Page 1 of 2

Stereo Width Controller Construction Information

Posted: Tue Apr 14, 2015 11:44 am
by mediatechnology
This is the construction thread for the Stereo Width Controller PC board.

Stereo Width Controller PC Board: http://ka-electronics.com/shop/index.ph ... arch=width

Image
Stereo Width Controller PC Board

Stereo Width Controller Schematic

Image
Stereo Width Controller Schematic

Assembly Instructions

Assembly instructions: http://www.ka-electronics.com/images/pd ... ctions.pdf
Schematic suitable for printing: http://www.ka-electronics.com/images/jp ... _small.jpg

Parts List and Bill of Materials

The Mouser Project Manager bill of materials for the Width Controller can be used to order directly from Mouser or used simply as a guide for components you source locally:

Note that the 10K linear taper pot is not included in the bill of materials.

Bill of Materials without THAT ICs: https://www.mouser.com/ProjectManager/P ... 96947b82d5

Bill of Materials with THAT ICs: https://www.mouser.com/ProjectManager/P ... c66b355bb9

Semiconductors

4) THAT1246 IC1, IC2, IC5, IC7
4) THAT1240 IC3, IC6, IC8, IC9
3) THAT1646 IC4, IC10, IC11
1) LM78L12 IC12
2) 1N4004 D1, D2
4) 1N4148

Electromechanical and Connectors


2) Relay DS2Y-S-DC12V RY1, RY2
11) 8 pin sockets
10) 3 pin Phoenix terminal blocks Mouser 651-1725669

Capacitors

15) 0.1/100V Mono Ceramic Mouser 581-SR201C104KAR C1-C15
6) 10uF radial NP or polarized 25V or greater C20-C25
1) 10uF radial polarized 25V or greater C32
2) 47uF/35V C30, C31

Resistors

2) 1R 1/4W R1, R2. (Flameproof/fusible is better but generic carbon film if that's all that's available. If you do use CF space it off the board.)
2) 100K 1% Mouser 271-100K-RC, R3, R4

PC Board Stuffing Guide:

Image
Stereo Width Controller PC Board Stuffing Guide


How it Works


The core circuitry of the LRS-1 Width controller is IC3, IC6, IC8 and IC9.

THAT1246 balanced line receivers along with THAT1646 outputs provide both buffering and balanced inputs and outputs.

Like any of the matrices using differential line receivers it is important that the driving source impedance is an op amp output.
If the source is not buffered, the source impedance becomes part of the matrix and degrades its accuracy.
The balanced inputs provide the low source impedance required by IC3, IC8 and IC9.

THAT1646 output stages provide L and R balanced outputs.

IC3, a THAT1240, provides a Side signal that is that is equal to (L-R).

(A 1246 was originally used for IC3 to provide the needed (L-R)/2 signal but to maintain signal level at the insert send a unity gain THAT1240 was used and attenuation added at IC7.)

IC4 and IC5 provide an external Side Insert with RY1 providing bypass.

A low pass filter in the external Side Insert makes the circuit become an Elliptic Equalizer for vinyl mastering providing low frequency crossover to mono.

IC6, a THAT1240, along with a 10K linear potentiometer (or stepped switch) create a Side signal with a voltage gain, "k," that ranges from -1 to 0 to +1.

This permits Side to be subtracted or added to Left and Right.
100K resistors bridging the wiper of the pot to both ends of the pot provide bias current return for the THAT1240 input if the wiper lifts during rotation or switching.
If a stepped switch is used, Goldpoints are supplied in make-before-break and the 100K's can be removed to increase taper accuracy.
RY2 defeats the Width circuit for signal comparison.

IC7 attenuates the Side signal by a precise -6dB to provide a k{(L-R)/2} output.

THAT1240 summation stages IC8 and IC9 either add or subtract Side from Left and Right.

The Side signal at the output of IC7 is subtracted from L and added to R.

"Side" is evenly distributed between Left and Right but has opposite polarities.
Thus, Left contains a side component that is L + 0.5(side).
Right has a R -0.5(side) component.

If k=1 (IC6 is not inverting polarity) and the attenuation of IC7 is 0.5 then:

L = L +0.5(side) -0.5(side) = L
and
R = R -0.5(side) +0.5(side) = R.
Thus with no Side signal to make it stereo, L then equals R and the resulting signal is mono.
Obtaining mono is a subtractive process.

If k=0 then:

L= L +0.5 side
and R = R -0.5 side.
The result is normal stereo.

If k=-1 (IC6 is providing polarity inversion) and and the attenuation of IC7 is 0.5 then:

L = L +0.5(side) +0.5(side) = L + side
and
R = R -0.5(side) -0.5(side) = R - side.
Thus the amount of side in left and right are doubled relative to normal stereo where the side terms are +0.5 and -0.5.
Expanding width, or the side component, is an additive process.

The X-Y diagrams below show how the diagonal side axis is either fully collapsed to produce mono or doubled in width relative to normal stereo to expand width.

The following are X-Y Oscilloscope photos with 1 kHz applied to the Left channel and 10 kHz applied to the right.
When the polarity flipper's output is 0V, the Left/Right inputs are unprocessed and provide a "normal" stereo image.


Image
A Stereo Width Control Using the +/-S Method, 1 kHz Left, 10 kHz Right, Set for Mono.

Image
A Stereo Width Control Using the +/-S Method, 1 kHz Left, 10 kHz Right, Set for Stereo.

Image
A Stereo Width Control Using the +/-S Method, 1 kHz Left, 10 kHz Right, Set for Wide.

Image
With the Width control centered crosstalk is excellent.

Image
With the Width control bypassed, crosstalk is essentially that of the sound card used to perform the measurement.

A Simple Elliptic Equalizer Using the Width Controller

Posted: Tue Apr 14, 2015 7:42 pm
by mediatechnology
This post shows how to make a simple passive first-order Elliptic Equalizer for Vinyl Cutting or subwoofer crossover using the Stereo Width Controller Insert.

Image
A passive low pass filter in the Side Insert Turns the Stereo Width Controller into an Elliptic Equalizer

The THAT1646 which drives this network has high current outputs.

The filter shown is either 150 or 300 Hz.
A 75/150 Hz filter can be made by making the caps 1 uF.
Other values of capacitance can be bridged across tip and ring to form a low pass filter in the side insert.
When computing resistor values for custom filters include the internal 25 Ohms per leg (50 Ohm total) output resistance of the THAT1646.

If the Width control is set to mono, the separation will converge to mono.
Intermediate settings of the Width control allow permit some LF separation within the pass-band of the filter.

The result is a low frequency crossover to mono (to reduce Elliptic vertical modulation) at low frequencies similar to the Neumann EE-70 or EE-77.

Image
Crosstalk curves with the Stereo Width Controller configured as a 150 Hz or 300 Hz Elliptic Equalizer.

This plot is a 300 Hz EE curve with the Width Control varied:

Image
Crosstalk curves with the Stereo Width Controller configured as a 300 Hz Elliptic Equalizer showing Variable Bass Width.

Re: Stereo Width Controller Construction Information

Posted: Sat Feb 13, 2016 6:33 pm
by Gold
I did an R&D build for this. Wayne sent me a PCB a while ago but I didn't have a chance to check it out. At first blush i didn't see how this was significantly different than using the M/S board. It's significantly different and well worth checking out.

Since I have elliptical EQ's already I didn't need that function. I've always wanted a high frequency widener. When I want to widen something I rarely use gain on the S channel. I usually boost a shelving EQ on the side channel to approximate widening the top end. This can actually do what I want. I don't think there is anything out there that does this.

Instead of putting a LPF in the insert to make an EE I put a HPF in to make the HF widener. It is like the first order passive LPF but inverted. The capacitors are in series like the 499Ω resistors. The resistor is in shunt like the 0.47uF caps. I wanted the crossover to be about 1K Hz. I used 0.22uF caps and a 604Ω resistor.

Listening tests made me happy but I have to do some more listening.

Image

Re: Stereo Width Controller Construction Information

Posted: Sat Feb 13, 2016 8:12 pm
by Gold
I think I screwed up the crossover frequency. Since this is a balanced filter the R value has to be divided in half. I believe with the values I have the crossover is at 604Hz. I need to change the caps to about 0.13uF to get 1K Hz. Or I could be wrong again...

Re: Stereo Width Controller Construction Information

Posted: Sun Feb 14, 2016 8:45 am
by mediatechnology
Thanks for the feedback on the Widener!

I come up with 0.13 uF per leg for a 1 kHz filter fc but I think the effective frequency is off by an octave due to the "0.5" gain term in the LR+-S process.
Electrically the gain of -6dB, the "0.5," occurs at IC7.

It's the same octave error we saw in the initial EE-77 vs. "EE-84" elliptic equalizer comparisons and soundfiles.
I come to the revelation I'm off by a factor of two here: viewtopic.php?f=6&t=622&hilit=vab+84&start=50#p7414
Much earlier in that thread I sent sound files: The "EE-84" files were an octave lower than the EE-77's and therefor not a fair comparison.

In this situation, not an EE with a LP filter, but a Widener using a HP filter, the octave error is still with us.
I think that for 1 kHz filter values the widening may begin at 500 Hz.
(Though it could be at 2 kHz after I finish coffee.
This clearly requires more thought which direction the error factor of two applies...
An experiment might save brain cells and coffee.)

You can also vary the shunt R within a range to tune it.
The minimum is determined by the THAT1646 and 604R is a good value.
The shunt R is loaded by the approximately 24K (+/-40%) differential impedance of the THAT1246 line receiver.
A good upper bound for the external shunt R might be 2K or more.

Re: Stereo Width Controller Construction Information

Posted: Sun Feb 14, 2016 9:50 am
by Gold
mediatechnology wrote:Thanks for the feedback on the Widener!
I come up with 0.13 uF per leg for a 1 kHz filter fc but I think the effective frequency is off by an octave due to the "0.5" gain term in the LR+-S process.
Electrically the gain of -6dB, the "0.5," occurs at IC7.
I must admit I still don't fully understand all the implications of the 0.5 gain term. I was under the mistaken impression that a 2 pole filter needed to be scaled an octave up but not the single pole. I need to read the thread a few more times.

The next thing I'm going to do is switch out the passive filter for unity gain Sallen-Key types. I'll try one and two pole at a few frequencies to get a feel how the different filter types sound. Since all the Sallen-Key filters I've seen are unbalanced can you ground one side of IC5 1246? I know as long as you have the caps installed you can ground one side of a 1646.

I am also going to switch out the pot for a rotary switch. I have 12 position switches. I think I'll start with a switch that covers the full range from Mono to Wide. If that seems too coarse I'll do another one that only widens and has twice the steps.

I think this thing is going to develop into a very unique processor.

Re: Stereo Width Controller Construction Information

Posted: Sun Feb 14, 2016 2:38 pm
by mediatechnology
I think this thing is going to develop into a very unique processor.
Thanks Paul!

You sure can use a single-ended SK filter.
Ground the 1646 output and 1246 "-" inputs.

I have a Width Controller on the bench right now and will try a 1 kHz fc filter to see what the effective frequency is.
The thing to measure to determine the effective frequency is to look at the Vertical and see where it begins to increase.
The 3 dB increase in the vertical - the L-R - will be the effective frequency.
The octave relationship should hold regardless of the filter order.
The more I think about it a 1 kHz filter HP will have an effective frequency of 2 kHz.
(In the EE example, which uses a LP filter to narrow width, an opposite relationship exists. A 300 Hz fc filter will have an effective 3 dB vertical crossover at 150 Hz.)

I should point out that crosstalk measurement using something like RMAA or a meter are misleading since they only measure level and not relative polarity.
That's the same reason why carts measuring only 20 dB sound wider than 20 dB: The crosstalk polarity is inverted and they are adding perceived width.

Re: Stereo Width Controller Construction Information

Posted: Sun Feb 14, 2016 2:58 pm
by Gold
mediatechnology wrote: The more I think about it a 1 kHz filter HP will have an effective frequency of 2 kHz.
I guess I got it almost right by mistake. I used a HPF with 0.22uF's and a 604Ω. That would be at about 603Hz. That doubled is about 1k2Hz.

Re: Stereo Width Controller Construction Information

Posted: Sun Feb 14, 2016 3:30 pm
by mediatechnology
OK, I did a reality check.

0.47 uF per leg (0.235 uF) and 680 Ohms is approximately 1 kHz - 3dB point. (996 Hz).
And the +3 dB point of Vertical (L-R) is also about 1 kHz.

It does sound nice.
Still need to think about that 2X error in the multi-pole EE...

Re: Stereo Width Controller Construction Information

Posted: Mon Feb 15, 2016 12:03 pm
by Gold
mediatechnology wrote:OK, I did a reality check.
It does sound nice.
I think we may be the first humans to hear this. I could have set something like this up in software but never bothered. I don't think too many others have thought to do this.