Topic: I managed to overload my FR-AV2 mini-jack mic input with a pair of Clippy mics

[TheJez]

That's the disconnect.  The 32bit float container can store data well above 0db.  You can demonstrate this by simply plugging in a line level source, set the preamp to 0db, and turn up the volume on the source until you're hitting above 0db.  Then load it into a DAW, and normalize it to -0.5db, and you'll see the dynamics return.

That's indeed a good test. I'll try to make a setup as you described when I get the chance.
Edit: I do understand 32bfp, so this test for me would primarily show when analog clipping would start to occur, and if this would be below or above 0dB (when the so-called 'Record Level' is set to 0dB)

[Gutbucket]

The Clippies / EM272 have a sensitivity of ~40mV/Pa

That's very sensitive. Helpful for achieving good levels and lower self-noise in quieter environments, but in high SPL conditions the resulting high output will overload the mic-input stage of many recorders. You'll need to switch to line-in to accommodate that.. assuming the mics themselves are capable of handling the SPL levels without distorting.

[VibrationOfLife]

The Clippies / EM272 have a sensitivity of ~40mV/Pa

That's very sensitive. Helpful for achieving good levels and lower self-noise in quieter environments, but in high SPL conditions the resulting high output will overload the mic-input stage of many recorders. You'll need to switch to line-in to accommodate that.. assuming the mics themselves are capable of handling the SPL levels without distorting.

They are not, that is his problem, not the recorder.  He is misguided in his mind.  But apparently he refuses to get that.

[nulldogmas]

The Clippies / EM272 have a sensitivity of ~40mV/Pa

That's very sensitive. Helpful for achieving good levels and lower self-noise in quieter environments, but in high SPL conditions the resulting high output will overload the mic-input stage of many recorders. You'll need to switch to line-in to accommodate that.. assuming the mics themselves are capable of handling the SPL levels without distorting.

They are not, that is his problem, not the recorder.  He is misguided in his mind.  But apparently he refuses to get that.

"I expect that I may have been also overloading the mics. One doesn't rule out the other."

[Gutbucket]

Its both.  That recorder's mic-input stage is unable to support microphones with a sensitivity of 40mV/Pa when things get loud.

If he were using mics that WERE fully capable of handling high SPLs AND had a sensitivity that high, the recorder's input stage would most certainly clip before the microphone does, unless the output was sufficiently padded down.

A couple come to mind:
Sennheiser MKH 800 TWIN: 40mV/Pa, 136dBmax SPL, 4V max output
DPA 4006-TL: 36 mV/Pa, 146dB max SPL, 8V RMS max output

I'm sure there are others.

[Ozpeter]

My simple way of understanding 32 bit float is that it provides a 1500db wide ribbon along which the 140db (or whatever the figure is) ribbon of audio is written.  It matters not whether the the narrow audio ribbon is positioned at the bottom, middle or top of the 1500dB 32 bit float ribbon - in post production it can be repositioned as required.

And if the mics or preamps or anything before the a/d converter(s) are overloaded, the 32 bit float part will record the resulting clipping with great accuracy...

[VibrationOfLife]

And if the mics or preamps or anything before the a/d converter(s) are overloaded, the 32 bit float part will record the resulting clipping with great accuracy...

GIGO.

[TheJez]

The Clippies / EM272 have a sensitivity of ~40mV/Pa

That's very sensitive. Helpful for achieving good levels and lower self-noise in quieter environments, but in high SPL conditions the resulting high output will overload the mic-input stage of many recorders. You'll need to switch to line-in to accommodate that.. assuming the mics themselves are capable of handling the SPL levels without distorting.

They are not, that is his problem, not the recorder.  He is misguided in his mind.  But apparently he refuses to get that.

"I expect that I may have been also overloading the mics. One doesn't rule out the other."

Thank you, nulldogmas. I indeed never ruled out that the mics were overloaded. Regardless if they were (which is likely the case) or not, also the recorder was overloaded. At least, that was what it indicated. Not sure why VibrationOfLife  thinks this was not the case. If the recorder says "analog input overloaded", then I assume it is overloaded.
On a side-note, I am curious how the recorder detects this. Would there really be a signal coming from the analog stage indicating 'I'm overloaded!', or would this be derived indirectly from the digital output of the A/D converters? Guess we'll never know.

[TheJez]

My simple way of understanding 32 bit float is that it provides a 1500db wide ribbon along which the 140db (or whatever the figure is) ribbon of audio is written.  It matters not whether the the narrow audio ribbon is positioned at the bottom, middle or top of the 1500dB 32 bit float ribbon - in post production it can be repositioned as required.

I've read this analogy before from you and it indeed gives a nice mental picture of 32bfp. However, the manufacturers won't put the 140dB ribbon just anywhere along the 1500dB ribbon. They will put it where it makes some sense, so no post-processing would be needed to still get some decent levels in low amplitude circumstances.
I was expecting that the max output of the ADC's would match the 0dB level within the 1500dB, but apparently this is not the case. Apparently (with this recorder) some fixed amplification is applied to make the stored samples a bit louder, i.e. to raise the 140dB ribbon a bit higher, likely to get decent levels in low amplitude circumstances. And by using the 'record level' (0..60dB digital amplification), the user can raise it even higher. (And yes, by doing so, possibly get samples above 0dB, which is perfectly fine and 'fixable' with 32bfp)

And if the mics or preamps or anything before the a/d converter(s) are overloaded, the 32 bit float part will record the resulting clipping with great accuracy...

Yes indeed!

[Ozpeter]

TheJez, indeed my posts get a bit repetitive    As for levels, my experience with relatively simple 32 bit float recorders, whose digital-only level controls I don't mess with, is that with other than unusually quiet sounds, it's often necessary to bring down levels in post.  As a wild guess with no supporting evidence, I suspect that -12dB in 24 bit recorder terms is treated as 0dB in 32 bit float by these recorders.  So when the analog end saturates, the 32 bit float file peaks at +12dB.  Next time I get a chance I might test that, making loud noises to annoy my neighbours...

[Gutbucket]

I am curious how the recorder detects this. Would there really be a signal coming from the analog stage indicating 'I'm overloaded!', or would this be derived indirectly from the digital output of the A/D converters?

Yes. It monitors peak voltage through the input stage.  When the signal voltage reaches the limits of headroom available from the analog input stage circuitry it pings the overload warning or light.