dirtyhanfri
Well-known member
JohnRoberts said:
I agree with you completely, but;"...linearity ut to 80KHz at least,..." seems a bit excesive to me, as a minimum spec, but, I´m not a golden ear guy anyway
MIX BUS SATURATION
- Thread starter simonsez
- Start date
Help Support GroupDIY Audio Forum:
I agree with you completely, but;"...linearity ut to 80KHz at least,..." seems a bit excesive to me, as a minimum spec, but, I´m not a golden ear guy anyway
I am not a golden ear guy either... exactly the opposite. I am all about in-band audio and objective (read measurable) phenomenon,
Allow me to clarify I am talking about the primary raw audio path (like a console front end/mic preamp). Even though we (at least I) can't hear above 20 khz, that doesn't mean we won't encounter signal up there... If the electronic path can not pass above 20 kHz cleanly, or filter above 20kHz content low enough to pass cleanly, we can get audible IM distortions. IMD caused by very HF sources can express at low audio frequencies and be very audible.
I gave the obvious example of a closely mic'd cymbals. strong diffuse content up well above 20kHz in a path that can not keep up, will generate IM distortion which will sound like LF mud/grunge. Another surprisingly HF source is to take the keys out of your pocket and jangle them in front of a good microphone. Once again if you have a slow path that isn't well LP filtered, you will hear the path mess up.
Note: Modern oversampled A/D convertors sufficiently LPF the input so they will generally just benignly LPF such sources.
BTW, the circuitry has to have adequate gain bandwidth to effectively LPF the signal.
I am not advocating flat frequency response to 80kHz, just that the circuitry be capable and preferably LP filtered. These days since most work product ends up in the digital domain, you will get LP filtered eventually, but if you allow a bunch of IM trash into the front end, the later stages will think that is your idea of music.
JR
PS: I concede this is a discussion about a sum bus. Prudent console design should band pass the signal in the front end, before it gets to the sum bus. An 80kHz one pole -3dB filter, would be -1 dB at 40kHz and -.1dB at 8kHz. Since all of these roll-offs accumulate, it seems prudent to target a nominal bandpass well above 20 kHz.
Allow me to clarify I am talking about the primary raw audio path (like a console front end/mic preamp). Even though we (at least I) can't hear above 20 khz, that doesn't mean we won't encounter signal up there... If the electronic path can not pass above 20 kHz cleanly, or filter above 20kHz content low enough to pass cleanly, we can get audible IM distortions. IMD caused by very HF sources can express at low audio frequencies and be very audible.
I gave the obvious example of a closely mic'd cymbals. strong diffuse content up well above 20kHz in a path that can not keep up, will generate IM distortion which will sound like LF mud/grunge. Another surprisingly HF source is to take the keys out of your pocket and jangle them in front of a good microphone. Once again if you have a slow path that isn't well LP filtered, you will hear the path mess up.
Note: Modern oversampled A/D convertors sufficiently LPF the input so they will generally just benignly LPF such sources.
BTW, the circuitry has to have adequate gain bandwidth to effectively LPF the signal.
I am not advocating flat frequency response to 80kHz, just that the circuitry be capable and preferably LP filtered. These days since most work product ends up in the digital domain, you will get LP filtered eventually, but if you allow a bunch of IM trash into the front end, the later stages will think that is your idea of music.
JR
PS: I concede this is a discussion about a sum bus. Prudent console design should band pass the signal in the front end, before it gets to the sum bus. An 80kHz one pole -3dB filter, would be -1 dB at 40kHz and -.1dB at 8kHz. Since all of these roll-offs accumulate, it seems prudent to target a nominal bandpass well above 20 kHz.
dirtyhanfri
Well-known member
I, again agree with you (thanks for the explanation, it helped me clarifying some things I've heard about actually)
But the statement made below goes much further than yours, and it's much less argumented.
We agree in the fact we need linearity beyond 20khz, but would you skip using a mic pre o comp that you like and sounds fine just because it's not linear up to 80Khz? That sounds crazy for me.... What is the bandwith of a la-2 or a 1176? I don't actually know
Should I buy a 1Mghz oscilloscope to check the behringer gear I can afford with the (little) earnings on my studio? ??? You know, If I start that struggle, I won't have time to work anymore...
Sorry for the irony, but sometimes I get sick with audiophoolia.
About the Bus saturation, I don't find any good on doing that on my mackie, it actually sucks... I think the mixers who have grown in digital domain (me included) use to have a wrong idea about "analog saturation". I try to mix and track every day with more and more headroom, it just makes the work easier later...
But the statement made below goes much further than yours, and it's much less argumented.
We agree in the fact we need linearity beyond 20khz, but would you skip using a mic pre o comp that you like and sounds fine just because it's not linear up to 80Khz? That sounds crazy for me.... What is the bandwith of a la-2 or a 1176? I don't actually know
Should I buy a 1Mghz oscilloscope to check the behringer gear I can afford with the (little) earnings on my studio? ??? You know, If I start that struggle, I won't have time to work anymore...
Sorry for the irony, but sometimes I get sick with audiophoolia.
About the Bus saturation, I don't find any good on doing that on my mackie, it actually sucks... I think the mixers who have grown in digital domain (me included) use to have a wrong idea about "analog saturation". I try to mix and track every day with more and more headroom, it just makes the work easier later...
Kingston
Well-known member
dirtyhanfri said:
They are as flat and wide as your transformers. LA2A needs some calibration of its global feedback, but both of my units are flat up to 40khz for example - using modern transformers. -3dB point is probably well above 80khz, but I never bothered to measure that high.
dirtyhanfri
Well-known member
Kingston said:
Interesting, thanks for the info
Don't confuse linearity with frequency response.
To clarify my personal design philosophy about this, this is "not" about providing significant frequency response well above 20kHz, but either cleanly passing above band HF signals or cleanly attenuating them (preferably attenuating them), so they can't generate audible artifacts at lower frequencies that we will hear. If the front end is properly designed (with LPF), even poorly designed gear in later stages will be protected from slew rate misbehavior. The best design strategy is often a combination of the two (pass and filter).
This is mainly an issue with active solid state circuitry to prevent rectification when circuits can't keep up with signal voltage rate of change. This is not generally a problem for transformer front ends since that transformer forms a bandpass filter protecting following circuitry from excessive slew rates.
-------
I need to clarify another earlier comment of mine.. You only need adequate gain bandwidth for active LPF. Passive LPF will generally not have any GBW concerns.
-------
My speed and GBW concerns are mainly so the path can benignly ignore above band crap not reproduce things we can not hear.
JR
PS I've been wrestling with audio phools for decades. Our public education system is not reducing the problem.
To clarify my personal design philosophy about this, this is "not" about providing significant frequency response well above 20kHz, but either cleanly passing above band HF signals or cleanly attenuating them (preferably attenuating them), so they can't generate audible artifacts at lower frequencies that we will hear. If the front end is properly designed (with LPF), even poorly designed gear in later stages will be protected from slew rate misbehavior. The best design strategy is often a combination of the two (pass and filter).
This is mainly an issue with active solid state circuitry to prevent rectification when circuits can't keep up with signal voltage rate of change. This is not generally a problem for transformer front ends since that transformer forms a bandpass filter protecting following circuitry from excessive slew rates.
-------
I need to clarify another earlier comment of mine.. You only need adequate gain bandwidth for active LPF. Passive LPF will generally not have any GBW concerns.
-------
My speed and GBW concerns are mainly so the path can benignly ignore above band crap not reproduce things we can not hear.
JR
PS I've been wrestling with audio phools for decades. Our public education system is not reducing the problem.
sr1200
Well-known member
If they taught audio classes in my school, i probably would have paid a helluva lot more attention in class! lol
The closest we got was in biology when we did the parts of the ear, and physics with the speed of sound.
The closest we got was in biology when we did the parts of the ear, and physics with the speed of sound.
+ zillion to that. I design mikes and Blind Listening Tests and will stick my neck out & claim the only possible audible effect of supersonic response is intermod giving mid grunge etc.
Here I think JR is rather optimistic. There are several tests on Sample Rate Converters and also recorders which show many with abysmal anti-aliasing filters. More grunge. Now you know why stuff with fs = 384kHz sounds differentJohnRoberts said:
But for the recorders, why worry about this when there are '24b 192kHz' recorders on sale with less practical dynamic range than a good Dolby B cassette recorder. 8)
Yup, perhaps I am a little optimistic. Yet another reason for my modest LPF in the front end, but I was mainly thinking about analog path rectification and IMD. Aliasing is another unwanted artifact that I hope HF oversampling by convertors mostly keeps in check.
For a little fun and games for all the lurkers reading this thread with digital recording rigs, have you ever run some simple above bandpass frequency sine waves to listen for unwanted artifacts? Maybe we need to whip up a modest distortion (say < .25% THD) sine wave generator, capable of 20-100+ kHz at a few volts p-p .Start with 20khz at maybe 1Vp-p and ramp up the frequency until you hear aliasing or new unwanted sounds. If it's clean to 100khz perhaps push the amplitude to 3V or more...
If you can pass this test you path should be free of HF overload from real world sounds.
Note: I do not suggest that digital paths should be able to casually ignore full amplitude 100kHz, IMO the analog front end should LPF it before it hits the digital conversion.
JR
For a little fun and games for all the lurkers reading this thread with digital recording rigs, have you ever run some simple above bandpass frequency sine waves to listen for unwanted artifacts? Maybe we need to whip up a modest distortion (say < .25% THD) sine wave generator, capable of 20-100+ kHz at a few volts p-p .Start with 20khz at maybe 1Vp-p and ramp up the frequency until you hear aliasing or new unwanted sounds. If it's clean to 100khz perhaps push the amplitude to 3V or more...
If you can pass this test you path should be free of HF overload from real world sounds.
Note: I do not suggest that digital paths should be able to casually ignore full amplitude 100kHz, IMO the analog front end should LPF it before it hits the digital conversion.
JR
maxiemixer
New member
- Joined
- Jun 11, 2012
- Messages
- 2
JohnRoberts said:
Thank's to JR. My statement is a direct experience of some very good mic preamps that I have used over the years. The one I like are the ones that fit in my statement. The difference sometimes cannot be obvious in the short run but in my experience there is a huge difference when you sum 24+ channels recorded with lesser mic preamps. I suggest to investigate more on that matter instead of looking for that elusive MIX BUS SATURATION magic from the past.
To be more explicit about it here is a list of mic pres that I like and use (check the specs of that stuff if you like)
- Neve 1073 (my first love!, unforgettable...)
- Neve 8232 recording desk (recently used on a recording session, sounds close to the 1073)
- GML 8304
- Amek/Neve 9098 DMA (This is currently my first choice)
- Focusrite Isa110 (I like them but I feel it a bit "cold"
- Symetrix 528 (my first real channel strip - bought in 1985, still in use. A bit slow, useless on snare...)
- Presonus D-8 (cheap class A preamps, for a different sound)
- Api 512c - very similar to the Symetrix, but way out faster - I love the snare trough that one!)
- SSL E mic pre - noisy but nice. I love the grit
- SSL G mic pre - useless the presonus D8 sounds similar (sorry to say) - I only use it straight from the insert send to a compressor or patched to the "tape" machine (wow, that gives away my age)
Since I use my master bus plugin chain I have never had anyone saying that the mix sounded too digital.
I'm trying to share my experience and give you a way to save money, but still in sound engineering if it sounds good IS good.
SSLtech
Well-known member
Comedy gold! The 81 and 82 seres use an inexpensive transformer (forget the brand, but it's one often used in DIY) into a single 5534 with a pot in the feedback loop to set the gain. -It has as much in common with a 1073 as a brick wall has with a mosquito. -What makes it "Sound close" is most likely the badge.maxiemixer said:
barclaycon
Well-known member
A couple of things.
First of all the 'cheapness' of the transformer doesn't neccesarily mean it has a bad effect on the signal. I had a conversation with Paul Wolf once about where API got their transformers from and he told me they were not exactly 'high-end' products and actually produced a lump at LF - but that it was quite appealing and added to the character of the design.
Secondly, there's something about passive mixing that does have an effect on the sound (I'm sure we've discussed this before). Testing signals under dynamic conditions (rather than steady state tones) and mixing signals together is difficult to prove.
One further thing, Rupert Neve always used to mention in his lectures about an incident where Geoff Emerick said there was a fault with a Neve channel and they couldn't find the cause.
Eventually they tracked it down to the lack of a compensation capacitor in the secondary of a transformer which caused a massive reasonant peak somewhere above 30kHz.
Rupert's point was that Geoff could hear the effect of something in the ultrasonic region.
But, I've got to say, if you ever heard the effect of an incorrectly terminated transformer, there's massive phase disturbance to the signal in the Khz region and I think it's that unpleasant smearing effect that most people would pick up on.
Those things on the sides of our heads can be quite useful!
First of all the 'cheapness' of the transformer doesn't neccesarily mean it has a bad effect on the signal. I had a conversation with Paul Wolf once about where API got their transformers from and he told me they were not exactly 'high-end' products and actually produced a lump at LF - but that it was quite appealing and added to the character of the design.
Secondly, there's something about passive mixing that does have an effect on the sound (I'm sure we've discussed this before). Testing signals under dynamic conditions (rather than steady state tones) and mixing signals together is difficult to prove.
One further thing, Rupert Neve always used to mention in his lectures about an incident where Geoff Emerick said there was a fault with a Neve channel and they couldn't find the cause.
Eventually they tracked it down to the lack of a compensation capacitor in the secondary of a transformer which caused a massive reasonant peak somewhere above 30kHz.
Rupert's point was that Geoff could hear the effect of something in the ultrasonic region.
But, I've got to say, if you ever heard the effect of an incorrectly terminated transformer, there's massive phase disturbance to the signal in the Khz region and I think it's that unpleasant smearing effect that most people would pick up on.
Those things on the sides of our heads can be quite useful!
Back in the day engineers tied to design a linear, clean audio path.barclaycon said:
More recently, old technology that was marginally linear, has be embraced as a source of distortion.
Good is in the ears of the beholder, but i suspect there are some old design engineers rolling over in their graves, as people corrupt otherwise clean paths by introducing distortion. .
yawn...
Top octave and near above poles, can and do have audible impact in-band... old news to responsible designers. Perhaps another reason to avoid transformers.
JR
barclaycon
Well-known member
Back in the day John, good audio engineers fine tuned audio products using their ears.
Sorry if you find that tiresome (yawn), but it's the reason that people still go after 'old stuff'.
Far from being horrified, I suspect that people like Rupert Neve would be delighted that their work has been valued and respected, and would be flattered that modern designers try to emulate their designs.
I'm not a commercial electronics designer by trade - I do have a degree in electronics, but I've managed to make a career in music recording and mixing over the last 36 years.
If I relied on the test results from something like an 8078 desk, I would probably never use one. But having heard what they 'sound' like, I would always look forward to it.
I don't care if something isn't 'entirely' linear.
It may be distorted - but if it is in a pleasing and musical way, then it becomes an enhancement.
That's a far more difficult thing to achieve than just a 'clean' signal path, however much you want to sneer about it.
Sorry if you find that tiresome (yawn), but it's the reason that people still go after 'old stuff'.
Far from being horrified, I suspect that people like Rupert Neve would be delighted that their work has been valued and respected, and would be flattered that modern designers try to emulate their designs.
I'm not a commercial electronics designer by trade - I do have a degree in electronics, but I've managed to make a career in music recording and mixing over the last 36 years.
If I relied on the test results from something like an 8078 desk, I would probably never use one. But having heard what they 'sound' like, I would always look forward to it.
I don't care if something isn't 'entirely' linear.
It may be distorted - but if it is in a pleasing and musical way, then it becomes an enhancement.
That's a far more difficult thing to achieve than just a 'clean' signal path, however much you want to sneer about it.
My philosophy is well stated... make the audio chain linear, and put the sound effects in their own box with an on/off switch.
Good cooks do not always use the same seasoning without tasting the dish, or begin with a dirty pot that always tastes the same... If you have multiple different cooks, designing different parts of the audio chain, all with their own different idea of optimal seasoning, the results can be variable, to put it kindly.
I have actually designed studio efx and I used my ears for that. When designing linear paths I find that bench test equipment can measure things that i can not hear.. So i improve those too. I then use my ears to check that the bench test didn't miss something.
Sorry I do not mean to sound dismissive, but I have approached these questions when forming my design philosophy decades ago. Nothing has changed my mind since then. IMO things that can be designed objectively should be, things that can't (like effects) are free game to do whatever works..
JR
PS: I have a (perhaps) funny joke for why the old designers turned over in their grave, but i don't want to be even more irritating that I already am.
Good cooks do not always use the same seasoning without tasting the dish, or begin with a dirty pot that always tastes the same... If you have multiple different cooks, designing different parts of the audio chain, all with their own different idea of optimal seasoning, the results can be variable, to put it kindly.
I have actually designed studio efx and I used my ears for that. When designing linear paths I find that bench test equipment can measure things that i can not hear.. So i improve those too. I then use my ears to check that the bench test didn't miss something.
Sorry I do not mean to sound dismissive, but I have approached these questions when forming my design philosophy decades ago. Nothing has changed my mind since then. IMO things that can be designed objectively should be, things that can't (like effects) are free game to do whatever works..
JR
PS: I have a (perhaps) funny joke for why the old designers turned over in their grave, but i don't want to be even more irritating that I already am.
barclaycon
Well-known member
I don't always agree with you John, but I value your wit and wisdom!
Regards
Regards
Andy Peters
Well-known member
barclaycon said:
I wish I could find it online, but I remember reading an interview with Rupert Neve where he mentioned that he wished he had modern devices back in the day when the classic pieces were designed.
Just sayin'.
-a
barclaycon
Well-known member
Oh right. That changes everything.
(!)
(!)
