Headroom Voltages Vs Sonics - More headroom isn't always a good thing?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Electrobumps

Well-known member
Joined
Aug 12, 2008
Messages
282
Location
LA
I'm an Engineer and Producer and electronics understanding is basic in terms of many on this forum.  I see designers that are changing circuits to 24v rails from 16v rails to offer increased headroom.  Without any justification other than increased headroom.

I understand why we need headroom and there is a lot of technical info to read and debate.  However I'm more interested in the sonics of headroom or more importantly lack of headroom. 

Lets take a Neve VR, wonderful console to work and we can slam the mix bus and it can sound great.  Do the same on an SSL J,  it will start clipping in a very nasty way.  Also a great console but needed more headroom.

1073 pre I will often crank the gain on a snare for some crunch and back the output right back as not to clip my DAW,  "technically" on paper stupid.  sonically this can sound amazing.    Do this on an SSL mic pre, it will sound like a disaster.

So here is my point, if we extended the headroom on a Neve VR mix buss the sound we get as we max it out would become less likely without having to gain everything up to reach that point.  On the other hand an SSL J series it would be useful to have more headroom and they did increase the headroom on the K Series.

I'm of the opinion when making clones of classic audio gear don't mess with the headroom.  Some of the best creative sounds  can be when you are smashing everything way beyond what would be considered normal operating levels and the saturation and distortions are truly wonderful.  Why put these out of reach with increased headroom?

So extra headroom can be useful, but adding it for the sake of having more headroom may well be detrimental the sonics of that gear that make it iconic.




 
Electrobumps said:
I'm an Engineer and Producer and electronics understanding is basic in terms of many on this forum.  I see designers that are changing circuits to 24v rails from 16v rails to offer increased headroom.  Without any justification other than increased headroom.
This is a very old marketing conceit and might have made a little sense several decades ago when electronic paths had higher noise floors, but for decades modern electronics are quiet enough to not need the extra swing.
I understand why we need headroom and there is a lot of technical info to read and debate.  However I'm more interested in the sonics of headroom or more importantly lack of headroom. 
The obvious sonic consequence of running out of headroom is saturation or clipping.  The audibility of this depends on how severe the clipping, and what frequencies are being clipped (LF clipping is easier to hear).
Lets take a Neve VR, wonderful console to work and we can slam the mix bus and it can sound great.  Do the same on an SSL J,  it will start clipping in a very nasty way.  Also a great console but needed more headroom.
or different gain structure
1073 pre I will often crank the gain on a snare for some crunch and back the output right back as not to clip my DAW,  "technically" on paper stupid.  sonically this can sound amazing.    Do this on an SSL mic pre, it will sound like a disaster.
I'll take your word for that.
So here is my point, if we extended the headroom on a Neve VR mix buss the sound we get as we max it out would become less likely without having to gain everything up to reach that point.  On the other hand an SSL J series it would be useful to have more headroom and they did increase the headroom on the K Series.
I don't know about those models but 'bus' is spelled with one 's'.
I'm of the opinion when making clones of classic audio gear don't mess with the headroom.  Some of the best creative sounds  can be when you are smashing everything way beyond what would be considered normal operating levels and the saturation and distortions are truly wonderful.  Why put these out of reach with increased headroom?
Different circuits have different overload characteristics. Tube saturation is widely regarded as euphonious, while solid state saturation is a guitar effect (fuzz tone).
So extra headroom can be useful, but adding it for the sake of having more headroom may well be detrimental the sonics of that gear that make it iconic.
Perhaps the circuit design and technology used in iconic legacy products defines the characteristic sound.

I have no opinion or advice about intentional saturation in audio paths, in mixers/consoles it is IMO evidence of improper gain structure. One modern digital console (Midas) incorporates a soft clipping circuit in their mic preamp that is embraced by some users.

Consoles should make it easy for end users to get good sound. Mix bus headroom is a matter of gain structure not PS rails. Any affordable modern electronic path will be limited to roughly 36V IC technology, with lots of final mixes feeding 5V or lower rail voltage A/D convertors.

JR
 
Lack of headroom as a sonic choice is useless if the topology doesn't sound good when out of juice.  Different question.    Headroom itself has little to do with the Neve versus the SSL.  The same is true of a Mackie versus an (older) Midas. 

Worth considering this theoretical (16 versus 24 doesn't buy much):

9V (or +/-4.5V) = +12.27 dBu
12V (or +/-6V) = +14.77 dBu
24V (or +/-12V) = +20.79 dBu
30V (or +/-15V) = +22.73 dBu
32V (or +/-16V) = +23.29 dBu
34V (or +/-17V) = +23.82 dBu
36V (or +/-18V) = +24.31 dBu
40V (or +/-20V) = +25.23 dBu
44V (or +/-22V) = +26.06 dBu
48V (or +/-24V) = +26.81 dBu
60V (or +/-30V) = +28.75 dBu
 
JohnRoberts said:
or different gain structure

Yes that is obvious, but when you have been mixing for 12 hours and you decide to do some big pushes into the chorus and you clip out the buss it can be a pain  to rework the levels across the console.  This was just to point out where more headroom would be desirable.  Loads of engineers complained about this as working in the normal SSL didn't always work,  Doesn't happen on the E and G and  SSL fixed it on the K.

JohnRoberts said:
I'll take your word for that.

Don't take my word for anything! Why not try something creative.  Try it and if you think it sounds terrible don't do it. 

JohnRoberts said:
I don't know about those models but 'bus' is spelled with one 's'.

So you decided to pick up on my spelling instead!  Bus or  Buss are both commonly used in the professional environment. 

https://www.waves.com/8-mix-buss-compression-eq-saturation-tips
https://www.audiomasterclass.com/newsletter/what-is-a-buss

Plenty more example if you want to google.

JohnRoberts said:
Different circuits have different overload characteristics. Tube saturation is widely regarded as euphonious, while solid state saturation is a guitar effect (fuzz tone).

If you say so!
 
Electrobumps said:
1073 pre I will often crank the gain on a snare for some crunch and back the output right back as not to clip my DAW,  "technically" on paper stupid.  sonically this can sound amazing.    Do this on an SSL mic pre, it will sound like a disaster.

IMO, it's more related to topology (A or AB-B).  I never saw anybody ask here for higher power supply voltage for BA283.
But, sometimes every additional dB of S/N or headroom counts. I designed for myself a microphone preamp for stereo pair setup working on +/-32V. YMMV.
 
EmRR said:
Lack of headroom as a sonic choice is useless if the topology doesn't sound good when out of juice.  Different question. 

moamps said:
IMO, it's more related to topology (A or AB-B).  I never saw anybody ask here for higher power supply voltage for BA283.
But, sometimes every additional dB of S/N or headroom counts. I designed for myself a microphone preamp for stereo pair setup working on +/-32V. YMMV.

These point makes sense. 

So we can say the typology of the circuit will dictate the character of the saturation, distortion, break up, sonic quality or however we technically or not so technically describe that character.

The original point...will less headroom bring these characters to appear at a lower program level, If we have the desired typology?

My main thought is in relation to vintage clones running on higher voltage rails,  and if this would this sonically change the point we hear saturation? 

From the numbers posted we would say only by a few dB.  So I now conclude this will have a marginal difference in the real world in most applications.


 
Lots of different questions and answers here.  Some waffle:

Early solid-state amp gain blocks (circa 1964 to 1972) generally exhibit a more monotonic type of distortion and saturation rather than the abrupt clipping that happens with  your typical modern IC op-amp when the signal approaches the rails. No amount of increased supply voltage will make that op-amp behave differently in that regard, other than pushing the clip point higher.  There still won't be a monotonic increase in distortion products relative to level.  This is down to topology: 

Use a transistor (bjt or J-Fet) that is loaded with a simple resistor rather than an active load (or a pair of active loads as with a current mirror), and you will get a softer saturation.  Use a single ended input (one transistor) rather than a nicely balanced diff-amp input pair and H2 doesn't get cancelled = mo' euphonic.

Internal operating level of the audio plays a big part in desk headroom.  The classic Neve desks had a fairly low internal level and relied on a wee bump (+4dB) in voltage step up from the output transformer along with the fact that primary of the output transformer was a choke load for the 3N3055 transistor which gave an almost doubling of the effective output voltage swing.  That's why a 1073 output stage can drive +26dBu with just a single-sided 24V supply.
Modules such as the 1081 achieved the same output swing by using a bigger step up in the output transformer (+10dB).  Same end result except for being either 'A' or AB' in class.

Ratio of the signal level to available voltage will tell you a lot about what to expect.  That and the topology.



 
Very interesting Winston.  Thanks.

I grew up on an 8036.  It’s interesting the different eq modules.  We had 1064 input modules and 1081.  1064s were said to be class A output and 1081s were  class AB.  They each had a unique sound 1073s I understand to be like 1064’s with more eq points. 
 
Electrobumps said:
Yes that is obvious, but when you have been mixing for 12 hours and you decide to do some big pushes into the chorus and you clip out the buss it can be a pain  to rework the levels across the console.  This was just to point out where more headroom would be desirable.  Loads of engineers complained about this as working in the normal SSL didn't always work,  Doesn't happen on the E and G and  SSL fixed it on the K.
OK I have written about exactly this (bus overload) here before.

Last century I designed an entire mixer series that addressed that exact problem, more for musicians trying to mix themselves from onstage who didn't have the luxury to reset the faders in the middle of their set because their L/R bus was saturating. In the  studio you have the luxury of time. (Some VCA fader system also support cranking in offsets across all the channels to buy similar headroom.)

I designed a low cost mixer series last century with an alternate gain structure where the typical +10dB of post fader gain in every channel before feeding the mix bus was actually only unity gain, while the fader was still calibrated indicating +10dB @ full up to keep the fader pushers happy. The missing 10dB of gain was actually added post the master sum bus so bingo the gain structure was transparent to the end user except for the extra 10dB of bus headroom. 
Don't take my word for anything! Why not try something creative.  Try it and if you think it sounds terrible don't do it. 
cute... how many decades did you design consoles?
So you decided to pick up on my spelling instead!  Bus or  Buss are both commonly used in the professional environment. 
yes bus is commonly spelled incorrectly (buss) by many, mostly on the  music equipment side. Other professional disciplines use computer "buses", or power distribution" buses" spelling it correctly (I researched this myself decades ago without using google. Buss is a verb, look it up  ::) ).

I even spelled it wrong myself before the 1980s, but eventually figured it out.
I do..



JR

PS: Of course there is no free lunch...When rejiggering the gain structure to put the +10dB gain post sum bus instead of pre, he bus noise floor is amplified by that post +10 dB. This was not a problem for my series of modest channel count mixers thanks to modern low noise op amps and low noise gain at the summing amp, while big dog consoles are too proud of their WFO bus noise floor spec to give away 10dB. In my last big console I put a 10dB pad across the feedback R of the master sum amp buying the 10dB of headroom with no noise floor penalty (but that console had over 100 inputs feeding the L/R bus.) 
 
fazer said:
I grew up on an 8036.  It’s interesting the different eq modules.  We had 1064 input modules and 1081.  1064s were said to be class A output and 1081s were  class AB.  They each had a unique sound 1073s I understand to be like 1064’s with more eq points.

I don't remember off hand what the differences in eq are between a 1064 and 1073 but the 1064 is basically the same topology as the 1073 and its ilk, it had the 183 and 184 (283 or 284) amp cards.  It was just housed in a 12" format rather than 8" so you get separate controls for eq point and amplitude rather than dual concentric.  They're nice channel modules for sure.

About 20 years ago I bought David Manley's 80 series desk from Eve-Anna Manley (they were sorting out the separation and divorce stuff at the time) and that desk had originally had 1064 modules but, somewhere along the way, they were changed out for 1081's.  I hated doing it but, the only way I could afford to buy the desk was by selling off 18 of the 24 channel amps.  I ended up with a few free (or silly money cheap) 1081's, a bunch of 1272's, plus the routing modules out of it so I suppose I shouldn't feel too bad.
Anyway, I wired up a couple of pairs of the 1081's and had switchable output stages using cards and transformers from the class A router outputs for the A or AB thing. 
I'd probably sell a testicle and a pinky finger just to have that frame with all the buss amps and routers right now  ;)


P.S.  Forgot, the desk also had 2254 compressors installed but I don't recall what happened with those.  Damn!



 
Electrobumps said:
1073 pre I will often crank the gain on a snare for some crunch and back the output right back as not to clip my DAW,  "technically" on paper stupid.  sonically this can sound amazing.

Is that 'output' control located after the output transformer or at the fader insert point? just curious...

Michael
 
Michael Tibes said:
Is that 'output' control located after the output transformer or at the fader insert point? just curious...

Michael

The regular 1073 does not have a control after the output transformer.

Cheers

Ian
 
A other have said, I think topology and class are the main drivers. Most modern gear uses op amps. These use huge amounts of negative feedback and class B or AB push pull output stages. In combination this means they are very linear right up to the point of clipping. After that they hard clip and sound horrid.

Older designs use class A single ended output stages and modest amounts of negative feedback. The effect of these differences are several. First, the smaller negative feedback means distortion tends to rise with level. This is much like what happens when instruments are played louder - they become richer in harmonics - so this level dependent rise in distortion tends to make things sound louder. Also, the harmonic structure of the distortion in class A circuits  very different to push pull class B circuits. Push pull circuits tend to cancel even order harmonics leaving the third as the major distortion component. This is a musical 5th so it can give instruments a 'power chord' feel. Class A circuits tend to have a dominant second harmonics( a musical octave) which tends to make instruments sound brighter (tubes are particularly good at this which may account for their popularity).  Most class A circuits can be pushed a long way into distortion before clipping occurs so they can sound good when driven 'harder'.

The Neve BA283 has a single ended class A output stage, as does the venerable V72 and the REDD 47. And of course they all use transformers at both the input and output which undoubtedly adds to the sonic palette.

Bottom line, raising the supply voltage on op amps just raises the maximum level before clipping but the character of the clipping is unchanged.

With class A stages, raising the rail voltage also means clipping level is raised but in addition there is a greater range over which potentially euphonic distortion is available so it may be a 'good thing'.

Cheers

Ian
 
ruffrecords said:
With class A stages, raising the rail voltage also means clipping level is raised but in addition there is a greater range over which potentially euphonic distortion is available so it may be a 'good thing'.

Cheers

Ian

Very informative thanks,  much better information than people getting picky about spelling and spouting off about how many consoles they designed.  ???


 
scott2000 said:
I picked up on a bit more than just that?

Maybe I had a wider Q dialed in on all the info :-\.....

Yeah maybe so, but my BS filter was broken.  So I found it difficult get the all the information.



 
Electrobumps said:
Very informative thanks,  much better information than people getting picky about spelling and spouting off about how many consoles they designed.  ???

Congrats, you’ve been cute again and disinclined the senior elders from ever helping you again.  Sometimes one knows too little to know what they are hearing, or who they are hearing it from.  All you ask has been covered over and over and over again here.  The research material is deep.  Head into the stacks instead of asking for a thesis at the front desk.  Yeah, it’s ‘bus’. 
 

Latest posts

Back
Top