbucket]

I can understand the motivations underlying both approaches.

If it is not the way it currently works, option 2 would seem the way forward in future model revisions.

[aaronji]

Possibly something of a hedge on user acceptance of the novel 3 ADC recombination scheme?

Could approach it two ways:
1) Signal follows a traditional ADC scheme in 24 bit mode, making analog limiter advantageous given that dynamic range limitation.  (Hedging bet, evolving the tech)
2) Signal follows multi-ADC scheme in 24 bit mode as it does in 32 bit mode, with 24 bit derived from that.  In which case fitting signal to 24 via ranging, limiting, or whatever can be all done digitally (Requiring user acceptance of multi-ADC scheme even if recording 24 bit)

Yes, I was thinking along the lines of option 1, not necessarily as a hedge, though. A lot of people are using these for film and they say post-production expects 24-bit. I also think that SD would avoid the expense of an analog limiter solely for marketing. Their more professional recorders haven't yet implemented 32-bit, which kind of reinforces the idea that 24-bit is not quite dead.

I wouldn't be surprised if option 2 becomes increasingly common in the future, however.

[Gutbucket]

Yes except either approach produces a 24 bit output which satisfies industry post-production requirements today.  16 and 24 bit formats aren't going away and I expect will continue to be accommodated even in future miniature on-talent type 32PF recorders where a reduced feature set would be expected.

There exists inertia with regards to professional expectations and user confidence as well as inertia in developing things across product lines.  Perhaps analogous to it having taken a few generations of new ship building for mercantile transportation to fully evolve from sail to steam.

[aaronji]

^ True.

Another factor that might affect the "inertia" is SD's patent. If the industry arrives at a consensus that their approach is the way to go (and, on paper at least, it seems to be), I guess that will lead to substantial licensing fees or development of alternative approaches that don't violate the patent, both of which would be barriers to rapid adoption on a broader  scale. It will be interesting to see how it plays out.

[voltronic]

Another factor that might affect the "inertia" is SD's patent. If the industry arrives at a consensus that their approach is the way to go (and, on paper at least, it seems to be), I guess that will lead to substantial licensing fees or development of alternative approaches that don't violate the patent, both of which would be barriers to rapid adoption on a broader  scale. It will be interesting to see how it plays out.

SD is not the only brand who figured out how to do this type of thing though, and they only have a patent on their specific design.  Zoom has their solution, and prior to that you had Zaxcom with their NeverClip tech which uses dual ADCs, albeit with 24-bit fixed point recording.

I recently learned that before any of those, StageTec had TrueMatch using multiple ADCs and 32-bit float point in their consoles and routers.

My only point is that I just named 4 companies who independently figured out some variation of using multiple gain-staged ADCs to capture greater dynamic range, and to my knowledge, none of them sued another for patent infringement because the designs were sufficiently unique.

[aaronji]

SD is not the only brand who figured out how to do this type of thing though, and they only have a patent on their specific design.  Zoom has their solution, and prior to that you had Zaxcom with their NeverClip tech which uses dual ADCs, albeit with 24-bit fixed point recording.

I recently learned that before any of those, StageTec had TrueMatch using multiple ADCs and 32-bit float point in their consoles and routers.

My only point is that I just named 4 companies who independently figured out some variation of using multiple gain-staged ADCs to capture greater dynamic range, and to my knowledge, none of them sued another for patent infringement because the designs were sufficiently unique.

SD is using three ADCs, unlike Zoom or Zaxcom (I don't know about StageTec), and a mathematical model that is novel. Zaxcom is known to have defended their patent and, rumor has it, the delay in the introduction of the F6 was due to a conflict with that patent. None of those other methods caught on to any large extent, and, as I mentioned, "If the industry arrives at a consensus that their [SD's] approach is the way to go", which is not a foregone conclusion...

[voltronic]

SD is not the only brand who figured out how to do this type of thing though, and they only have a patent on their specific design.  Zoom has their solution, and prior to that you had Zaxcom with their NeverClip tech which uses dual ADCs, albeit with 24-bit fixed point recording.

I recently learned that before any of those, StageTec had TrueMatch using multiple ADCs and 32-bit float point in their consoles and routers.

My only point is that I just named 4 companies who independently figured out some variation of using multiple gain-staged ADCs to capture greater dynamic range, and to my knowledge, none of them sued another for patent infringement because the designs were sufficiently unique.

SD is using three ADCs, unlike Zoom or Zaxcom (I don't know about StageTec), and a mathematical model that is novel. Zaxcom is known to have defended their patent and, rumor has it, the delay in the introduction of the F6 was due to a conflict with that patent. None of those other methods caught on to any large extent, and, as I mentioned, "If the industry arrives at a consensus that their [SD's] approach is the way to go", which is not a foregone conclusion...

See the link I posted.  StageTec appears to have 4 gain taps feeding 4 ADCs.

"Despite of the enormous effort needed, for example, by a fourfold-stacked 32-bit A/D converter, the Stage Tec converter requires even less supply power than the conventionally structured combination of microphone preamps and A/D converters."

I'm curious to see where you read that rumor about Zaxcom and Zoom.  There are some in the TV/film sound world who consider Zaxcom to be overly aggressive on their patents (though I won't use the "T" word).  Like anything else, you have to decide how far to push things, I guess.  It seems from the outside looking in that the only thing similar between the Zoom F6 and the Zax Nomad or MAXX is the dual ADCs, but they are otherwise rather different.  I'm no attorney though, and I haven't read the patents because I admit some of that stuff is over my head.

I get what you are saying about SD's implementation potentially being the way to go, but for legal reasons it seems like competitors will have to find some other way, as these other companies have.  Either way, I'm interested to see where all of this goes.

[Gordon]

what firmware are folks running?  I've been on v4.00 since I got it back in Sept.  No issues at all but I figure this is as good of a time as any to try out a newer version.

[jbell]

I updated to the most current version and home testing everything works fine!

what firmware are folks running?  I've been on v4.00 since I got it back in Sept.  No issues at all but I figure this is as good of a time as any to try out a newer version.

[aaronji]

Post-ADC.

As I posted on the previous page, I was curious if Paul's comment applied only to 32-bit mode or if it applied to both 32- and 24-bit. I e-mailed that question (with a link to the conversation in this thread) to Sound Devices. Here is the response:

I tracked down Paul this afternoon to elaborate, and "post-ADC" gain applied to 32-bit recording mode on the new -II series.  When you switch to 32-bit float mode, the recorder power cycles and reconfigures the audio path, as it's all FPGA processor based.

I thought that was still a little ambiguous, so I asked, "To be clear: 24-bit: two gain stages, one analog and one digital; 32-bit: all gain post-ADC. Is that correct?" and was told yes.

Hopefully, Paul can weigh in again if this is in some way inaccurate.

[jerryfreak]

If that’s the case that’s quite possible the two gain stages may have distinctly different sounds if one is using an analog and stage in the other is not

Not that one or the other would be necessarily preferable, but just different

[justink]

Post-ADC.

As I posted on the previous page, I was curious if Paul's comment applied only to 32-bit mode or if it applied to both 32- and 24-bit. I e-mailed that question (with a link to the conversation in this thread) to Sound Devices. Here is the response:

I tracked down Paul this afternoon to elaborate, and "post-ADC" gain applied to 32-bit recording mode on the new -II series.  When you switch to 32-bit float mode, the recorder power cycles and reconfigures the audio path, as it's all FPGA processor based.

I thought that was still a little ambiguous, so I asked, "To be clear: 24-bit: two gain stages, one analog and one digital; 32-bit: all gain post-ADC. Is that correct?" and was told yes.

Hopefully, Paul can weigh in again if this is in some way inaccurate.

Now we need a “24bit (recorded close to zero) > normalized” and a “32bit > normalized” comp.

[carpa]

So it seems to be similar to the Zoom F6....no analog gain adjustable before recording.