Passive summing boxes necessity vs wants

[Michael Tibes]

A common disconnect between digital mixing and analog, is that nominal 0VU signals on analog meters are 20+dB below clipping or saturation.  OdB(FS) in the digital domain is clipping, so adding together a bunch of hot digital signals will overload.

This might be common misunderstanding because it is only valid if you do a D/A conversion and the converter clips at 0dBfs (which proabably all normal D/As do). The headroom of the digital summing bus depends on the mathematical model, whether it's fixed or floating point math. For floating point busses the headroom is incredibly high, some hundred dB or so. In any imagianable situation it can't clip.
The theoretical resolution of a floating point bus is also much better than analog in the aspect that neither very low nor very loud signals get degraded. This still doesn't say anything whether analog summing sounds better or not though, but if digital was inferior then in my opinion it is not a headroom problem.  I believe that we actually like the imperfections of analog stuff

Michael

 

[JohnRoberts]

A common disconnect between digital mixing and analog, is that nominal 0VU signals on analog meters are 20+dB below clipping or saturation.  OdB(FS) in the digital domain is clipping, so adding together a bunch of hot digital signals will overload.

This might be common misunderstanding because it is only valid if you do a D/A conversion and the converter clips at 0dBfs (which proabably all normal D/As do). The headroom of the digital summing bus depends on the mathematical model, whether it's fixed or floating point math. For floating point busses the headroom is incredibly high, some hundred dB or so. In any imagianable situation it can't clip.
The theoretical resolution of a floating point bus is also much better than analog in the aspect that neither very low nor very loud signals get degraded. This still doesn't say anything whether analog summing sounds better or not though, but if digital was inferior then in my opinion it is not a headroom problem.  I believe that we actually like the imperfections of analog stuff

Michael

yes, even a fixed point digital sum bus has a carry (overflow) bit that could be used for another 6 dB of dynamic range.

My point was more about the nominal 0VU for analog systems being much cooler than digital that is referenced below full scale.

Hopefully as modern digital paths get better operators run the stems in cooler.

Within reason a digital mixer could detect for mix bus saturation and re-scale the rest of path to prevent clipping, at least until it consumes the full output dynamic range.

My point is that digital combining is arbitrarily good so passive summing seems like fashion decision. At least from a technical/performance perspective.

JR

 

[ruffrecords]

My point was more about the nominal 0VU for analog systems being much cooler than digital that is referenced below full scale.
JR

I can understand why people regularly become confused over 0VU and 0dBFS but I think this is mostly the fault of DAW manufacturers.

Many analogue systems work at a nominal 0VU = +4dBu and have (at least)  16dB of headroom before clipping. Many use VU meters of one sort or another combined with a LED that flashes on peaks at some level approaching clipping.

So, I have a proposal for a standard that should eliminate all the confusion. Let us say that all analogue systems have a 0dBFS equal to +20dBu. All digital system should be calibrated so that their 0dBFS occurs with a +20dBu input and 0dBFS also creates a +20dBu output. All DAW meters should then be calibrated to show 0VU at -16dBFS and so should all analogue systems (most already are). Everybody then uses 0VU as their nominal operating level with an allowance of 16dB for peaks.

I am surprised some august body like the AES has not already proposed something like this.

Cheers

Ian

 

[JohnRoberts]

My point was more about the nominal 0VU for analog systems being much cooler than digital that is referenced below full scale.
JR

I can understand why people regularly become confused over 0VU and 0dBFS but I think this is mostly the fault of DAW manufacturers.

Many analogue systems work at a nominal 0VU = +4dBu and have (at least)  16dB of headroom before clipping. Many use VU meters of one sort or another combined with a LED that flashes on peaks at some level approaching clipping.

So, I have a proposal for a standard that should eliminate all the confusion. Let us say that all analogue systems have a 0dBFS equal to +20dBu. All digital system should be calibrated so that their 0dBFS occurs with a +20dBu input and 0dBFS also creates a +20dBu output. All DAW meters should then be calibrated to show 0VU at -16dBFS and so should all analogue systems (most already are). Everybody then uses 0VU as their nominal operating level with an allowance of 16dB for peaks.

I am surprised some august body like the AES has not already proposed something like this.

Cheers

Ian

It would be simple enough if digital paths also provided absolute (output) metering capability, but they don't seem to care... At this point it is quickly becoming academic.

JR

 

[Andy Peters]

So, I have a proposal for a standard that should eliminate all the confusion. Let us say that all analogue systems have a 0dBFS equal to +20dBu. All digital system should be calibrated so that their 0dBFS occurs with a +20dBu input and 0dBFS also creates a +20dBu output. All DAW meters should then be calibrated to show 0VU at -16dBFS and so should all analogue systems (most already are). Everybody then uses 0VU as their nominal operating level with an allowance of 16dB for peaks.

I am surprised some august body like the AES has not already proposed something like this.

That all makes perfect sense, and it faces the same kind of inertia we saw when they tried to decide whether Pin 2 or Pin 3 was hot. At least in that case, it's a binary option, so flipping a coin or spinning a wheel was a reasonable selection method.

 

[ruffrecords]

So, I have a proposal for a standard that should eliminate all the confusion. Let us say that all analogue systems have a 0dBFS equal to +20dBu. All digital system should be calibrated so that their 0dBFS occurs with a +20dBu input and 0dBFS also creates a +20dBu output. All DAW meters should then be calibrated to show 0VU at -16dBFS and so should all analogue systems (most already are). Everybody then uses 0VU as their nominal operating level with an allowance of 16dB for peaks.

I am surprised some august body like the AES has not already proposed something like this.

That all makes perfect sense, and it faces the same kind of inertia we saw when they tried to decide whether Pin 2 or Pin 3 was hot. At least in that case, it's a binary option, so flipping a coin or spinning a wheel was a reasonable selection method.

I am an AES member. I'll contact them and propose it if nothing else just to see what reaction I get.

Cheers

Ian

 

[JohnRoberts]

So, I have a proposal for a standard that should eliminate all the confusion. Let us say that all analogue systems have a 0dBFS equal to +20dBu. All digital system should be calibrated so that their 0dBFS occurs with a +20dBu input and 0dBFS also creates a +20dBu output. All DAW meters should then be calibrated to show 0VU at -16dBFS and so should all analogue systems (most already are). Everybody then uses 0VU as their nominal operating level with an allowance of 16dB for peaks.

I am surprised some august body like the AES has not already proposed something like this.

That all makes perfect sense, and it faces the same kind of inertia we saw when they tried to decide whether Pin 2 or Pin 3 was hot. At least in that case, it's a binary option, so flipping a coin or spinning a wheel was a reasonable selection method.

I am an AES member. I'll contact them and propose it if nothing else just to see what reaction I get.

Cheers

Ian

I don't want to poison the well but it isn't very easy.

I wasted a couple years trying to get the AES standards group to help resolve a concise definition for Q/bandwidth of boost/cut EQ sections ( a major problem with published EQ curves for loudspeaker crossovers). Right now different DSP platforms can return different results depending on the filter approach. In fact this lack of standard definition pre-dates digital, all the sundry analog 1/3 octave equalizers deliver different Q/bandwidth despite claiming to be the same 1/3rd octave wide.  (Well they aren't every one different but there are a handful of different filter topologies that deliver different results).

I got as far as having them accept my submission to discuss at a standards meeting (they usually meet during AES shows or conventions.) I think my request got bumped down  to some standards sub comittee (if you check the AES website you can see how many their are and read the minutes). I tried to follow the meeting reports for a few years then gave up.

If I was still in that industry I might get more aggressive, but generally it's a little like politics where you need to build a coalition of enough (market share) manufacturers.

I recall any number of failed attempts to come up with sundry industry standards. Midi ended up being adopted by a handful of large manufacturers and then the industry followed. VHS just kicked BetaMax's proverbial but in the market.

Pick a more important fight, because they are not sitting around looking for new ideas to pick up and run with. I thought a definition for Q was important and lost energy to pursue it.

JR

 

[ruffrecords]

I don't want to poison the well but it isn't very easy.
JR

Quite likely. However, I suspect an large number of users will be interested in such a standard and if I can mobilise enough of them it will help. I already made a start by posting on SOS.

Cheers

Ian