paralle HiVi b3s - what happens to the Notch?

[zenon]

In theory if you had a given notch filter designed for a speaker and you were to add another driver wired in parallel how does the Notch filter change?

I think you would:
R5 / 2
L4 / 2
C3 * 2

Correct?

I plan to adapt Zaph's b3s design into a bipole.
http://www.zaphaudio.com/audio-speaker18.html

[holdent]

Zaph uses a parallel notch filter for this project, not a series filter, so the RCL values won't change; they would for a series filter.

Lalena.com has good parallel and series notch filter calculators with a good explanation of how they are used. See:

http://lalena.com/Audio/Calculator/ParallelNotch/
http://lalena.com/Audio/Calculator/SeriesNotch/

[sreten]

Originally posted by holdent

Zaph uses a parallel notch filter for this project, not a series filter,
so the RCL values won't change; they would for a series filter.

Hi,

I do not know how you reckon this to be true, its not.

A) R5 / 2, L4 / 2, C3 * 2 would be correct for parallel wired drivers.

Hoever this is complicated by the removal of R1 and L2 for a bipole.

The above A) would be correct for R1/2 and L2/2.

But with R1 omitted the circuit will cause a larger dip than intended.

Try downloading the free TinaTi circuit emulator and play with it.

/sreten.

[zenon]

I guess another question is, how "flat" should I shoot for?

For example, say the notch was attenuating 6dB. Now, since wiring in parallel and doubling surface area will result in a 6dB output increase (3dB + 3dB). Would I want the notch to attenuate 12dB (6dB + 6dB)?

OR

Since it's a Bipole we assume at those high frequencies the rear driver won't be adding to the front driver's output equally as it will for the lower octaves and thus conclude the Notch filter should attenuate 6dB.

I think the latter is what I have to shoot for, I mean if the rear driver contributed equally at all frequencies then I would still need baffle step correction. Since this would be equivalent to having both drivers on the front baffle!! Which isn't the case!

I'm not familiar with the TinaTi software but I will have to look more thoroughly tonight for it.

[sreten]

Hi,

A bipole does not need baffle step correction.

The notch filter needs to be tweaked to work without R1 and L1.

Estimate driver inductance at notch frequency to use in Tina Ti.

/sreten.

[zenon]

That Tina looks real neat. Can you help me with my analysis?

I think what I have to do is simulate the whole circuit at 8kHz then eliminate the BS compensation and figure out what the value of R5 must be to have an equal total circuit resistance at that frequency.

Then at that point I apply the rules you suggested above.

I pulled the driver impedance off the graph Zaph provided of the raw driver measurement.

TIA.

[zenon]

I didn't have much luck with that model. I think this however is what I'm looking for. I modeled the speaker using T/S parameters as an inductor with resistance. Entered DC resistances for all the inductors and tried to match the voltage across the driver to be equal.

Assuming the model is correct I think I was able to find a good equivalent!

[zenon]

image 2

[sreten]

Hi,

You don't need to connect anything to the op-amp.

You can plot the frequency response directly in Tina.

model two circiuts (omit L1, one circuit has R1, the other doesn't)

Get the responses to look the same (by reducing R5 somewhat).

/sreten.

[DSP_Geek]

Originally posted by holdent
Lalena.com has good parallel and series notch filter calculators with a good explanation of how they are used. See:

http://lalena.com/Audio/Calculator/ParallelNotch/
http://lalena.com/Audio/Calculator/SeriesNotch/

I have some doubts about their parallel notch filter. The value of the resistor doesn't change with the specified attenuation, and the DCR of the speaker cannot be specified, so that makes the network suspect.