Something new from Zaxcom

[Wandering Ear]

Hmm, so at the receiver what is it you see if you use the neverclip option? Does it just show an attenuated signal? But those bottom four bits are reintroduced at the receiver output if it is analog...so you have an attenuated signal with those four bits reintroduced from further down the signal chain...

Neverclip is a system using two AD's at the input to increase the dynamic range of the input. It is not an option, and there is no indication for it. Think of it as a high dynamic range AD converter.

Sounds like you are confusing it with iso attenuation in the nomad, which is different. That allows signals above 0dbfs to be recorded by attenuating the iso track.

[Tim M]

With the addition of NeverClip to our body packs we are going to be giving 2 options to the production sound mixer on how to use the function.

 

Option 1: Use the soft knee compressor in the transmitter to keep the added dynamic range in the proper range for recording to a mix track. (This is how the TRX742 is set up now. Please see the Glen Trew review in Sound and picture for his comments on sound quality)

 

Option2: Use iso track attenuation in the transmitter to fit the entire dynamic range( about 123 db in this product) into the transmitter channel and onto the backup recorder in the transmitter. This option does not use any Limiters, compressors, companders or noise reducers. It will transmit the full 123dB dynamic range as if via a hard wire to the receiver producing an output from the receiver that can never clip.

 

The new transmitter versions are fully Zaxnet compatable with remote control, wireless QC audio transmission, Power Roll  TM  (Pat Pending) and internal backup recording (Patented).

 

The new transmitters are now in beta test and will be delivering in the next few weeks. The NeverClip TM (Pat Pending) audio structure in our transmitter is an important advancement in the art of wireless microphone audio. We will have demo units out to our US dealers soon so all can hear what it can do. Please wait to contact our dealers until our official Zaxcom product release is announced.

 

Glenn

Well yes it is there but how you utilize it is the question, as Glenn states in the quoted post. It seems if you go analog the best option is to use it as a soft knee.

[Wandering Ear]

Well yes it is there but how you utilize it is the question, as Glenn states in the quoted post. It seems if you go analog the best option is to use it as a soft knee.

possibly, but that is assuming the da and analog outs have a significant enough noise floor to add to the attenuated track. I don't actually know that.

[Tim M]

possibly, but that is assuming the da and analog outs have a significant enough noise floor to add to the attenuated track. I don't actually know that.

Well true, but that combined with the analog inputs on the Nomad and I think you've got enough noise to counteract using the attenuation selection in the transmitter...actually this option and then going back to analog sounds worse than if you were to use a normal lectro full range transmitter. You've got a low level recording with noise introduced back in. We all know the preamps on at least the Nomad can be downright dreadful at times, especially the 2nd one. This is all assuming the Nomad will be your main recording and the recording on the transmitter won't be used. To me it still sounds like using the soft knee option (which is still better than any option on the market)  is the way to go if you have to go back down to analog for the remaining inputs after the initial 6.

[John Blankenship]

...

We all know the preamps on at least the Nomad can be downright dreadful at times, especially the 2nd one.

...

 

Apparently, we don't all know that, as this is the first I've heard of this assertion -- and I've owned a Nomad for quite a long time now.

 

This has certainly not been my experience, so if this is the case with your Nomad, it needs a trip back to Zaxcom.  

[Wandering Ear]

We all know the preamps on at least the Nomad can be downright dreadful at times, especially the 2nd one.

I find the pre-amps on the nomad to be excellent. Sounds like you might have a hardware problem if your #2 pre sounds bad.

However, i don't know why you would run at mic level if you are running analog, and i doubt the line level is adding as much noise as you are suggesting.

[Zack]

Dreadful? Wow, what would a H4n classify as then? Nomad has always sound very clean to me.

[Tim M]

I find the pre-amps on the nomad to be excellent. Sounds like you might have a hardware problem if your #2 pre sounds bad.

However, i don't know why you would run at mic level if you are running analog, and i doubt the line level is adding as much noise as you are suggesting.

Actually I had a talk with Glenn about this half a year ago. He said that the 2nd pre amps on the Nomad are known for being a little louder than the others...and yes this was using line level not mic level. It does need a trip back for more than one reason, but I've always been too nervous to do so unless it is a crippling issue.  Still though, back to the subject matter if the last four bits are reintroduced through analog processes it makes sense to go with the soft knee option, unless somebody can explain that is irrelevant etc etc.

[RScottATL]

The preamps of ANY modern professional mixer input FAR surpass the preamps of any wireless input, so no you don't need Neverclip input preamp to experience the benefit that a Neverclip wireless system has to offer. The wireless preamp stage is constantly the weakest link in the entire signal path and Zaxcom claims to have solved it with Neverclip transmitters.

AES eliminates the Neverclip preamps of the Nomad, but more importantly they are eliminating D/A and A/D processes which aren't doing your signal any favors, from a digital purist point of view. In this entire process, though, the most gains are coming from the Neverclip preamps because they are eliminating the weakest link in the chain.

[Mark Orusa]

In my attempts to understand the implementation of NeverClip in the TRX900, I made a signal flow diagram:

(Click to view/download a PDF in Google Drive)

The diagram is based off of information given by Glenn ( and ) and my own assumptions. As many aspects of it's design and operation have not been revealed, it is most likely incorrect in some way.

If anyone has a correction or something to add, please let me know,

Mark O.

[studiomprd]

Tim: " the preamps on at least the Nomad can be downright dreadful at times, especially the 2nd one. "

HUH ?

" but I've always been too nervous to do so unless it is a crippling issue. "

and HUH ?

if the gear needs servicing, we get it serviced!  that is part of being a pro...

 

Mark: " although the input range of 126 dB fits... it is over the 0dBfs reference... "

HUH ?

" it is most likely incorrect in some way. "

most likely...

[Mark Orusa]

Mark: " although the input range of 126 dB fits... it is over the 0dBfs reference... "

HUH ?

" it is most likely incorrect in some way. "

most likely...

I'm assuming the output of the transmitter is a 24 bit audio signal with a maximum SNR of 144db. With only a 126db input range, why do we need to compress or attenuate the signal?

 

As the name implies, NeverClip is meant to prevent a signal from clipping the preamp. On the surface that is all you need to know, but if you dig deeper you have to acknowledge the particular relationship between the preamp and the AD converters. It takes two AD converters to handle the 126db input range, and with Zaxcom's implementation (a special, secret way of combining the two signals) they create a digital signal greater than the 144db limit of a 24 bit audio signal. Why, exactly, only Zaxcom can say.

 

NeverClip could just shift the top end of the 126db input signal to the top of the 144db output, but it doesn't. If it did we wouldn't need to think about compressing or attenuating the output before recording or transmitting.

 

Mark O.

[Ze Frias]

In my attempts to understand the implementation of NeverClip in the TRX900, I made a signal flow diagram:

 

(Click to view/download a PDF in Google Drive)

 

The diagram is based off of information given by Glenn ( and ) and my own assumptions. As many aspects of it's design and operation have not been revealed, it is most likely incorrect in some way.

If anyone has a correction or something to add, please let me know,

Mark O.

 

 

Hey Mark,

 

Love the visual element your graph provides. Just a few points/questions:

- I thought the Zaxcom wireless used digital transmission exclusively. As such, why does the signal go through a DAC before hitting the RF Amp?

- For that matter, the audio recorders in the transmitters are digital as well, why are they also after the DAC at the end of the transmitter's signal path?

- In the paragraph you state "The TRX 900LANC uses two digital to analog converters to allow for greater input signal...". I'm sure you meant that it uses two analog to digital converters, not digital to analog converters.

 

I'd love to clarify how NC works in the Zaxcom wireless systems as well. I think I have a firm grasp of how it works in the Nomad and Maxx per previous threads that go well into detail. I'm just not too sure in the wireless.

[glenn]

I think I was confused by that diagram ;-)

 

Glenn

[Mark Orusa]

Thanks Jose, I had a bit of a brain fart. Yeah, no DA on the end, plus I moved the limiter where it should be - near the preamp.

 

(Download as a PDF)

 

To help understand why the digital signal must be attenuated before transmission or recording, it is helpful to consider a non-NeverClip transmitter. The TRX900LA has a 106 db input range. A signal reaching the limit of the transmitter's input will be at the limit of the 24 bit audio signal, 0 dbFS. Now, throw in NeverClip with it's additional 20 db input range and you end up with a signal as much as 20 db over 0 dbFS.

 

Glenn, feel free to share your knowledge. You don't have to share the secret sauce recipe, just a little bit about gain structure and operation. 

 

Mark O.

[glenn]

The dynamic range of the converter that goes to 126 dB is only 106 in the wireless and should start 20 dB higher than the lower converter.

 

Glenn

[RScottATL]

Nothing is ever "over" 0dBFS. The point of reference can be shifted down from the typical -20. Glenn has often described the NeverClip system as adding usable range into what is normally the noise floor of the ADC and preamp. This allows us to set our typical recording levels much lower without any fear of having a noisy signal, and thus gives us more headroom. 0dBFS never changes. The 24 bit container is perfectly fine for accepting audio from the Neverclip signal path.

[Mark Orusa]

If the DSP in the TRX is anything like Nomad it operates as a 32bit floating point system which has a SNR of 192 db. Similar to a NeverClip signal in Nomad, the NeverClip signal in a TRX must be attenuated before it reaches an output.

 

My question is why is it designed so the level from NeverClip has to be attenuated? The 126 db range fits inside the 144 db range of a 24 bit signal, so why the attenuation?

 

Mark O. 

[RScottATL]

Mark O, the only way to add "headroom" to a signal is to attenuate it, as there is nothing above 0dBFS. It's the classic problem of why a CD with more dynamics has to play back "quieter" than one with no dynamics.

It all depends where in that 144db package those 126db fall. Usually, digital audio works from the top down, only giving as much headroom as needed, hence, why digital clipping must be prevented with a limiter. As well, you're not just playing in the digital space as you're working with A>D converters that add more noise as they deal with finer signals, so there are electrical reasons to keep things "maxed out".

The end reason why there needs to be an attenuation at the end may simply be because of the order of operations used in the DSP. Maybe one attenuation at the end is more DSP efficient than two attenuations on the individual A>D signals prior to combining. Who knows. It's part of the sauce, so I'm sure there's a good reason for it.

I wouldn't assume the TX uses the same architecture as the Nomad mix engine. The Nomad is a mixer so that 32bit floating point is important for summing a lot of 24 bit tracks. The transmitter has no such responsibility.

Is it really an issue that it needs to be attenuated to work properly? Just add some gain. If you don't need the full 126db of range, you can always use a smaller Neverclip dynamics setting, for instance if your input preamp is the concern. If you're just adding gain digitally, though, there will be no issue.

What problems are you concerned about because of the attenuation?

[Constantin]

Nothing is ever "over" 0dBFS. The point of reference can be shifted down from the typical -20.

Thank you for this. I have stated this many times in various NeverClip threads, but somehow the idea of NeverClip being able of recording above 0dBFS seems to have manifested itself.

[Victor Rubilar]

Thank you for this. I have stated this many times in various NeverClip threads, but somehow the idea of NeverClip being able of recording above 0dBFS seems to have manifested itself.

it canot record above 0 dBFS cause the system is designed to work at a standard 24 bit .

0 dBFS correspond to a 24 bit word full of 1s. but never clip  works as if you can input over 0dBFS. Indeed the mixer receives a reconstructed signal from two 24 bit  converters, at 32 bit float (i.e.  24 bit multiplied by an exponent) . This 32 bit Float reconstructed signal has more than 24 bits, and represents analog voltage levels higher than the 24 bit converters clipping voltage level  (i.e. represents levels over 0dBFS), and it needs to be atenuated prior to be recorded into a 24 bit format and prior to transmission.

 

  The diagram lacks a 20 dB amp  before the high AD converter, or instead, a  20 dB pad before the lower converter.

  After the conbiner the signal is 32 bit Float. I explained the conbiner process in the other thread.

  the atenuator/compressor  fits the "big" 32 bit float signal into 24 bits, and then ignore the exponent, to convert the signal to normal 24 bits.

  the dinamic range diference with the big recorders  is due to the lower  quality of the preamp and converter of the wireless system.

 

  the level graphs are confusing, I recomend  this for the input:

 

- a bar representing 144 db (total of 24 bits), signaling 38 db of floor noise and 106 db dinamic range above the      noise floor.

- then a " + " sign.

-then another bar like the first.

-then a " = " sign.

-then a bar representing 164 db (144+20, maximum of neverclip system), signaling two levels of posible noise floor at 38 and 58 dB. and a text saing it depends on the signal going lower or higher than 144 dB.

 

  for the output:

 

- a bar of 164 dB like the last one

- a text : " 20 dB atenuation"

- a bar of 144 dB, signaling two levels of  posible noise floor, at 18 and  38 dB.

[Victor Rubilar]

actually, Howy told me these new transmitters can be used as non-neverclip ones via software...

it must be posible deactivating the signal switching in the software driven conbiner, leaving only the high path. I dont know if it is posible with the available software.

[Constantin]

then a bar representing 164 db (144+20, maximum of neverclip system)

where do you get the 164dB from? The dynamic range of NeverClip enabled inputs is 137dB or 126dB for transmitters.

[Max Futterman]

And 24-bit audio has a mathematical limit 144dB.

[studiomprd]

... and "the diagram"  is only speculation...