EQ Phase Responses

[Mlewis]

statement about the phase response of the hi's and lows

that seems to be a sensible idea so long as the high and low filters are perfectly complimentary in terms of phase shift but wouldn't the mid band stick a spanner in the works with its phase response?

i came across an interesting note over at Rane about their accelerated slope tone controlls:

http://www.rane.com/pdf/acceler.pdf

This is a quote from the Weiss EQ1Mk2-LP brochure:

Contrary to common belief, non-minimum phase IIR systems are realizable, this usually involves a minimum-phase section followed
by a linear all-pass section with any arbitrary phase response. Thus, a linear-phase system would have a minimum-phase system followed by an all-pass system designed to have a phase response which, when added to the minimum-phase, will result in a linear phase response. This can (theoretically) be done in the analogue as well as the digital domain. However, there are certain practical problems which make this method difficult, if not impossible.

what are the difficulties mentioned? is it the need for analogue time delay or actually creating all-pass filters with the complimentary phase responses of the minimum-phase filters?

 

[gyraf]

Problem would probably be getting the allpass tracking the frequency, Q, and boost/cut amount..

Jakob E.

 

[Samuel Groner]

Well, the text you quoted is not perfectly precise. You can get as close as you want to linear phase response by adding up a finite number of all-pass filters, but only an infinity number of all-pass filters will result in true linear phase response. And even if you'd buy all NE5532s to get there this EQ would be pretty unusable because it would have infinite delay--you'll never get your audio back that you sent in.

In the digital world, we are able to (symmetrically) truncate the impuls response of a hypothetical analogue filter to get true linear phase without infinite delay. That is the so-called FIR (finite impuls response) filter.

That seems to be a sensible idea so long as the high and low filters are perfectly complimentary in terms of phase shift.

Unfortunately they aren't (assuming minimum-phase filters).

Samuel

 

[Larrchild]

You can get as close as you want to linear phase response by adding up a finite number of all-pass filters, but only an infinity number of all-pass filters will result in true linear phase response. And even if you'd buy all NE5532s to get there this EQ would be pretty unusable because it would have infinite delay--you'll never get your audio back that you sent in.

Bob Orban's analog broadcast processors, with their phase rotator circuitry, came damn close to the abovementioned description! :grin:

That's why most sane broadcast engineers turn them off.

 

[SSLtech]

-Sounds like Bob Orban's work turns YOU off, Larry...

:green:

Keef

 

[clintrubber]

W.r.t. Oram-article, I read it a few days ago and while I don't recall it exactly, there were a few 'special moments'. But I'm not sure who's to blame, as in the SOS-author overplaying his hand w.r.t. tech-details.

Why is it that op-amps have such a bad name in purist audio circles?

If you were to take a standard 741 op-amp (which the data sheet claims has 70dB of gain) and try building an amp with 70dB of gain, you'd find it had 4% distortion, a noise floor of -50dB, and an HF roll-off 3dB down at 9kHz -- appalling. In the Trident Fleximix design, I took two 741s (which was the op-amp of the day), ran them at 35dB of gain each, and got a response flat to 20kHz with 0.05% THD and a low noise floor.

I get his point, but to illustrate it by the fact that somebody would want to use an opamp for providing gain equal to the open loop gain is not elegant.

Valves also had that inherent odd harmonic distortion, while transistors produced even harmonic distortion, which wasn't very pleasant from a guitarist's point of view.

This sounds like a case of reconstructing an interview from notes on which a full cup of coffee was spilled, so let's not spend too much time here :wink:

But still a nice read; I was actually pleasantly surprised to find such an amount of electronics-tech-talk in an SOS article.

Regards,

Peter

 

[Larrchild]

-Sounds like Bob Orban's work turns YOU off, Larry...

I really respect the novel (and patented) things he does to make an FM transmitter 2 db louder, lol.
But he's really down on "Golden Ears" types. Tubes , especially.

Regarding the "other "O" word, I think if we look back at what console eq we like throughout the years, and which we don't..analysis of phase characteristics must be mandatory. So the article was revealing in that regard.

 

[Mlewis]

[quote author="Samuel Groner"]
In the digital world, we are able to (symmetrically) truncate the impuls response of a hypothetical analogue filter to get true linear phase without infinite delay. That is the so-called FIR (finite impuls response) filter.
Samuel[/quote]

I'm fairly sure that Weiss don't use FIR filters for their EQs. They use complimentary IIR filters to create a linear phase response. I’ll do a bit more research before sticking my head above the parapet on this tho!

http://www.weiss.ch/eq1/manuals/eq1lp.pdf
http://www.weiss.ch/eq1/brochures/eq1lp.pdf

 

[mikep]

[quote author="Mlewis"]
what are the difficulties mentioned? is it the need for analogue time delay or actually creating all-pass filters with the complimentary phase responses of the minimum-phase filters?[/quote]

IIRC the Am@k 9 0 9 8 eq, designed by Rudolph Neeb, has something similar to this is the hi and low shelf circuits. If I understand it correctly, the "glow" and "sheen" buttons actually turn the phase correction bits OFF. I don't know how close it comes to perfect linear phase, tracking, etc. anyone know some more about this EQ?

mike p

 

[Mlewis]

[quote author="mikep"]
IIRC the Am@k 9 0 9 8 eq, designed by Rudolph Neeb, has something similar to this is the hi and low shelf circuits
[/quote]

Sounds intersting. i'd always assumed that the "glow" and "sheen" buttons were some kinda snake-oil, subtle high and low boost circuits but having never heard the EQ in action was just guessing really.

i've got the Schem for the HMF and LMF filters but i can't find the shelves. Anybody got a link?

 

[gyraf]

Why all this name masking? It only results in the same questions being asked over and over again, and searches to be futile..?

Jakob E.

 

[mikep]

[quote author="gyraf"]Why all this name masking? It only results in the same questions being asked over and over again, and searches to be futile..?
[/quote]

I don't know! I thought that was the propper etiquette around here when discussing a commercial design. to not attract the wrong kind of attention or something? Ive never seen it discussed directly.

I do have the schematic on CD-rom somewhere, lets see if I can find it...

mike p

 

[gyraf]

I think the name masking is kinda misunderstood - it messes up the flow of information by making the posts unsearchable.

I wish people would stop doing this, unless off course they have a under-the-radar schematic or something that they decidedly want un-searchable..

Jakob E.

 

[Mlewis]

[quote author="mikep"]
I do have the schematic on CD-rom somewhere, lets see if I can find it...
mike p[/quote]

masking or no masking, thanks for looking for the schematic anyway.

 

[mikep]

[quote author="Mlewis"]
masking or no masking, thanks for looking for the schematic anyway.[/quote]

errrhm. unfortunately this is one of things that hasn't surfaced since the last move... still haven't found it.

 

[SSLtech]

[quote author="mikep"]IIRC the Am@k 9 0 9 8 eq, designed by Rudolph Neeb, has something similar to this is the hi and low shelf circuits. If I understand it correctly, the "glow" and "sheen" buttons actually turn the phase correction bits OFF. I don't know how close it comes to perfect linear phase, tracking, etc. anyone know some more about this EQ?[/quote]
Boy, labeling the buttons "sheen" and "glow" really did a number on everybody's understanding, didn't it!

Recently, I got about a half dozen half-decent engineers who regularly use the 9098i console together, and asked them what they thought the "sheen" and "glow" buttons did. -I got as many different suggestions as there were people who used it.

Then I ran a sweep with the Neutrik analyser on the EQ, several settings, with the Sheen off and then on, likewise for the glow. I also ran THD+N measurements.

What they do is very simple. They open up the bandwidth on the top and bottom sections respectively, so that no matter what frequency you have them selected to, the shaped section extends wll above or below the audio band.

There's no magic, there's no mystery, it's just a VERY wide bandwidth setting.

-Now, the fact that I got answers -before I ran the sweeps- ranging from "it adds second harmonic distortion" all the way to "it makes the EQ very precise" and "it's like the opposite of a notch filter" (whatever the hell THAT's supposed to mean!) doesn't really say much about the users whom I asked.

The upshot is that nobody actually NOTICED that the bandwidth got so friggin wide when you press those buttons. -THe only reason that I ran the test was because I thought that was what I was hearing, but there were so many other "dude.. like no way... it totally adds second harmonic" replies, that I wondered if I was missing something.

Anyhow, it says a lot for marketing and the power of suggestion, that pleasant words like "glow" and "sheen" are all it takes to convince some people that magic is being worked... to the point that they don't even notice that their bandwidth control isn't doing anything any more, and the HF ± control is almost a volume control wqhen you bring the frequency all the way down...

One more indication that people all too often hear what they want to hear based on what they read on paper.

No magic. -A good EQ, a fabulous notch filter. The low-Q settings are very useful and pleasant sounding, but for the love of god, when will most people GET IT?

(sigh)

Keef

 

[Mlewis]

Ha Ha! thats hilarious! reminds me of the tales of DFA boxes (ie. Does F*ck All boxes).

i can hear it now "no...no... it just doesn't have enough glow and honestly, you haven't got a clue, there's nowhere near enough sheen on that"

i still quite fancy havin a look at the schematic if there is one though.

 

[clintrubber]

[quote author="SSLtech"]What they do is very simple. They open up the bandwidth on the top and bottom sections respectively, so that no matter what frequency you have them selected to, the shaped section extends wll above or below the audio band.[/quote]
As in simple bell/shelf-switches ?
I've one of those old 9.5" Boss-Pro PQ-50 boxes around with such a switch on the Q-knob of the highest & lowest band. I'll re-label ! :wink:

 

[SSLtech]

[quote author="clintrubber"]As in simple bell/shelf-switches ?
I've one of those old 9.5" Boss-Pro PQ-50 boxes around with such a switch on the Q-knob of the highest & lowest band. I'll re-label ! :wink:[/quote]
No not bell/shelf... they stay in Bell. -Remember, a shelf can still have a tight Q with all the attendant impulse response complications... however, a super-wide bell extends many octaves above and below the center frequency, whether it's HF or LF.

Keith

 

[clintrubber]

[quote author="SSLtech"][quote author="clintrubber"]As in simple bell/shelf-switches ?
I've one of those old 9.5" Boss-Pro PQ-50 boxes around with such a switch on the Q-knob of the highest & lowest band. I'll re-label ! :wink:[/quote]
No not bell/shelf... they stay in Bell. -Remember, a shelf can still have a tight Q with all the attendant impulse response complications... however, a super-wide bell extends many octaves above and below the center frequency, whether it's HF or LF.

Keith[/quote]
Thanks for explaining :thumb: