does the exact order of modules in the chain matter?

emosms

Member
2010-09-28 1:19 am
Hi,

I am going to assemble a chain of modules on slots (smth like motherboard with modules) to condition subwoofer signal.

Does the order of modules matter? F.ex., due to construction matters I consider:

1. Buffer/summer of L+R chanel + Linkwitz transform
(both circuits sharing dual opamp, linkwitz transform needs a buffer)
2. Lowpass filter.
3. Phase shifter (elaborate schematics, requires dual opamp, has buffer)
4. Subsonic.

Eventually can switch 3 and 4. I want the phase shifter 3-rd or 4-th, so that the phase pot shares a dual digital pot whit the volime pot
(before the power amp stage). It will be easier pcb-wise if the digipot is closer to positions 3/4.

These are mostly construction/design considerations.
I assume, that the order we apply the signal conditioning does not really matter.

I only care if the input/output of each module has reasonable impedance.
And - the volume pot to be at the end of the chain for better signal to noise ratio (line level input for the whole chain)

Is that all ok :D :D ?

Kind Regards
 
Last edited:

emosms

Member
2010-09-28 1:19 am
I also thought about subsonic filter somewhere in the beginning.

But, the only modules that could be combined together are summer/buffer and the linkwitz. The Linkwitz circuit consist of 1 opamp, no buffer (explicitly stated it needs a buffer).

On the other hand, if there might be some low freq pulsations due to all that electronics, the subsonic at the end might filter it...
Anyway, to have such low freq pulsations appearing in the preceding modules is a problem.
It should not occur and means that there is something wrong... :D.

Eventually putting the subsonic as number 2, then lowpass and phase shifter (phase pot combined in a dual digipot with the volume pot)
Another + for subsonic at the end - I can omit it altogether... Just a bit concerned that I may damage the subwoofer driver.

p.s. the modules circuits:
- phase: http://sound.westhost.com/project103.htm //has input buffer, output impedance 100 ohms?
- subsonic: http://www.sound.westhost.com/project99.htm //stated input impedance 47 k, recomended source 100 ohms(but can go ok with 10k), output impedance - 100 ohms?
- low pass. Actually I plan to stack two of these in series, to get higher order (24db oct): http://www.sound.westhost.com/project155.htm
// stated input impedance of ...1k for the lowpass circuit. Output impedance 100 ohms?
- linkwitz, needs a buffer: http://www.sound.westhost.com/project71.htm // 100 ohms output impedance?
the buffer is the first opamp from figure 1, the Linkwitz is fugre 2. The buffer from figure 1 is also a signal summer
 
Last edited:

Zero D

Member
2009-08-06 11:11 am
Here's the order i would do with the sound.westhost circuits.

1 - Input Mixer and Phase Switch

2 - Linkwitz Transform Circuit

3 - Low-Pass Filter

4 - Subsonic Filter

The Low-Pass Filter minimum -3dB frequency is a HIGH 1.0kHz. I expect that you want it a LOT lower than that, for eg 80Hz ? I can work out the values of C1/2/3 to change it, if you'ld like ? It will still be variable though. Let me know what f you want it to start from.

Linkwitz transform is normally only used with Closed box subs, NOT ported. What are you planning to do ?
 

emosms

Member
2010-09-28 1:19 am
Linkwitz is to fix the simulation response curve of a driver.
40L ported box, f.ex. this one: ES 250.5 - Hertz Energy car audio subwoofers
To get 0 db at 30 hz

Low pass - I will be glad if you help. Better give a guidance on how to recalc values. Stacking two in series is ok?
I want that the subwoofer can't be really heard alone.. Due to the room plan, it is going to sit between left and right cabinets, 2 meters away of each...

You are right about the frequency.
The left and right channel are closed box drivers. I expect F0 at around 80 hz.
Anyway, I want to make it variable (my expectations might be wrong),or, to adjust it to my liking.
Also, to play with arduino - 4 pots should not be a problem :)

I was looking at the more elaborate phase shifter circuit. What if just flipping the phase 180" does not give the best result in the room?

The stereo speakers are closed box and they go at 6db/oct bellow F0. The low pass as I plan it now - 24/db oct. There is some uneven overlaping bellow 80hz
The variable phase shifter may help to get the best results for the lower bass.
I would have variable low pass, subwoofer seporate volume, and the phase shifter...
The stereo cabinets are not cut 24db/oct bellow the crossover point as the sub LPF does.

Regards
 
Last edited:
You can't Linkwitz Transform a ported box.

Below the port tuning, the output from the back of the cone meets the output from the front of the cone. This means the cone is pretty much free to flap around at very low frequencies, as if there was no cabinet at all.
A Linkwitz Transform boosts the low end of a sealed box in order to extend the low-frequency cutoff. This is possible because sealed boxes don't unload at low frequencies like ported boxes do.

Applying low-frequency boost below the port tuning is a very good way to kill drivers from over-excursion.

Chris
 

emosms

Member
2010-09-28 1:19 am
You can't Linkwitz Transform a ported box.

Below the port tuning, the output from the back of the cone meets the output from the front of the cone. This means the cone is pretty much free to flap around at very low frequencies, as if there was no cabinet at all.
A Linkwitz Transform boosts the low end of a sealed box in order to extend the low-frequency cutoff. This is possible because sealed boxes don't unload at low frequencies like ported boxes do.

Applying low-frequency boost below the port tuning is a very good way to kill drivers from over-excursion.

Chris

It is very close to the port tunning.
BassBox suggest port tuning at 28.smth Hetz.

I guess I could apply +3db at f3 (26.smth hertz)

The cone movement simulation shows problems starting at 21hz/ 80W rms.
Pretty ok at 25 hz.

Linkwitz puts some severe gain in the lows for closed box designs.
Requiring monster oversizing in terms of power (both amp and driver)

I just want to equalize the response curve with some +3db (and double the power...)
Nowthere is a peak of +3db at 50hz and -3db at 26/7 hz

The graphs:

[IMGDEAD]http://s24.postimg.org/6433po2np/driver_params.jpg[/IMGDEAD]
 

Zero D

Member
2009-08-06 11:11 am
@ emosms

Yes i know what a Linkwitz Transform does, that's why i mentioned it's not for reflex builds, nor recommened either. ONLY for closed box subs ! See here for eg ESP - The Linkwitz Transform Circuit & especially here, Note 9 - 12 dB/oct highpass equalization ("Linkwitz Transform", Biquad) Active Filters Siegfried Linkwitz invented it.

You can change the Variable Filter capacitors C1/C2/C3 to, C1/C2 replace both with 1uf & C3 to 0.82uf which will then allow a 25Hz - 275Hz variation. This gives you plently of variation for matching.

Flipping the phase 180 with the phase shifter circuit isn't the only thing you can do with it. You can alter the phase anywhere from zero - 360. Just 180 with the Flip switch. Up to 180 with the varying control. 180 - 360 by using both together.

Closed box speakers roll off @ 12db an octave. The Variable LP Filter to the sub also rolls off @ 12db an octave. Flipping and/or adjusting the sub phase with the phase shifter, should get you a good match between the sub & closed boxes.

If you just want to EQ then you don't need the LT circuit. Just the other ones, plus EQ.
 

emosms

Member
2010-09-28 1:19 am
Sounds reasonable.

Clipping from bass boost, +3db....
I will have +-15v psu for the opams

Ok, if put the subsonic first, can I rework it to be a L+R summer?
- Project 99 - Subsonic Filter
The input does not look quite well. I don't want to add buffer opamp (if possible)

Here is where I will get the L and R from, it has a buffer:
- http://s24.postimg.org/urx04gigl/tone_control.jpg

This is a stereo tone control (lo, hi). I recalculated the caps, so the bass operates at 50-60hz instead of 30hz.
Only the stereo cabinets signal will go through.
- pot 47k
- original caps 220nf (110nf in series for the calcs)
- caps for 50hz - 120n (60nf ins seres for the calcs)

Finally... I don't want to use eqalizer. My whole concept is to stuff ALL the electronics in the sub box, including 3 power amps pcb's and a system for remote control.

Could I build one channel (30hz) filter with fixed boost of 3db?
Then putting that module into the chain, buffer included. (instead of LT)
The bandpass filter won't boost all the frequencies 30hz to subsonic, just a knee around 30 hz.

P.S. about the LPF: I planned to stack two circuits in series: http://www.sound.westhost.com/project155.htm
to get 24db/oct filter instead of 12db/oct. Is that ok?

Regards
 
Last edited:

Zero D

Member
2009-08-06 11:11 am
@ emosms

Clipping from bass boost, +3db

If you mean in the circuits OpAmps ? then no.

Project 99 subsonic first = NO. Because ideally it requires a buffer.

Tone control circuit = ok if you want to.

Project 155 LPF = Yes you could put 2 in series, but you would ideally need a Quad potentiometer ! If you can get one, fine. Actually, replacing both C1/C2 with JUST one 0.82uf & C3 to 0.39uf will approximately give you a 26Hz - 285Hz variation. It depends on the tolerances of All the components in the filter. I had simmed a slightly different filter configuration, with the values i posted before !

Could I build one channel (30hz) filter with fixed boost of 3db?

I don't know, could you ? Or do you mean, you would like someone to design one for you ?

A single 30Hz fixed, or variable, high Q graphic EQ would do that. BUT, how do you actually know you need a 3db boost @ 30Hz ?
 
..................
P.S. about the LPF: I planned to stack two circuits in series: Project 155
to get 24db/oct filter instead of 12db/oct. Is that ok?

Regards
LPF, use an MFB which doubles up as a summing channel since it is inverting.
section 7
http://www.ti.com/lit/an/sloa049b/sloa049b.pdf
just add a second R1

also discussed in ESP
http://www.sound.westhost.com/articles/active-filters.htm#s4

You can cascade active filters.
The Qfinal = Q1 * Q2
and works with Butterworth, or Bessel, or Linkwitz Riley.
Gain (final) at roll off is Gain1 + Gain2 i.e two Butterworth B2 in cascade (equals LR4) give an F-6dB equal to the The basic B2 F-3dB frequency.
 
Last edited:

emosms

Member
2010-09-28 1:19 am
Thank you both !!
I really must understand how to design filters someday...

Can't really spot MFB in Eliot's low pass: Project 155
The quad pot won't be a problem, I will be using digipots and rotary encoders (arduino).

But the design of the filter still concerns me.
Actually the usable range will be 50hz-100hz
(I don't believe the cutting frequency would be above 80hz and bellow 60hz).

My digipots are 50k, and they are pretty close, (48-49k) to what specified.
The LPF circuit uses ..... 10k (I just noticed that...... :(, after buying more 50k digipots...)

I need to redesign the circuit to vary in one octave range (50-100hz)

If I use 10k out of 50k digipot, I will have only 12 steps of the pot.
Provided that a linear pot is needed (by design) it would be pretty ok.
Somewhere around 5hz change per step.

But.. I will really need to go through the design and recalc elements, to know what I am doing :(.
 
Last edited: