NeverClip... in simple pictures

[RadoStefanov]

Also very important to mention that even without the iso attenuation you will benefit gtom the extra 17db dynamic range. Trx900la is 106dB the new one supposedly will be 123db.

[soundslikejustin]

" Justin,

You are doing a great job explaining never clip. "

I'd assume they've got "

" the internal mixer of the Nomad is 32 bit - it has room for this 'extra' digital signal. "

where is Glenn to give this interpretation his blessing ?

" this graph should be a sticky, "

why?

Thanks Senator, for being a jerk.

The Nomad's 32 bit mix engine is noted on the website. This was also covered in multiple Nomad threads (do a google search with site: jwsoundgroup.net in it) and in the Coffey Sound Nomad/NeverClip presentation. It is also how ProTools and many other DAWs run their internal processing, making it almost impossible to clip the internal mix busses.

As for the preamp to a-d alignment, I said in my post I wasn't sure of the specifics, and to 'contact the manufacturer' if the poster was curious. What I had outlined in that post makes the most sense.

I'm trying to help people. Maybe you could quit being a smart ass and do the same.

[soundslikejustin]

So if I have ISO Attenuation enabled, the outputs will sound the same when I'm monitoring, and if I monitor the inputs, they will sound attenuated? I'm not really monitoring the inputs as I go, so no real difference as far as monitoring goes, correct? I understand everything else I think but I'm a little fuzzy when it comes to monitoring the attenuated/unattenuated signal.

I'll stop before I start typing a foreign language, I'm starting to confuse myself again.

If you're listening to an input, and it goes into the range above what the outputs can handle (the headphone amp included) it will sound distorted.

If you're listening to an output, it is as you've always listened.

If you're listening to an ISO with attenuation then yes, it will sound softer than your outputs and inputs.

Without ISO attenuation, listening to inputs and tracks (assuming they are pre-fader) will be the same.

[studiomprd]

" What I had outlined in that post makes the most sense. "

but is it the way it really is...

 

" Nomad's 32 bit mix engine is noted on the website...how ProTools and many other DAWs run their internal processing,. "

 

that is 32 bit Floating Point  DSP,  which is a bit different...

you can't interchange PCM digital with FP

[soundslikejustin]

" What I had outlined in that post makes the most sense. "

but is it the way it really is...

" Nomad's 32 bit mix engine is noted on the website...how ProTools and many other DAWs run their internal processing,. "

that is 32 bit Floating Point DSP, which is a bit different...

you can't interchange PCM digital with FP

Maybe it's not the way it really is, but perhaps the onus could be on those that want the exact information (that is/was not particularly relevant to this image/explanation of signal level) regarding the preamp/dual a-d relationship should contact the manufacturer. This image was regarding what happens to the signal AFTER it had gone through this stage.

I'm not trying to make out that I know everything about the system, as I'm sure some parts are protected under IP and won't be released. I've simply taken info that Glenn has given out and recombined it in a simple, graphical way.

As for Nomads internal mixer, here's a copy-paste from the site:

♦32 bit floating point DSP processing

If I wasn't using the exact correct terminology I apologise, but it is true that the reason Nomad can accept signals above what the 24-bit outputs can handle is due to a higher bit-level mixing DSP.

[VM]

The Zaxcom preamp is very low noise, very low distortion. The limiting factor in the past has been the range of the A-D chip. Now, it's not. I'd assume they've got the absolute end of the preamp aligned with the absolute ends of the a-d system as to maximize signal and minimize noise, but I don't know the specifics, so that might be a question for Glenn if you're curious.

 

With pro recorders I have no low noise problems with "normal" and loud sounds. Only with very tiny sounds, and those are not concerned by NeverClip.

It seems to me that the good point may be the ability to record without limiter, but not the dynamic range. The more so as the range of human ears (nearly 120 dB)  are supposed to be inferior to 24 bits (144 dB).

[bralleput]

+1 VM

Hardly ever have 'Clips' on my inputs ,but : I work drama: high noise can be anticipated... 

[Constantin]

With pro recorders I have no low noise problems with "normal" and loud sounds. Only with very tiny sounds, and those are not concerned by NeverClip.

It seems to me that the good point may be the ability to record without limiter, but not the dynamic range. The more so as the range of human ears (nearly 120 dB) are supposed to be inferior to 24 bits (144 dB).

As I understood it NeverClip is exactly for the very quiet sounds. The point being that with any other recorder (except for the Cantar, by the way, which has been using a triple A/D converter for many years) you would probably boost the gain for the very quiet scenes. Then there's an unexpected shout and this may hit the limiter or clip. With NeverClip this is less likely to happen while still retaining full fidelity on the quiet and loud parts. And since there's no gain change, the background doesn't either, making it that little bit easier for post

[studiomprd]

" With NeverClip this is less likely to happen "

and I agree...

[bralleput]

Wow Mike!You agree?!

[Actualsizeaudio]

Actually Pro Tools was 48bit fixed for PTHD, and 32bit floating for regular native PT. Now in PT10 it's all 64bit floating.  Metric Halo ULN8s are 80bit fixed. You can very easily clip the mix bus of the older native Pro Tools. Maybe not with one track, but once you start piling on tracks it clips quick in my opinion. The Metric Halo boxes are another story, they have crazy headroom at the mix bus, and mixes do sound better summed inside one than they do when summed though a 32bit floating point mix bus. 

 

For a 12 track recorder I would think a 32bit bus would be plenty, and highly advantageous. Thanks for the diagram, it makes the concept much clearer. 

 

The limiters, and the compressors on the unit are fantastic IMHO. Very powerful for field work, and so much more transparent than any analog field mixer I have ever used. 

Thanks Senator, for being a jerk.
The Nomad's 32 bit mix engine is noted on the website.  It is also how ProTools and many other DAWs run their internal processing, making it almost impossible to clip the internal mix busses.
 

[Christian Spaeth]

since there's no gain change, the background doesn't either, making it that little bit easier for post

But there is a gain change, at least when you use Neverclip on the ISOs. These are reduced in gain by the amount you specify on the recorder. So it is actually more work for post, but with the benefit of non clipping tracks.

[Constantin]

But there is a gain change, at least when you use Neverclip on the ISOs. These are reduced in gain by the amount you specify on the recorder. So it is actually more work for post, but with the benefit of non clipping tracks.

That's true, but it's not what I meant. On a regular recorder you'd boost the gain for really quiet scenes, but then there's a problem when there's just one loud sound. So you'd either momentarily tear the gain down, or let hit the limiter. In either case it may be tricky for post to fix it, and it will take longer than boosting everything by a set amount. Or you'd set your levels for that one loud sound, but then for the quiet stuff you might be getting close to the system noise. So in this case I wouldn't really call it a gain change

[Victor Rubilar]

neverclip was partially explained time ago, in an older post, I make a sintesis here for clarification of the theory.

after the preamp, the signal is splited in 2, then each copy goes to a separated converter, both comverters identical and sinchronised, except for one of them has a preamp (about 20 db) inserted before its imput. That leaves you with 2 identical digital audio signals but with diferent level, one High and one Low (20 db diference), both with the "native" noise spec of a normal 24 bit design

I think Never clip works this way.

- the system now imput the High signal to the 32bit float mixer.
-Because It is digital, there is a previous signal buffer where the system can Look ahead, so the system knows when the High level signal comes near its cliping level, and when the signal reach a predefined level threshold about 3 db from the 24 bit cliping level, the system seamlesly switch the input signal to the Low (still far from the cliping level) signal, but adds it a digital gain of about 20 db, so the wave form continue from the point where the High singnal was stoped to be inputed to the mixer. When the Low signal descent below the predefined threshold the sistem switch to the High signal again(I think this is the aplying patent of zaxcom).


So the system ends with a 32 bit float signal (made of sections of the Low signal with 20 db of digital gain, and sections of the High signal) with a dinamic range exeding 20 db of a single converter design.

the final 32 bit float sections of the waveform with the Low level signal has 20 db higer noise floor (because of the digital gain) than the sections made of the High level signal, so the noise floor goes up and down with the switching of signal, but it only goes up when the sound material level is hot (about 20 db below 24bit clipping and higher) so the "increased" noise is allways inaudible (masked). This is realy not an issue comparing with a normal system because this sections with higer noise still have the normal noise spec for a single converter design. And the sections with the lower noise now have 20 db lower noise compared to the maximum posible peaks of the final 32 bit float audio.

Finaly when recording with Iso atenuation you get a 24bit signal with a noise floor that is normal for hot signals, and is 20 db lower for
low signal.

 

As I understood it NeverClip is exactly for the very quiet sounds. The point being that with any other recorder (except for the Cantar, by the way, which has been using a triple A/D converter for many years) you would probably boost the gain for the very quiet scenes. Then there's an unexpected shout and this may hit the limiter or clip. With NeverClip this is less likely to happen while still retaining full fidelity on the quiet and loud parts. And since there's no gain change, the background doesn't either, making it that little bit easier for post

Partially agree. The most importan use,  why they call it Never Clip,is for hig dinamic range material you are far from clipping the signal in the A/D conversion process. You can record 20 db lower than before but manteining the normal noise level, and avoiding clipping for suddenly increasings of the sound material level.
the use for recording quiet sound is less important since the microphones introduce higer noise than the recorder, for that aplication is preferable a less noise microphone, If the signal is quiet and stable then the system will never switch the converters, and you will get a normal quality recording, so Never Clip is for Hig dinamicaly changing material .

[Christian Spaeth]

Good summary, Victor!

[VM]

Then there's an unexpected shout and this may hit the limiter or clip. 

Good point, thanks. I have to hear it now.

[studiomprd]

anyone else notice who has not posted in this thread ??

[Mark Orusa]

Victor, you just gave away the secret, patent-pending Neverclip process! 

 

 

Mark O.

[VM]

anyone else notice who has not posted in this thread ??

Yes !

[Boomboom]

... Too occupied doing promotion in other brand-related posts maybe ? lol

[Victor Rubilar]

Victor, you just gave away the secret, patent-pending Neverclip process! 

 

 

Mark O.

 

well, not so secret I think. I actually dont know the official neverclip proccess, I just made conclusions from the info other users posted in this forum and what Mr. Glen Sanders said in his video of Never Clip.  From the available info anyone can conclude the same. I did a sintesis just for the peole that still does not believe that Never Clip works.  

By the way, If my conclusions are the real deal, the tecnological advancement of never clip is  making the process automatic. The use of 2 converters with diferent preamp level setings is a common practice in Live event recordings. The diference is that you record the  2 signals in separate tracks. Then in Post you replace the clipped sections with the "lower level safety track" and make up the level .

[Mark Orusa]

Yours is the first explanation, and a good one at that, of what I believe is the most critical part of the NeverClip system. This part of NeverClip is always active and thus always affecting a mic input. It is where the noise of the system (i.e. Nomad self noise, preamp noise) comes into play. It is also the area where a competitor would claim lies weaknesses of the system, or at least refute the claims of benefits of the system.

 

Rob Finch (username rofin) of SQN seems to one of the few who are willing and capable of pursuing technically relevant questions regarding the true benefits of a system like NeverClip. He asks some valuable questions (that still haven't been answered):

 

Mark O.

[Victor Rubilar]

Yours is the first explanation, and a good one at that, of what I believe is the most critical part of the NeverClip system. This part of NeverClip is always active and thus always affecting a mic input. It is where the noise of the system (i.e. Nomad self noise, preamp noise) comes into play. It is also the area where a competitor would claim lies weaknesses of the system, or at least refute the claims of benefits of the system.

 

Rob Finch (username rofin) of SQN seems to one of the few who are willing and capable of pursuing technically relevant questions regarding the true benefits of a system like NeverClip. He asks some valuable questions (that still haven't been answered):

 

Mark O.

 

Mr. Rob Finch made good questions, but he is confused by the lack of info. He see the picture only from a conventional point of view, and Mr Sanders doesnt respond clearly, Maybe I spoiled his secret, i dont know. I didnt intended that. The noise thing was spoiled by other user baguely, he said somthing like "the extra noise is not problem for the things are going so hot", so I concluded the whole thing. But the thing is partially like R. Finch wrote: 2 converter with difernt gains.       He is just not thinking in 32 bit float, that is why he doesnt understand.

 

Mr. Finch ask for the maximun imput level, and Mr. Sander says that is irrelevant  XD (what a troll response). the thing is that de maximum imput level for the system is the same, but the system works " as if " it can accept a voltage 20 db higher, but in realyty ther is a pad in  one converter, or an amp, or beter said, a diferent sensitivity set up of the analog circuitry in one of the converters (maybe a diferent feedback resistor) so both paths, high and low remains very similar in sound quality.

  So for the user, the imput works as if it can accept 20 db more voltage, Thus, extending the dinamic range, and sending real 32 bit float amplitud signals to the internal mixer.  the confusion that Mr Finch has is due to him assuming the conventional formula of dinamic range and not the real "concept". He knows the process of the dual converter high and low path, but he has not thank on it in 32 bit float, where you can make up the gain of the low path, to continue the wave form from the limits of the 24 bit word.

[Mark Orusa]

I was just kidding about revealing NeverClip secrets. Glenn has consistently avoided explaining this aspect of NeverClip, claiming it's a patent pending process. As I understand the patent process, you have protection while a patent is pending, so his excuse is null.

 

Mark O.

[John Blankenship]

I was just kidding about revealing NeverClip secrets. Glenn has consistently avoided explaining this aspect of NeverClip, claiming it's a patent pending process. As I understand the patent process, you have protection while a patent is pending, so his excuse is null.

Not at all null. The "patent pending" period holds a competitive advantage. It can be a really big advantage should a competitor wish to try to circumvent the patent.

Whereas, once a patent is issued the patent information is available to anyone, not so during the "patent pending" period. During that time other manufacturers are on notice about patent protection but they don't yet know the specifics of the patent. That information is not universally available until the patent is issued, thus giving the manufacturer with the "patent pending" a market advantage during this period.