Jump to content

Help with understanding Zaxcom ZHD wireless system


Mike H

Recommended Posts

2 hours ago, Jason Todd said:

Interesting. So the range boost in XR mode might be due to better hardware on the LA3's and won't necessarily give me a range boost using the newest firmware on my QRX200's and LA2.5's?

Not sure. Just spewing some thoughts based on what I remember Howy said.

Link to comment
Share on other sites

  • 6 years later...

I hate to bring up a topic that caused a flame war last time, but I'm struggling to find a definitive answer about how Zaxcom wireless interacts with sample rate, and how to interpret the sample rate specs on some of their transmitters.

I've seen several (older) reports that Zaxcom wireless samples at 32KHz, which therefore caps the frequency response of those units at 16kHz.  What I'm not clear on is whether this is true across the board, if it applies only to specific modulations (XR?), or specific transmitters.

Zaxcom doesn't seem to specify much about their wireless modulations (the best information is here:  https://zaxcom.com/modulation-overview/), but they do specify ADC audio specs for their transmitters:

  • ZMT4-x:  24 bits / 48 kHz
  • ZMT4:  32 bits / 32 kHz
  • TRXLA5:  24 bits / 48 kHz
  • TRX743:  24 bits / no sample rate specified

For the most part, there's no sample rate or frequency response specs for the receivers, but the RX200 lists:  

  • AES3 balanced output 32 kHz
  • Frequency Range 20Hz to 16kHz

RX200 is intended as a camera receiver, so I wouldn't assume that all the receivers follow this spec, but it does seem to confirm that Zaxcom uses 32kHz sample rate at least some of the time.

That summarizes the official information I've been able to find.  What I'm wondering is:  Are the ADC specs on the transmitters a proxy for the sample rate they are transmitted at, or would their frequency responses always be truncated to 16kHz for transmission anyway (in which case, why bother specifying the *ADC* numbers)?  I notice that 32 bits @ 32kHz requires roughly the same bandwidth as 24 bits @ 48kHz, so I can see how it might work technically, but I haven't been able to find any firm information about the sample rate during transmission.

I'm particularly interested in how this might affect the TRX743 vs ZMT4 — I would pick the dedicated TRX for my boom over the flexibility of the ZMT4 if I knew the frequency response was better.

I would also like to know about the various modulations:  Does ZHD (48 or 96) sacrifice the top end frequencies to fit audio into a smaller bandwidth?  In that case, would XR be a better choice for fidelity?

I don't want to restart the flame war about whether this "matters" or not.  What I'm interested in is knowing the capabilities and limitations of the equipment that I'm buying.  I would judge that the difference between 32kHz and 48kHz is mostly academic for typical recording situations ... but there's a couple situations where I *would* care about it (mainly, specialized SFX recording, and classical music, both of which I occasionally have call to work with).  So, faced with a situation where the lower frequency response of a 32kHz sample rate might matter, I'd like to know which, if any, Zaxcom equipment I would want to choose or avoid.

Link to comment
Share on other sites

3 hours ago, The Documentary Sound Guy said:

I hate to bring up a topic that caused a flame war last time, but I'm struggling to find a definitive answer about how Zaxcom wireless interacts with sample rate, and how to interpret the sample rate specs on some of their transmitters.

I've seen several (older) reports that Zaxcom wireless samples at 32KHz, which therefore caps the frequency response of those units at 16kHz.  What I'm not clear on is whether this is true across the board, if it applies only to specific modulations (XR?), or specific transmitters.

Zaxcom doesn't seem to specify much about their wireless modulations (the best information is here:  https://zaxcom.com/modulation-overview/), but they do specify ADC audio specs for their transmitters:

  • ZMT4-x:  24 bits / 48 kHz
  • ZMT4:  32 bits / 32 kHz
  • TRXLA5:  24 bits / 48 kHz
  • TRX743:  24 bits / no sample rate specified

For the most part, there's no sample rate or frequency response specs for the receivers, but the RX200 lists:  

  • AES3 balanced output 32 kHz
  • Frequency Range 20Hz to 16kHz

RX200 is intended as a camera receiver, so I wouldn't assume that all the receivers follow this spec, but it does seem to confirm that Zaxcom uses 32kHz sample rate at least some of the time.

That summarizes the official information I've been able to find.  What I'm wondering is:  Are the ADC specs on the transmitters a proxy for the sample rate they are transmitted at, or would their frequency responses always be truncated to 16kHz for transmission anyway (in which case, why bother specifying the *ADC* numbers)?  I notice that 32 bits @ 32kHz requires roughly the same bandwidth as 24 bits @ 48kHz, so I can see how it might work technically, but I haven't been able to find any firm information about the sample rate during transmission.

I'm particularly interested in how this might affect the TRX743 vs ZMT4 — I would pick the dedicated TRX for my boom over the flexibility of the ZMT4 if I knew the frequency response was better.

I would also like to know about the various modulations:  Does ZHD (48 or 96) sacrifice the top end frequencies to fit audio into a smaller bandwidth?  In that case, would XR be a better choice for fidelity?

I don't want to restart the flame war about whether this "matters" or not.  What I'm interested in is knowing the capabilities and limitations of the equipment that I'm buying.  I would judge that the difference between 32kHz and 48kHz is mostly academic for typical recording situations ... but there's a couple situations where I *would* care about it (mainly, specialized SFX recording, and classical music, both of which I occasionally have call to work with).  So, faced with a situation where the lower frequency response of a 32kHz sample rate might matter, I'd like to know which, if any, Zaxcom equipment I would want to choose or avoid.

The 743 has a nicer preamp than the zmt, my zmt3 phs are noisy to the point that I don't use them for interior drama, let alone classical - don't know about zmt4s - I thought the cut off was 15khz - nothing a nice cable wouldn't fix

 

 

Link to comment
Share on other sites

1 hour ago, jozzafunk said:

The 743 has a nicer preamp than the zmt, my zmt3 phs are noisy to the point that I don't use them for interior drama, let alone classical - don't know about zmt4s - I thought the cut off was 15khz - nothing a nice cable wouldn't fix

 

 


For the 15kHz cutoff, are you referring to the 743, the ZMT4, or all Zax wireless?

And, from what I've read, one of the selling points of the ZMT4 is supposed to be a better preamp.  But I trust and appreciate having your first-hand knowledge of the 743

Link to comment
Share on other sites

Pretty sure zaxcom is always 32kHz . If you use XR. But who cares?. I would not recommend using zhd for range.   Just don’t make sound for children shows. Adults can not hear over 16khz anyways. 
As far as noise goes I am not as concerned about it.Honestly no preamp should be noisier then the condenser mic. 
even $20 once. 

But fast transient loud peaks are handled a lot better by 74x transmitters. 
I believe the ZMT never clip is not “real” dual AD. It employs a compressor in order to work. 

Link to comment
Share on other sites

Thanks.  That's useful info.  It's all about knowing the gear.

Re:  ZMT neverclip:  I'm wondering whether the 32 bit / 32 kHz spec in the ZMT is part of the Neverclip design.  If it's a 32bit *floating point* ADC, that would eliminate the need for the dual ADC design that is more common for Neverclip?  But if it's compressing things anyway, maybe not.  All I can do is speculate ... or hope someone else here has inside information.

Link to comment
Share on other sites

1 hour ago, The Documentary Sound Guy said:

Thanks.  That's useful info.  It's all about knowing the gear.

Re:  ZMT neverclip:  I'm wondering whether the 32 bit / 32 kHz spec in the ZMT is part of the Neverclip design.  If it's a 32bit *floating point* ADC, that would eliminate the need for the dual ADC design that is more common for Neverclip?  But if it's compressing things anyway, maybe not.  All I can do is speculate ... or hope someone else here has inside information.

Not how it works. The hardware has to match 32bit which is pretty much impossible.
about the 32bit: might be an advertising gimmick. 136dB dynamic range can be easily send over 24bit. And I doubt the 136dB is a realistic number. There is for sure an expander implemented.
32 bit over digital wireless sounds like insane amount of data even with the compression. 


 

Link to comment
Share on other sites

1 hour ago, RadoStefanov said:

Not how it works. The hardware has to match 32bit which is pretty much impossible.
about the 32bit: might be an advertising gimmick. 136dB dynamic range can be easily send over 24bit. And I doubt the 136dB is a realistic number. There is for sure an expander implemented.
32 bit over digital wireless sounds like insane amount of data even with the compression. 


 

Ok, yes, I'm glossing over some engineering details.  But, it's not without it's benefits; from what I understand, transmitting 32-bit float is basically how Sound Devices GainForward works (i.e. it eliminates gain staging at the transmitter by making sure it can't clip digitally and leaving enough analogue headroom that analogue clipping is also nearly impossible).  It's not that it avoids analogue clipping entirely or the inherent limitations of dynamic range, but it eliminates the "hard clipping" of exceeding 0dbFS, which is really the limitation that dual ADCs / Neverclip is intended to work around.

In reality both 32-bit float and Neverclip are just a case of choosing a more practical amount of headroom above "0", with some ingenious wizardry to expand dynamic range enough that the extra headroom doesn't cause noise floor issues.  Don't believe me?  The entire Neverclip workflow is premised on the fact that Zaxcom recorders use 32-bit float internally, which means they can process audio "above" 0dbFS without hard clipping before it hits the record tracks at 24 bits, and without pushing the signal closer to a 24-bit noise floor.  Presumably, they could also do this in 24-bit by simply leaving digital headroom above full scale on the meters ... but that wasn't the approach they took.

Link to comment
Share on other sites

11 minutes ago, The Documentary Sound Guy said:

Ok, yes, I'm glossing over some engineering details.  But, it's not without it's benefits; from what I understand, transmitting 32-bit float is basically how Sound Devices GainForward works (i.e. it eliminates gain staging at the transmitter by making sure it can't clip digitally and leaving enough analogue headroom that analogue clipping is also nearly impossible).  It's not that it avoids analogue clipping entirely or the inherent limitations of dynamic range, but it eliminates the "hard clipping" of exceeding 0dbFS, which is really the limitation that dual ADCs / Neverclip is intended to work around.

In reality both 32-bit float and Neverclip are just a case of choosing a more practical amount of headroom above "0", with some ingenious wizardry to expand dynamic range enough that the extra headroom doesn't cause noise floor issues.  Don't believe me?  The entire Neverclip workflow is premised on the fact that Zaxcom recorders use 32-bit float internally, which means they can process audio "above" 0dbFS without hard clipping before it hits the record tracks at 24 bits, and without pushing the signal closer to a 24-bit noise floor.  Presumably, they could also do this in 24-bit by simply leaving digital headroom above full scale on the meters ... but that wasn't the approach they took.

I bet you money it does not transmit 32bit. It records it at best and it does not make any difference compared to 24bit. 
by the way when they say 32 bit they actually mean 25 bit. 
its 24 bit with one bit floating. 

And by the way gain forward is “all due respect” the dumbest term of all time. 🙂

Link to comment
Share on other sites

I'm not betting.  I was speculating, and trying to figure out how it actually works.  But ... since you asked, Sound Devices actually does provide a somewhat comprehensible technical explanation here: https://www.sounddevices.com/gainforward-explained/
 

Here's the most relevant parts:

 

Quote

GainForward architecture differs from systems that require control of gain at the transmitter and which send a limited dynamic range signal over the RF link. Instead, GainForward transmits the entire dynamic range of the analog audio over the RF link and adjustments to gain are done at the receiving end. All analog FM wireless systems and most digital systems require gain control at the transmitter. GainForward systems do not.


 

Quote

 

Floating-Point RF Transmission

With GainForward, the signal’s entire dynamic range is preserved and sent digitally via RF over the air to the receiver using floating-point math. Somewhat similar to how 32-bit floating point audio files have increased dynamic range, a related “Floating-Point RF” method is used to send digital audio over the RF link. Audio samples are converted from fixed-point coming out of the ADCs into a floating point representation. At the receiver, the reverse process takes place and the floating point signal is converted back into fixed-point audio. This entire RF encoding and transmission topology is proprietary to Sound Devices. It has evolved over many years of work and listening tests, and is the very heart of every A20-Mini system.

 


So ... I guess it's floating point but not 32-bit.  I can see that being useful way of getting the "soft clipping" of floating point without pushing up bandwidth requirements needlessly.

I have no idea what you mean by 24-bit with 1 bit floating.  That would be mathematically identical to a 25-bit integer; adding a single bit "exponent" is the same thing as adding a binary digit.  More likely, they do floating point representation that is smaller than 32 bits.  24-bit float (FP24) is unusual but it exists, typically with a 16 bit base and 7 bit exponent.  Or, maybe they are using a custom floating point representation that suits audio better (I can see something like a 21-bit base and 3-bit exponent and no sign bit covering more dynamic range than would ever be needed, with the same "soft clipping" and only minimal loss in precision compared to 24-bit integer).

I agree GainForward is a dumb name.  I didn't choose it.

Link to comment
Share on other sites

Zaxcom wireless is always intentionally limited to 16KHz of audio frequency response.  We do this for multiple reasons. Better transmission reliability, higher bit resolution and most importantly to filter out all of the junk that can exist above 16KHz that can come from ultrasonic sources. Combine that with the fact there is no usable content there and its a no brainer to eliminate the interference that post does not want there in the first place.

 

Neverclip uses 2 A-D converters to provide 20 dB of extra headroom so you will "Never Clip".  Still with the 136 dB of dynamic range, it is possible to need the use of a compressor if the preamp gain is crazy hot. As a catch all we have a compressor just in case. It is seldom if ever used.

 

The Zaxcom transmitters have local and remote gain control. This provides a 5-6 dB noise floor advantage over wireless mics that have no control of the pre-amp gain in the transmitter.  Adjusting the gain of the transmitter signal forward of the transmitter pre-amp does nothing to restore the lost dynamic range of a transmitter with no pre-amp gain control.  The concept that the lost audio quality of  a mismatched transmitter preamp gain setting to a microphone output can be repaired or improved once the audio has left the transmitter is bogus. A low level microphone signal needs optimum gain to properly drive the A-D converter for the best noise and distortion performance.  A pre-amp with no gain control will still be sub par no matter how the good the signal path is to the receiver or where the gain is adjusted forward of the transmitter preamp. Just like in the old days, the trim control of a Cooper mixer made wireless input signals louder not better. 

Link to comment
Share on other sites

21 hours ago, The Documentary Sound Guy said:

I'm not betting.  I was speculating, and trying to figure out how it actually works.  But ... since you asked, Sound Devices actually does provide a somewhat comprehensible technical explanation here: https://www.sounddevices.com/gainforward-explained/
 

Here's the most relevant parts:

 


 


So ... I guess it's floating point but not 32-bit.  I can see that being useful way of getting the "soft clipping" of floating point without pushing up bandwidth requirements needlessly.

I have no idea what you mean by 24-bit with 1 bit floating.  That would be mathematically identical to a 25-bit integer; adding a single bit "exponent" is the same thing as adding a binary digit.  More likely, they do floating point representation that is smaller than 32 bits.  24-bit float (FP24) is unusual but it exists, typically with a 16 bit base and 7 bit exponent.  Or, maybe they are using a custom floating point representation that suits audio better (I can see something like a 21-bit base and 3-bit exponent and no sign bit covering more dynamic range than would ever be needed, with the same "soft clipping" and only minimal loss in precision compared to 24-bit integer).

I agree GainForward is a dumb name.  I didn't choose it.

Now let’s talk a little more about 32 bit 

there are 2 different formats:

32bit floating - which is equivalent of to 24bit

and

true fixed 32 bit.
 

if you do 32 bit floating point  you don’t have to dither since 32 bit floating point does not have higher bit depth then 24 bit.
32bit fixed is actual 32 bit and  has to be dithered down. 
 

Link to comment
Share on other sites

  • 2 weeks later...

Regarding the 32 bit spec on the page:

 

That’s only the A/D converter. Pretty much all modern A/D converters are 32-bit for a few reasons, one to maximize the accuracy when converting an analog sample to digital, and you can address more data per CPU cycle. It’s then converted to a 24 bit waveform transmitted over the air. 
 

32 bit wireless transmission makes no sense since a signal would be unreliable and likely have unacceptable latency to fit within FCC bandwidth limits. 
 

So in short, the 32 bit spec you read is something completely different. 

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...