Klipsch Jubilees and Roy's brand new Xilica Settings

[MMurg]

On 9/1/2023 at 5:20 PM, mikebse2a3 said:

 

Yes it seems that they have been scattered around the forum from time to time and maybe they can be gathered again and pinned somewhere if Roy is OK with that.

 

I also caution everyone that not all Manufactures DSP respond exactly the same to the same settings and should be taken into account when transferring settings made on one brand/model into a different brand/model. 

 

For example the EV DX38 and EV DC-One respond exactly the same to the same settings but settings derived on them will not directly transfer to the Xilica models which require different parameters for the Shelv Filters to have an equivalent output signals desired.

 

miketn

 

I figured that since there is a way to convert Q to bandwidth and vice versa for PEQ filters, that there might be a similar conversion for slope to Q and vice versa for shelf filters (and then to/from bandwidth using the previous formulas for a Xilica DSP).  I think I may have found it, but I have one point of uncertainty.  I also don't have an EV DSP with which to confirm this.  I found a formula to convert the "slope factor" to Q and vice versa here: https://ez.analog.com/dsp/sigmadsp/f/q-a/64769/how-to-design-shelf-and-butterworth-hp-or-lp-6db-oct-12db-oct-filter.  Unfortunately, the "slope factor" is not the same thing as the "slope" that EV is using.  My best guess is that the "slope factor" is kind of equivalent to the "order" of the slope of the transition region (i.e. a slope factor of 1 would be 6 dB/octave).  So, if you took the EV slope values and divided by 6, then you should have the slope factor.  With that, you should then be able to use the following Excel formulas for a conversion spreadsheet.  The dB gain is also required in these formulas.  Of course, you would then need to convert the Q value to bandwidth using the previously posted formula to plug the value into a Xilica DSP.

 

To Convert Slope Factor and dB Gain to Q for a Shelf Filter in Excel (replace A1 with the cell location containing the Slope Factor value and replace B1 with the cell location containing the dB Gain value)
=1/(SQRT((10^(B1/40)+1/10^(B1/40))*(1/A1-1)+2))
 

To Convert Q and dB Gain to Slope Factor for a Shelf Filter in Excel (replace A1 with the cell location containing the Q value and replace B1 with the cell location containing the dB Gain value)
=(10^(B1/40)+1/10^(B1/40))/((1/A1^2)-2+10^(B1/40)+1/10^(B1/40))
 

If anyone knows if my assumption about the EV slope to slope factor relationship is correct or not, or if you have both an EV and a Xilica DSP and can check this, please let me know.

 

-----------------------------------------------------------------------------------

Edit:  I was doing more looking around some more and I found an interesting tidbit here: https://github.com/WebAudio/web-audio-api/issues/2428.  It talks about the parameters that can be used to define shelf filters in a certain program.  I've pasted it here and highlighted what I found interesting.

 

lowshelf f0[k] width[q|s|d|o|h|k] gain
Double-pole lowshelf filter.
highshelf f0[k] width[q|s|d|o|h|k] gain
Double-pole highshelf filter.
Reference:
q Q-factor (default).
s Slope (shelving filters only).
d Slope in dB/octave (shelving filters only). Also changes the
definition of f0 from center frequency to corner frequency (like Room EQ
Wizard and the Behringer DCX2496).
o Bandwidth in octaves.
h Bandwidth in Hz.
k Bandwidth in kHz.

 

The text in red show that the slope can be defined as s (slope factor) or d (slope in dB per octave).  Using one or the other effects how the filter frequency is defined (center or corner).  So, the conversion I proposed above when coming from an EV setting using dB/Octave is likely not that simple.  While I think the conversion from slope factor to Q and vice versa is likely correct, it is likely not useful to go from EV slope settings to Xilica settings for shelf filters without more information.  😞 

 

[mikebse2a3]

20 hours ago, MMurg said:

The text in red show that the slope can be defined as s (slope factor) or d (slope in dB per octave).  Using one or the other effects how the filter frequency is defined (center or corner).  So, the conversion I proposed above when coming from an EV setting using dB/Octave is likely not that simple.  While I think the conversion from slope factor to Q and vice versa is likely correct, it is likely not useful to go from EV slope settings to Xilica settings for shelf filters without more information.  😞 

 

I will try to post more information as time allows but I will add this observation I made during the programming experimentations of the Lo-Shelf Filters from the EV-DX38/EV DC-One units to the Xilica XD4080 using there editor software for response curve simulations.

 

miketn

 

 

 

[Edgar]

It looks like Xilica used the parameter definitions for shelving filters from the RB-J Cookbook. Robert Bristow-Johnson defined the "shelf slope" parameter as "... (for shelving EQ only). When S=1, the shelf slope is as steep as it can be and remain monotonically increasing or decreasing gain with frequency."

 

Years ago I designed a DSP package for EV that included shelving filters. I do not know if they continued to use my definitions of the parameters in subsequent products, but if they did then you can assume that the EV "shelf slope" parameter is equal to 1 in all cases. The difference between the "6 dB" shelf and the "12 dB" shelf is that the 6 dB version is based upon a 1st-order transfer function, and the 12 dB version is based upon a 2nd-order transfer function.

[PrestonTom]

To further confuse the issue.

 

Shelving filter parameters can be all over the place depending on the manufacturer. For most the "frequency" of the shelf is the 3 db down point (behringer and yamaha etc). However for some it is the midway point of the transition (I forget if that is "mid" freq on a linear space or log space). I wasted a number of hours "learning about this feature". I think miniDSP may do this, but don't quote me on that.

 

Not only that (maybe it was already mentioned), for specifications done on PEQ, although the Q and CF can be similar, the rolloff  with frequency may have a difference in steepness beyond the initial -3dB down point  (ie, different shape).

 

As always don't assume assume it might be a simple conversion. If exact values are not needed it may not make much of a difference, in which case ignorance is bliss.

[Edgar]

10 minutes ago, PrestonTom said:

To further confuse the issue.

 

Shelving filter parameters can be all over the place depending on the manufacturer. For most the "frequency" of the shelf is the 3 db down point (behringer and yamaha etc). However for some it is the midway point of the transition (I forget if that is "mid" freq on a linear space or log space).

 

In my designs, it was the "natural" frequency in the prototype Laplace-Domain transfer function. That frequency may not show any obvious relationship to the 3dB frequency or the "mid" frequency, depending upon the values of the other parameters -- for example, what is the "3dB" frequency in a shelf that only boosts 2dB? But my definition is at least self-consistent.

[OO1]

On 9/1/2023 at 5:13 PM, mark1101 said:

 Straight from the man.

wow

[mikebse2a3]

Roy’s Underground {Jubilee with K402 & TAD 4002 v1.0 (8/19/2008)} parameter settings from EV DX38 program and the resulting curves in the Race Editor Software for the EV DX38.

 

miketn

 

 

 

 

NOTE:  The required Lo-Shelf parameter changes necessary for the Xilica XP/XD-4080 DSP output curve simulation to match Roy’s original EV DX38 output curve simulation.

 

EV DX38: LOSLV (FREQ: 2.5K), (Q/Slope: 12), (Level db: -1)

Xilica:  Lo-Shelf (FREQ: 2500Hz), (1.80 Oct), (Q=0.751), (Level -1.00dB)

 

EV DX38: LOSLV (FREQ: 5.4K), (Q/Slope: 12), (Level db: -10)

Xilica: Lo-Shelf (FREQ: 4200Hz), (1.80 Oct), (Q=0.751), (Level -10.00dB) 

 

Note:  Tap on the images to enlarge for point to point comparison and even though the scale parameters are displayed differently a careful point to point comparisons will show the exact matching of the output curves from the EV DX38 and Xilica XP/XD 4080 of the two programs with these Lo-Shelf parameter changes in the Xilica XP/XD 4080 from Roy’s original EV DX38 parameters.

 

 

[mikebse2a3]

I thought it might be helpful to simplify for demonstration purposes to just do the simulations of the (EV DX38 LOSLV)/(Xilica XP/XD 4080 Lo-Shelf) filters to demonstrate how well the parameters changes in the Xilica match the EV when implemented.

 

miketn

 

 

EV DX38: LOSLV (FREQ: 5.4K), (Q/Slope: 12), (Level db: -10)

Xilica: Lo-Shelf (FREQ: 4200Hz), (1.80 Oct), (Q=0.751), (Level -10.00dB) 

 

 

 

 

 

 

 

NOTE: Both of the LOSLV/Lo-Shelf filters combined in this comparison below.

 

EV DX38: LOSLV (FREQ: 2.5K), (Q/Slope: 12), (Level db: -1)

Xilica:  Lo-Shelf (FREQ: 2500Hz), (1.80 Oct), (Q=0.751), (Level -1.00dB)

 

EV DX38: LOSLV (FREQ: 5.4K), (Q/Slope: 12), (Level db: -10)

Xilica: Lo-Shelf (FREQ: 4200Hz), (1.80 Oct), (Q=0.751), (Level -10.00dB) 

 

 

 

 

 

[mikebse2a3]

NOTE: I’ve reposted this from an old thread of mine here in case anyone uses one of these units.

EV DX38 comparison to EV DC-One to demonstrate that both units parameter set points respond identical to each other.

 

For any DSP Active Crossover users that might be interested in either unit here are some test comparison showing how these units perform very close to each other on the same (KHJ-K402/TAD4002) program settings.

The EV DX38 = Red Curve

The EV DC-One = Blue Curve

Test between my units show the EV DC-One has about 0.7db to 0.9db more sensitivity/gain across the frequency spectrum shown.

One other observation is an insignificant slight difference in the 32Hz PEQ filter's actual center frequency with the (EV DC-One = 32.2Hz) and the (EV DX38 = 31.8Hz). 

miketn

 

Here are the LF Channels comparison:

 

 

 

Here are the HF Output Channels comparison.

The EV DX38 = Red Curve

The EV DC-One = Blue Curve

miketn