The Studer "90°" Dome Filter Stereo to Mono Quadrature Summer

[AnalogJoe]

I have been working on these filters for some time now, but I had to stop for a while since I've been mad busy; I might get something published soon so I'll let you know. But your improvements are great, Wayne.

I realize I just did a "Fermat's last theorem" thing right there with the empty promises, but I do have something near ready.

[mediatechnology]

I tried to match the Studer component values using the equations from Weaver.

studer90m.JPG

Six of the eight resistors came close, but two were off by more than 10% ?

Calculated phase difference from the Studer schematic

studer90a.JPG

Calculated phase difference from the Weaver values

studer90t.JPG

Thanks Joe.
Frank gets the credit for straightening out the errors in Studer's implementation.

[AnalogJoe]

These 90° filters are not very common but have some intersting applications. I initially got interested on them to do IQ modulation and IQ summing for an application I had around 3-4 years ago. The work you have done on testing them on audio is great.

By the way, what got you hooked on them? I saw the thread you provided to "Gearspace", but I guess I was too lazy to read everything and make the connections to this thread.

[mediatechnology]

By the way, what got you hooked on them? I saw the thread you provided to "Gearspace", but I guess I was too lazy to read everything and make the connections to this thread.

emrr originally sent me the link to GroupDIY IIRC and we had a post here once about equal weighting of sum and difference in a compressor sidechain. Then the keyboard to mono thread at Gearspace. It seemed to fit both apps. But I guess overall just the shear oddity of it attracted me to it.

[Steven]

A very quick introduction: my name is Steven (really) and I'm from the Netherlands.
I just came across this website and found some interesting subjects, such as this topic on quadrature summing to mono. I wasn't aware of this summation method, although I work in the area of installed public address sound (70/100 V systems), including background music distribution in mono.
I know of the subjective quality loss of mono sound compared to the stereo original, but i had never thought of using quadrature filters as a solution.

I have used quadrature filters to create overlapping induction loops for hearing aids. There the phase shift between the loops is required to avoid that the sound disappears when moving from one loop to the next, because otherwise the fields would cancel at the crossover border.

Anyway, related to this, I think some people here might be interested to know that there is a freeware program to calculate very accurate quadrature filters. They are also used in single sideband receivers and/or transmitters, and it was developed for this purpose.

It's called QuadNet from Tonnesoftware.com, http://tonnesoftware.com/quad.html

Steven

[mediatechnology]

A very quick introduction: my name is Steven (really) and I'm from the Netherlands.
I just came across this website and found some interesting subjects, such as this topic on quadrature summing to mono. I wasn't aware of this summation method, although I work in the area of installed public address sound (70/100 V systems), including background music distribution in mono.
I know of the subjective quality loss of mono sound compared to the stereo original, but i had never thought of using quadrature filters as a solution.

I have used quadrature filters to create overlapping induction loops for hearing aids. There the phase shift between the loops is required to avoid that the sound disappears when moving from one loop to the next, because otherwise the fields would cancel at the crossover border.

Anyway, related to this, I think some people here might be interested to know that there is a freeware program to calculate very accurate quadrature filters. They are also used in single sideband receivers and/or transmitters, and it was developed for this purpose.

It's called QuadNet from Tonnesoftware.com, http://tonnesoftware.com/quad.html

Steven

Thank you for joining us!

I have used quadrature filters to create overlapping induction loops for hearing aids. There the phase shift between the loops is required to avoid that the sound disappears when moving from one loop to the next, because otherwise the fields would cancel at the crossover border.

That's brilliant.

Anyway, related to this, I think some people here might be interested to know that there is a freeware program to calculate very accurate quadrature filters. They are also used in single sideband receivers and/or transmitters, and it was developed for this purpose.

I recently downloaded his meter scale drawing utility and thought that URL looked familiar.

I'm going to try the quadrature utility. Thanks for the link. I was completely unaware of it.

[emrr]

Thanks for pointing out these things Steven!

[Steven]

Instead of using two sets of multiple single order allpass filters in cascade for a quadrature filter, it is possible to use half the number of second order sections, in this way saving on opamps. Unfortunately, most low-Q second order allpass filters have a significant attenuation, more or less inhibiting cascading.
Some 30 years ago, I designed a second order version that has a gain of 1 at a low Q, and uses normal E12 value resistors and capacitors. High-Q filters are not impossible to use, but a quadrature filter would need more sections, negating the benefit of using second order filters.
Another difficulty with second order allpass filters is their much higher sensitivity to component tolerance. If high accuracy is not needed, this can be tolerated.
See the circuit below for a comparison of two filters with more or less the same performance, one based on second order sections, the other on twice the number of first order sections. This simple circuit covers only one decade of frequency range with 90 degrees phase difference. For a wider bandwidth, the number of sections need to be increased.
The cascaded first order filters have unity gain by definition, regardless of tolerances. The second order section almost has flat gain, it dips only 30 mdB (gain = 0.9965) with the component values shown.

Steven

[mediatechnology]

I'm going to start running some experiments using I+Q summation for a stereo compressor sidechain.

True Power Summing works very well for a stereo compressor but requires two detectors with level trims to provide matching.
The only disadvantages I see are that it requires two RMS detectors and, more importantly, a stereo sidechain ahead of the detectors prior to their summation.
So if one wants to EQ the sidechain a stereo unit is required and, if external, a lot of patching.

Experiments in this thread with I+Q show that the envelope and total signal power are very close when stereo and I+Q are compared.
If close enough there are two advantages.
One RMS detector is eliminated and no trim is required to match levels.
And, most importantly, the pre-detector sidechain is now mono.

I'm thinking about calling it the "Poquito Compressor" for "a little bit of compression."

With a simplified mono path a pre-detector sidechain-only EQ using Bax, Tilt or a sliding HP/LP filter could be put on-board.