Did everyone lie about all pass filters?

[user 41968]

Back in the early 70s, a filter that maintained 90° between its two output was called a "dome filter." I made making two strings of single-pole all-pass filters. For the application, ±5° was close enough so the turnover frequencies were interleaved in roughly 3 to 1 ratios. Use more stages and smaller ratios to hold the 90° arbitrarily tight. These filters were used in radio communications for decades to make quadrature detectors that worked over octaves of frequency range. I used mine to maintain quadrature in a variable-frequency motor drive (from half-speed to double speed) for a 3M "Iso-Loop" tape machine. Using the dome filter, plus a high-pass filter to increase drive voltage linearly with frequency as well, two audio power amps driving the motor windings directly gave it full torque over the 4:1 speed range.

Such a filter is at the heart of the attached circuit I found in a quick web search. For what you're doing, drive the lines marked "L" and "R" together. Output of the filter is on lines marked "I" and "Q" (traditional for In-phase and Quadrature).

Thanks Bill for posting my rendition of the Studer 90° Filter with Weaver alignment. I hope it proves useful to someone. I may do a board for it soon so others can experiment.

At a minimum it can be used to draw pretty flowers on a goniometer.

 

[user 37518]

Thanks Bill for posting my rendition of the Studer 90° Filter with Weaver alignment. I hope it proves useful to someone. I may do a board for it soon so others can experiment.

At a minimum it can be used to draw pretty flowers on a goniometer.

Wayne? I am actually working on an improved version of this 90° constant phase shift filter. My first encounter with "Dome" filters came during my PhD, I wanted to add a constant 90° phase shift to one microphone signal and then mix it with the in phase signal, my goal was to be able perform an arbitrary phase shift depending on the ratio of the "quadrature" vs "in phase" signals, essentially the same of what IQ modulators do, hence, I worked a bit on this filter.

 

[user 41968]

In the 90's, I designed a spectrum-steering pan-pot, which consisted in boosting HF content on one side whilst substracting the same amount from the opposite side. 100% mono compatible.
I submitted it to Soundcraft, who took the time to evaluate it, but, although the cost was modest, concluded that it would not constitute a competitive advantage. I found out it requires some education to understand the benefits; when people see that the signal does not fully disappear on one side, they think it's broken.

I tried that in 2007 and the result was interesting. A Matrix-Based Pan Pot With Variable Width - Pro Audio Design Forum

I used a HP filter in the insert and provided a panning demo from a mono source. The link to the sound file is in the thread. The matrix panner itself was constant-voltage and the outputs inherently mono-compatible.

WRT your comments Abbey about the Studer 90° Filter solving a non-existent problem I tend to disagree having built one and actually listened to it.

There are numerous sound files in the thread below with stark differences between L+R and "I+Q" mono fold down. I+Q with problematic material almost always sounds better. In addition, on the files I looked at, I+Q had higher RMS levels than (L+R)/2.

https://www.proaudiodesignforum.com/forum/php/viewtopic.php?t=1055
The Weaver alignment phase response: The Studer "90° Filter" Stereo to Mono Summer/Recorrelator - Page 2 - Pro Audio Design Forum

If the material I used for demo quadrature material was lost due to cancellation it was de minimis when compared to the original stereo presentation. (L+R)/2 almost always sounded the worst of the three presentations. With material not needing "correction" the 90° filter almost never sounded "worse" than L+R. If anything they tended to sound more similar.

 

[user 41968]

Wayne? I am actually working on an improved version of this 90° constant phase shift filter. My first encounter with "Dome" filters came during my PhD, I wanted to add a constant 90° phase shift to one microphone signal and then mix it with the in phase signal, my goal was to be able perform an arbitrary phase shift depending on the ratio of the "quadrature" vs "in phase" signals, essentially the same of what IQ modulators do, hence, I worked a bit on this filter.

That's me.

Take a look at the Weaver alignment. Studer's math was a little off.

 

[user 41968]

It occurred to me that as we wrap our heads around this thing there may be a misunderstanding that has developed about the utility of summing out-of-polarity elements. After all, other than an over-the-top synth patch, how often does this occur? I'd say rarely. "Sympathy for the Devil" during the piano intro is the example in a commercial release I found. (Curiously the out-of-polarity element becomes in-polarity a few bars later.)

The 90° filter's primary benefit is not for completely out-of-polarity elements but those that are conventionally hard-panned with a knob or switch.

L = 0.5M + 0.5S
R= 0.5M - 0.5S

When something is hard-panned to L, R = 0.
When S is removed by summation what was panned to L appears in Mono at -6 dB.
All those Beatles songs with hard-panned elements, when collapsed to mono, get pulled back in the mix.

The 90° Filter gives M and S equal -3dB weight.
When summed in quadrature elements that would have been pushed back in the mix are brought forward relative to their L+R mono level.
The added S also increases signal power and the sum has similar balance to the stereo presentation.

 

[pvision]

Another example of how this forum generates the most interesting & diverse discussions, and where tangential references add to the overall picture without detracting

 

[abbey road d enfer]

WRT your comments Abbey about the Studer 90° Filter solving a non-existent problem I tend to disagree having built one and actually listened to it.

All those Beatles songs with hard-panned elements, when collapsed to mono, get pulled back in the mix.

Unearthing this thread.
My comment was in reference do decent mixes, not these aberrations.
I agree that the 90° shifter is useful for making a sloppy record listenable, but as a monitoring tool, it is counterproductive because it makes a stupid mix acceptable.

 

[user 41968]

but as a monitoring tool, it is counterproductive because it makes a stupid mix acceptable. (emphasis added.)

I don't recall suggesting that.

 

[abbey road d enfer]

I don't recall suggesting that.

Indeed, but that was the argument for the german broadcast engineers. That's why it was built in mixers.

 

[user 41968]

The commercial examples I've seen, from Studer and Lawo, were modules that appeared to be for AM/MW mono air chains.

They actually built these into consoles?
Wouldn't the justification to include them be to generate a mono broadcast feed from a source that is too wide and mixed upstream where they have no control?

 

[abbey road d enfer]

They actually built these into consoles?

Yes, they did. Many broadcast mixers were equipped with it, inserted in the mono monitoring chain, which was a specific speaker/amp, separate from the main (stereo) monitors.

Wouldn't the justification to include them be to generate a mono broadcast feed from a source that is too wide and mixed upstream where they have no control?

That would be correct, but should be insertable only when necassary.

 

[abbey road d enfer]

It occurred to me that as we wrap our heads around this thing there may be a misunderstanding that has developed about the utility of summing out-of-polarity elements. After all, other than an over-the-top synth patch, how often does this occur? I'd say rarely. "Sympathy for the Devil" during the piano intro is the example in a commercial release I found. (Curiously the out-of-polarity element becomes in-polarity a few bars later.)

One very common happenstance is many effects (phaser/flanger, delay...) that are announced as "stereo" actually deliver the wet signal with opposite polarity on the L&R outputs. Of course, when summed to mono, the effect completely disappears.
So it looks like the 90° shifter could be a worhwhile addition to these effects boxes.

Now, since you have built and listened to this filter, what happens auditively with the shifted signals?
We all know that a signal distributed out-of-phase between L&R channels results in the weird "where is it coming from" effect, but there is no loss of information.
Does the 90° filter provides some kind of "stereo" sensation or is it weird?
Could it be a better stereo synthesizer than those based on comb filters?

 

[user 41968]

One very common happenstance is many effects (phaser/flanger, delay...) that are announced as "stereo" actually deliver the wet signal with opposite polarity on the L&R outputs. Of course, when summed to mono, the effect completely disappears.
So it looks like the 90° shifter could be a worhwhile addition to these effects boxes.

Now, since you have built and listened to this filter, what happens auditively with the shifted signals?
We all know that a signal distributed out-of-phase between L&R channels results in the weird "where is it coming from" effect, but there is no loss of information.
Does the 90° filter provides some kind of "stereo" sensation or is it weird?
Could it be a better stereo synthesizer than those based on comb filters?I put a relay in the 90 degree filter to forward the I and Q outputs to L and R and it provided a shifted pseudo-stereo image that was somewhat "fixed" and very weird.

I put a relay to forward I and Q to L and R and the imaging of mono or stereo sources was weird.

A long time ago I did a stereo synthesizer using a delay line which provided comb-filtered sum and difference with a frequency-dependent dynamic delay similar to the DynaFlanger I helped develop with Bill Hall at MicMix.

The third approach I tried was a spectral panner which I think we discussed in this or another thread.

Of the three the spectral panner was the most natural at synthesizing stereo, the comb-filtered less natural but "ambient" and "swimmy" without a hard center and the I and Q image hard-pulled and weird.

I think the best example of using the 90 degree filter would be the stereo flanger example folded down to mono as I+Q and then panned to or off-of center. With that option you get the flanger effect without it cancelling downstream in mono.

 

[abbey road d enfer]

OK, thanks for your answer.