help appriciated designing correction filter for speakers

[erikb1971]

Hi all

The last months I have been busy designing and building a set of very small (2 liter cabinet) near field monitors.
It has an Audience A3s full range driver and a passive radiator on the back.
After along road I managed to get them from sounding plain awful to not too bad at all. Still, they miss a bit of low end. I have been playing around in Cubase and I came up with a relatively simple eq setting that makes them sound really sweet! Since I will also be building a pair of these for a friend of mine as regular pc speakers, it would be nice to sort of build in this eq. The eq settings are:

1. a 2nd order butterworth high shelving filter set at 117 Hz, -7.4dB with a q of 6.5
2. a notch filter at 16Hz with a q of 1.28

attached a picture of the eq settings I have used. 

Hope you can help me out or wake me up

Thank you anyway in advance
Erik
     

 

[moamps]

Hi,

The passive design of this filter is too much trouble, IMO. If you are going active, look into
Linkwitz transform here
http://www.linkwitzlab.com/filters.htm

Regards,
Milan

 

[wtmnmf]

This could be your solution:  http://www.minidsp.com/

 

[erikb1971]

Hi,

The passive design of this filter is too much trouble, IMO. If you are going active, look into
Linkwitz transform here
http://www.linkwitzlab.com/filters.htm

Regards,
Milan

Tnx! The transform filter is way out of my league...too bad, should have stayed in school

But I have seen something else interesting: the passive shelving low pass filter. Attached you see the eq settings I used to get my preferred sound. I have added to that the formula of the passive filter, and put in the values used in the filter (A, Fp & Fz). It looks to me that that could quite achieve what I want to achieve for as far as the high shelving... Or am I being stupid (again)?

 

[erikb1971]

This could be your solution:  http://www.minidsp.com/

Hey Tnx.. I have seen that one and I am keeping this option as a sort of last resort

 

[PRR]

You do NOT want to try for 30Hz response in a 2-liter box. It will slap itself to death before you reach listenable broadband level.

Very fine monitoring is done on systems that quit at 50Hz.

6db peak at 50Hz, 3db rise at 120Hz, similar to your first graph of suggested correction. Is a good phatt rise, but rejects subsonics.

 

[PRR]

.

 

[erikb1971]

You do NOT want to try for 30Hz response in a 2-liter box. It will slap itself to death before you reach listenable broadband level.

Very fine monitoring is done on systems that quit at 50Hz.

6db peak at 50Hz, 3db rise at 120Hz, similar to your first graph of suggested correction. Is a good phatt rise, but rejects subsonics.

I just thought I'd simplify the problem and and solve one problem at the time, that is why I left out part of the curve in my second picture. I have seen these speakers try to do 30hz,  and it is not a pretty sight (and still you do not hear it).  Your solution is both simple and seems to do exactly the trick! Slowly, I am starting to accept the fact that I might have to go to active filtering to get them right.
Thanks a lot for your input! 

 

[ricardo]

As a genuine loudspeaker guru in my previous life, I second PRR's comments and recommendations.

In fact, looking at the A3's specs, I'd be tempted to have the EQ peak at 70Hz and roll-off quickly below that.

In more than 10 yrs of blind listening tests, small speakers with 70Hz cutoff have often beat MUCH larger and more extended speakers with comments like, "more tuneful & detailed bass".

Of course you need to get the 70Hz alignment right but classic 4th order reflex as detailed by Aboriginal Elder Neville Thiele is excellent.

 

[erikb1971]

What do you mean with:

Of course you need to get the 70Hz alignment right

Cheers

Erik

 

[PRR]

There was no reason to use such low feedback impedance.
Change:
C3 = 0.01uFd
R3 = 120K
R4 = 100K

Now you can use nearly any dang opamp. TL072 is the obvious breadboard mule to find a bass response. If you can hear the high-end of a '072 being soft or harsh, then try sexy chips. (But nearly everything you listen to has been through several '072s before it comes to you.)

> tempted to have the EQ peak at 70Hz

My feeling also. Change C2 C1 in the ratio 0.03uFd * (50/70), about 0.02uFd.

C3 should change in a similar way. This is a cheap (and AFAIK novel) way to phatten the bump, rather than a fairly narrow boost just above the cut-off. Get the deepest notes right, then fiddle C3 +/-10% or 20% to get the next-up octave phatt enuff.

> get the 70Hz alignment right

This 2.5-pole filter may take very different shape if built with worst-case 20% caps and 10% resistors. This does not need 1% parts (4 and 6 pole filters sometimes do). 2% resistors are fine and now very cheap. 10% caps are not expensive, 5% can be had.

When a small woofer is driven hard, there can be a pretty hard breakover between boost and splatt. 90Hz may be too thin, 50Hz may be too slappy. When working VERY hard (as 2 liter boxes are wont to do) you may be down to a 65Hz/75Hz decision.

Get a set of 80K 90K 100K 110K 120K resistors. Preferably some in-betweens like 95K. Or buy a range of "5% values" from the 2% tolerance resistors. Put them in place of R1 and R2. Lower R is higher F.

 

[ricardo]

> tempted to have the EQ peak at 70Hz

> get the 70Hz alignment right

The correct textbook solution is to use what da Thiele-Small pseudo gurus call a B6 alignment.  The ported cabinet will be tuned to 70Hz with box size & magnet damping chosen to have response -9dB@70Hz.  The 2nd order electronic filter peaks 6dB@70Hz and the combination has a 6th order Butterworth response -3dB@70Hz.  You may not be able to do this with your Audience A3 unit.

However, with speaker design, the most important test is to listen to the end result rather than believe a beach bum masquerading as a guru.

As you have already conducted listening tests on your EQ, I suggest PRR's circuit gives you the boost where it is audible while protecting the speaker where it won't be heard but will distort.

Personally, I would cut his filter even further to peak at 70Hz while following your EQ boost above that.

The A3 is a very small unit and, despite what Audience claim, needs careful handling, especially in your very small box.

Do you know what is the resonant frequency of your Passive Radiator in that small box?  This affects response, distortion and what is sensible electronic EQ.

 

[erikb1971]

Thanks a lot both of you... looks like I will be making me a nice breadboard filter for the first time in my life!

The fs of the passive radiator according to the specs is 29hz.

This is the measured response of my combined unit (driver and pr) and my suggested correction to it.

Not that the dips at 9,5khz and 20khz were made purely on visual.. I do not hear the difference...  so far

 

[erikb1971]

So the basic of the filter should look like this:

now to add the power!

 

[erikb1971]

would this be a workable power supply schematic?

 

[ricardo]

The fs of the passive radiator according to the specs is 29hz.

That is the resonance of the Passive Radiator by itself in free air.  You need the resonance of the Radiator in your small box which will be much higher.

You can find this by carefully sweeping the complete speaker and noting the frequency where the passive radiator is moving most.  At that frequency, the main unit will also have a dip in its movement.  The accurate way to find this is to do an impedance curve of the complete speaker.

Can DEQX do this?

The importance of this tuning frequency is that you MUST NOT boost below it.

 

[PRR]

> So the basic of the filter should look like this:

Output returns to _ground_, not the to resistor divider.

Two or four 9V batteries is an ideal supply for breadboarding. You may not like it; in which case you have some 9/10th-good batts for other uses.

 

[erikb1971]

first time I am ordering without kit.. eekkss.. 0.6w or 0.25w resistors? 
And can I put 2x 0.01uF caps in series to make the 0.02?

 

[PRR]

> eekkss.. 0.6w or 0.25w resistors?

If you go the MAXimum for the opamp, there's 36V across 100K.

36V*36V/100,000 = 0.013 Watts. Double this for safety, 0.026 Watts.

Now you look in the catalog for the lowest-price good quality resistor larger than 0.026W. You will find they ALL are over 0.026W. These days, 1/4W is as cheap as any.

(In fact these resistors are unlikely to have more than 2V for 10% of the time. 0.00004W resistors would work. MOST line-level audio resistors may be selected for comfortable size and popular price, the actual dissipation is much less than the smallest resistor you can buy or handle.)

> can I put 2x 0.01uF caps in series to make the 0.02?

No. Two 0.01u caps in series gives 0.005u. Try parallel.

 

[erikb1971]

Tnx for keeping up with the beginners questions!