[buzztiger]

Hi ! I have built my subwoofer from an old SVS driver that i upgraded. I use a DIY 400 Watt amp to power it. However i hear noises when i feed it with frequency below 16Hz. I think its bottoming out. I need a subsonic filter to protect my driver. I came across a subsonic filter here http://sound.westhost.com/project99.htm . This is not suitable. I was then looking at Linkwitz Riley high pass filter. But it has confused me. The active filter posted at this site http://www.subwoofer-builder.com/active-filter.htm differs from http://www.linkwitzlab.com/filters.htm#3 so i don't know which one to build.


Pls help. Has anyone else made a simple subsonic filter ?

 

[collo]

Hi there Buzztiger,

the active filter on my site is based on the circuits at the Linkwitz site.

I just added together all the elements listed on his site to get the full circuit.

The design uses the hipass section of linkwitz's 24 dB/octave crossover. (section3 on the linkwitz page)
To the front of this is added the buffer stage (section1 on the linkwitz page)
At the end, the line driver stage is added (which is an amalgum of sections 10 and 11 on the linkwitz page)

Adding these elements together gives a filter that doesn't load down the line that is driving it, and is happy to drive whatever load you place on it. In addition, the +/- 6dB adjustment might just save you needing a cleanbox.

All the explanations on the linkwitz site apply to this circuit as well.

regards
Collo

 

[buzztiger]

Wow answered by Collo himself ! Thank you so much for your reply.

I don't think i need the +-6dB gain on the line driver because i thought of having a potentiometer to control the volume of the subwoofer after this subsonic filter . Can i just substitute the line driver with the circuit shown in section 11 of Linkwitz website ? There's 2 circuits there A and B, should i be using B ? Thanks.

 

[collo]

That should be OK.

I would go for option B, since that is closest to what I built, so I know it works.

Your circuit would be.....

 

[buzztiger]

Thanks. Just bought all the parts for the filter. I wanted to use a regulated power supply for this circuit but I only have 9-0-9 auxiliary out from my transformer in the amp. So i only have +12V and -12V dc. This means that i can't use any of the voltage regulator chips. Would using unregulated power for this circuit affect the sound quality ?

 

[collo]

I think 12v would be OK. You could use 12 volt regulators if you wanted.
Some good reference info in this power supply article at Elliot Sound Products

 

[buzztiger]

The problem is that the voltage regulators require that the voltage input to them be a min of 14V before it regulates the voltage to 12V.

 

[collo]

hmm, I see..

You could always try it unregulated.
If it's noisy, you could add an external plugpack supply

 

[buzztiger]

I think i will just do that.

 

[collo]

Sounds good. Let us know how it turns out....

 

[SturmMD]

The opamps have good power supply noise rejection so you may be able to get away without regulation. Otherwise just use 9V and -9V regulators if you have +/-12V for example. The circuit can be modified for single supply operation but then it has a very large turn on thump you have to be careful about without any circuits to mitigate it.

Also you don't need a linkwitz riley filter type for a SSF. LR is used for crossovers between two speakers so that they add up to a flat response. It will eat away too many dB's before the crossover point. You should just use a 2nd order butterworth HPF. This means you'll only need 1 of the two inner filter sections. Recalculate the R and C values using this link. Set it to 2 poles
http://www.daycounter.com/Filters/Sallen-Key-HP-Calculator.phtml
You can choose the frequency by adding a 2nd order butterworth filter in WinISD and changing the value until the cone excursion graph limits to the lower peak to about the same as the upper peak.

 

[collo]

Interesting point about the order of the filters.

I had assumed that a steeper filter would let you hang on to the SPL for longer before you had to cut the power to control excursion.

Here is a WinISD comparison between the two styles of filter mentioned above..

The frequencies were adjusted until the excursion was limited to Xmax+10% for each filter.

As expected, there is an advantage to the 4th-order filter in SPL, but the difference is so small that you wouldn't worry about it.

The main difference is in the group delay (as theory predicts).

With the simpler design and lower group delay, the 2nd-order filter certainly looks to be a better choice. Looks like I'll have to change my recommendations..

Further comments ???

 

[Mike P.]

The 4th order group delay exceeds the second order group delay at about 17 hz. The question is, is the 13 m/s difference noticeable at frequencies that are 20 hz or less?

 

[SturmMD]

If you're comfortable with the 4th order LR's group delay in the region below 20Hz you might consider a 4th order butterworth. This will gain you a solid 1dB over the LR over the entire octave above the cutoff frequency. The differences between the filters are most obvious in the Filter/Eq Transfer function magnitude plots.

Also consider the HPF characteristics of your component in the signal chain. For example I have measured the following and include them in WinISD to work on a SSF design:
EP2500: 1st order HPF at 4.1Hz
BFD: 1st order HPF at 3.7Hz
my Receiver's sub output: 1st order HPF at 2.5Hz

If you just add these you will see that a slightly lower SSF frequency is necessary. The higher order butterworths are preferred because they preserve a precious 1dB over a large range above the cutoff.

Also, the circuit as shown with a 1uF and 22K input has a 1st order HPF at 7.2Hz. You may want to lower this by increasing the C or changing R to 47K. Add this to your WinISD

 

[collo]

Looks like some more reading on the different implimentations of sallen-key filters would be rewarding. If I can just find my old copy of the filter design cookbook...

Thanks for pointing out the HPF effect of the RC network on the input buffer. Something I missed on the Linkwitz site!

Those rolloffs for the EP2500 and BFD are worth knowing. How reliable are those figures? How did you measure your AV amp?

 

[SturmMD]

Those rolloffs for the EP2500 and BFD are worth knowing. How reliable are those figures? How did you measure your AV amp?

They will vary primarily because of component tolerances. I would guess that they would be within 10% of my measurements because of combined tolerances of electrical components.

I used REW and a sweeps up to 200Hz to find those cutoffs. I saved that data to text and opened it up in excel. I then used the solver to fit the filter cutoff frequency and order to the data. It works very well. You just have to decide what part of the data is important for the fitting. I limited the fitting range from about the -6dB point up to about 2-3 octaves above the cutoff frequency. This provides the greatest accuracy in the range that you are interested in: 10-20Hz. Even though they may only contribute less than 1dB in this range, there are several so they add up and should be considered.

For the EP2500, I just was careful with the gain setting before looping it back to my sound card. You must use unbridged mode and also the ground (black) output from the amp must go the ground of your sound card. Maybe try it out with a 10k resistor in series first. If the polarity is wrong the lights on the EP2500 will light up. For the receiver I set the mains to small and the crossover at 80Hz then just hooked REW to stereo inputs and measured the sub line output. For fitting this data you need to include the LPF in the curve fitting.

 

[collo]

Thanks for that - I can see how all these could add up to a significant rolloff

 

[buzztiger]

I just cooked 2 of my LM741 opamps. Connected the +12 to the to the input by mistake. Just went back to the store to buy more LM741 and they don't have stock of it till next week. Thinking of substituting with tl072. This should be ok right ?

Wow so many posts while i was busy soldering. Makes it seem like i should stop building the LR filter. Maybe move onto the butterworth filter. SturmMD thanks for the link to the calculator. The input buffer and output line driver can still remain the same as in the circuit for LR filter right ? Btw how can i find out whats the 3dB frequency ? I only know that the cutoff frequency should be 16Hz.

 

[SturmMD]

The TL072 will work. The 22K resistors in the input buffer section can be lowered to like 10K so that the highpass from at the input is 3.3Hz instead of 7.2Hz.

The 3dB frequency is the cutoff frequency for the calculator.

To simplify the assembly you can even do without the separate line driver. Just put the 200 ohm output resistor on the output of the butterworth filter section. This means that only 1 dual opamp tl072 is necessary for a 2nd order filter (including input buffer).

 

[collo]

.. I only know that the cutoff frequency should be 16Hz.

Changing from a 4th-order filter to a 2nd-order will mean that you will need to model again in WinISD to find the new design frequency.

SturmMD, does one stage of the Linkwitz-Riley filter constitute a Butterworth filter (ie ratio of resistors is 2:1 Q=0.5) or is it different?
Nice tip about getting by on one op-amp if gain is not required!