A Second Order Elliptic Equalizer for Vinyl Mastering

[mediatechnology]

The EEQ boards pass test...


Second-order Elliptic Equalizer First Build

Crosstalk Curves for 75, 150, 300 Hz

Left-only input.
Top trace is left out.
Bottom traces are the right channel output with the EEQ switched from 75 to 150 and 300 Hz.


Second-order Elliptic Equalizer Crosstalk Curves for 75, 150, 300 Hz


Comparison of 6 dB vs. 12 dB Per Octave Lateral Versus Vertical Response

Left-only input. EEQ set at 300 Hz.
Top trace is lateral (mono) response.
Bottom traces are the vertical output with the EEQ plotted at 6 and 12 dB per octave.


Second-order Elliptic Equalizer Lateral vs Vertical at 300 Hz with 6 or 12 dB per octave slope


Comparison of 6 dB vs. 12 dB Per Octave Crosstalk Response

Left-only input. EE set at 300 Hz.
Top trace is left out.
Bottom traces are the right channel crosstalk with the EEQ plotted at 6 dB and 12 dB per octave


Second-order Elliptic Equalizer Crosstalk Comparison at 300 Hz with 6 or 12 dB per octave slope

I'll be posting some sound files soon.

Conclusions

1) The steeper highpass slope of the EEQ's second-order vertical response permits, for the same amount of filtering, a lower EEQ frequency to be used.

2) The steeper low pass response of the second-order crosstalk curve improves midrange separation and imaging.

3) Factors 1 and 2 in conjunction lower the crosstalk significantly to maintain midrange stereo imaging.

4) The summed output is mono compatible.

Applications

The applications for the EEQ include vinyl cutting, "bass focus" to improve low-end on headphones, earbuds and small speakers, subwoofer steering and vinyl playback warp reduction.

[mediatechnology]

I updated the Elliptical Equalizer comparison table I had posted in an MS thread. (viewtopic.php?f=6&t=622&p=7414#p7414)


Elliptical Equalizer Comparison. Neumann EE-70/EE-77, VAB-84, LR±S Width, MS Mid Side, MS Mid Side with Correction Allpass.

[mediatechnology]

For those interested in a PC board for the Elliptic EQ send me a PM.

I don't currently have inventory of bare boards but have pre-orders for a couple and need commitments for a few more before I submit an order for more.

EEQ-12 Elliptic Equalizer: https://www.ka-electronics.com/kaelectr ... tic_EQ.htm

[AnalogJoe]

Wayne, that looks like a nice EQ.

AFAIK, there is no such thing as a first order elliptic filter, meaning that the polynomials in the transfer function are never below 2nd degree in order to be an elliptic filter, however I think you are refering to the final roll off which can be 6dB per octave which corresponds to a first order roll-off.

As for the Douglas Self book, I disagree with him with what he calls Elliptic filters, he uses elliptic filters with 0dB passband ripple, that is not an elliptic filter, that is a Chebyshev Type II filter, in order to be an elliptic filter there must be ripple both in the pass and stop bands, large or small but there must be ripple. Actually, the elliptic filter can be considered the father of all filters, if you reduce the stop band ripple to zero you get Chebyshev Type I, if the pass band ripple is zero you get Chebyshev Type II, and if both the stop band and pass band ripple is zero you get Butterworth.

[KMN]

Wayne,

I've been playing with your circuits in LTspice. I believe I have come up with a good way to listen to the information removed. This may be useful for deciding whether the chosen filter roll off point is too low or too high based on musical contents presence within the stripped material or if no music is present maybe you could try a higher corner frequency.

It requires another filter of the time alignment type.....whatever you call that thing.

The idea is pass the encoded side channel information through both the high pass filter and the time alignment filter.... then...
- Subtract the high pass filtered side from the time aligned side for left ear noise component.
- Subtract the time aligned side from the high passed side to get the right ear noise component.

Stereo low frequency noise. The thing is, who can even hear warp frequencies? I think the biggest benefit will be listening to ensure there is little to no music present.

I think the above is correct but not 100% sure and wanted to run it past the forum.

Kevin

[mediatechnology]

What I do to listen to what is being mono'd by the EEQ is to feed the left input, dead patch the right channel input, mute the left channel output and monitor the right output.
Doing this allows me to hear the deliberately introduced LF crosstalk from L into R.

For warp removal I think there is a simple way to quantify what is being removed.
Sample the difference signal (vertical) pre-filter and the difference post-filter then null the two.

The difference signal pre and post-filter is available on the EEQ board.
Nulling the two requires an additional outboard THAT1240.

I have an EEQ on the bench right now awaiting final test - it would be easy to try it.

[mediatechnology]

Wayne, that looks like a nice EQ.

AFAIK, there is no such thing as a first order elliptic filter, meaning that the polynomials in the transfer function are never below 2nd degree in order to be an elliptic filter, however I think you are refering to the final roll off which can be 6dB per octave which corresponds to a first order roll-off.

As for the Douglas Self book, I disagree with him with what he calls Elliptic filters, he uses elliptic filters with 0dB passband ripple, that is not an elliptic filter, that is a Chebyshev Type II filter, in order to be an elliptic filter there must be ripple both in the pass and stop bands, large or small but there must be ripple. Actually, the elliptic filter can be considered the father of all filters, if you reduce the stop band ripple to zero you get Chebyshev Type I, if the pass band ripple is zero you get Chebyshev Type II, and if both the stop band and pass band ripple is zero you get Butterworth.

Thank you.
I think you may be confusing "elliptic filter" with "elliptic equalizer" which is easy to do.

Elliptic Equalizer is a vinyl mastering term of art not in any way related to the Elliptic Filter topology.

There'a an explanation of the etymology here: https://ka-electronics.com/shop/index.p ... duct_id=65

An "elliptic equalizer" is used during vinyl mastering to prevent playback skipping on music having wide stereo bass. The bass-to-mono elliptic EQ gets its name from the width of ellipses in the recorded groove or Lissajous patterns on an oscilloscope. If the width of the recorded groove becomes too narrow from stereo bass, which produces large "vertical modulation" of the cutter head, the playback stylus can be pinched out of the groove.

[AnalogJoe]

Wayne, that looks like a nice EQ.

AFAIK, there is no such thing as a first order elliptic filter, meaning that the polynomials in the transfer function are never below 2nd degree in order to be an elliptic filter, however I think you are refering to the final roll off which can be 6dB per octave which corresponds to a first order roll-off.

As for the Douglas Self book, I disagree with him with what he calls Elliptic filters, he uses elliptic filters with 0dB passband ripple, that is not an elliptic filter, that is a Chebyshev Type II filter, in order to be an elliptic filter there must be ripple both in the pass and stop bands, large or small but there must be ripple. Actually, the elliptic filter can be considered the father of all filters, if you reduce the stop band ripple to zero you get Chebyshev Type I, if the pass band ripple is zero you get Chebyshev Type II, and if both the stop band and pass band ripple is zero you get Butterworth.

Thank you.
I think you may be confusing "elliptic filter" with "elliptic equalizer" which is easy to do.

Elliptic Equalizer is a vinyl mastering term of art not in any way related to the Elliptic Filter topology.

There'a an explanation of the etymology here: https://ka-electronics.com/shop/index.p ... duct_id=65

An "elliptic equalizer" is used during vinyl mastering to prevent playback skipping on music having wide stereo bass. The bass-to-mono elliptic EQ gets its name from the width of ellipses in the recorded groove or Lissajous patterns on an oscilloscope. If the width of the recorded groove becomes too narrow from stereo bass, which produces large "vertical modulation" of the cutter head, the playback stylus can be pinched out of the groove.

Whoops, my bad! you are absolutely right I thought you meant elliptic filter.

[KMN]

What I do to listen to what is being mono'd by the EEQ is to feed the left input, dead patch the right channel input, mute the left channel output and monitor the right output.
Doing this allows me to hear the deliberately introduced LF crosstalk from L into R.

Hmm. I may have to give that a try too. Good stuff!

Sample the difference signal (vertical) pre-filter and the difference post-filter then null the two.

I think this may be similar to what I was attempting to describe, perhaps stated a bit more elegantly.

[KMN]

Sample the difference signal (vertical) pre-filter and the difference post-filter then null the two.

Maybe you could feed this noise signal to some kind of event indicator meter or LED circuit.