Powered Zaphaudio style HiVi B3N speaker using a Chipamp w/ active filters

[DaMeat]

I have looked at Zaph's little B3S speaker, and have been thinking about incorporating his design into a powered speaker using some LM3886 chips I have laying around. I would like to use the B3N (round flange) instead of the B3S (square flange) and I accept the small differences this will introduce...

This is what I would like to do:

Minimum amount of components all together (Amps, filters, and speakers), and I would like all the filters before the amp if possible Also, if possible, I would like to stay away from inductors and only use RC filters (Cheaper and more accurate?). I have an SMT station so I don't mind doing this all in SMT and laying out the circuit board myself (Hopefully everything will be on one 2 layer circuit board including the power supply.

Speaking of power supply, I might incorporate a switching power supply to power the Chipamp, I am doing this to lower the total power consumption and not have the need for that big 22V transformer that everyone uses... Can someone tell me the drawbacks of trying this?

For the speakers:
From Zaph's design I believe he has an 8kHz notch filter, and a BSC (Baffle Step Compensation) filter.

For the 8kHz, can I simply use a RC Notch filter (Twin T) like shown on this page (All the way at the bottom):
http://ourworld.compuserve.com/homepages/Bill_Bowden/page9.htm

Or is this filter too aggressive?

For the BSC I am planning on something similar to the variable equalizer type of filter from this page:

http://sound.westhost.com/bafflestep.htm

But I will use 2 discrete resistors instead of the potentiometer.
Also, could someone tell me if 906Hz is the freq. Zaph used for his BSC? (115/.127)

I will also incorporate a 200Hz high pass filter as Zaph suggests, and build a sub using a Polk MM120 I also have lying around... (I will build a better sub for this system after the speakers are done and I am happy with them)

Is there a better way to incorporate all of these filters into less components (If I need to use an op amp before the Chipamp that is also a possibility, I have seen some good fast ones recommended on this site.)

I plan to build 4 of these for my computer, and hopefully they will be loud enough to fill my medium sized living room with some good tunes...

Any comments, suggestions, feedback is welcome... I gave myself a very small budget, but a lot of time to complete this (Around 4-5 weeks I am hoping... so multiply that by 3 and I think I should be done by August )

 

[gmikol]

I've been looking at doing the same thing as you. It kind of surprises me that Zaph hasn't published an applicaiton note covering this...

I'm using the Active Filter Boards from here.

They do a lot more than just notches and BSC, but at $5 a piece, it's hard to beat. If you're interested in these, you might need to act fast. I think Bob is gonna close the GB soon, if he hasn't already.

You're right that you don't need any inductors if you're going active, but I don't see why you need an adjustable baffle-step compensation. That's something that is primarily defined by the shape of the baffe.

Also, I think a Twin-T might be too aggressive and too narrow for the 8k breakup node. The more closely matched the components are, the deeper the rejection will be. And I don't think you wan't to count on component mismatches to get the response. You might want to consider some other topology to deal with the 8k peak.

Using an SMPS is a great way to save weight and efficiency, especially in this relatively low-power range, but then why use a chip (LM3886) that is ony 60% efficient at full power. What about some sort of Class-D amplifier? Tripath-based kits are probably the least expensive. UCD180's are pricey and overkill for B3S/B3N.

As for loud enough...without any passive devices robbing power, these are about 79dB @ 1W/1m. The LM3886 can do 50W into 8Ohm (We'll call it 32 to make the math easier.) 79 + 3*5 = 95 dB @ 1m per speaker. 107dB @ 1m with all 4 going full-power. Loud enough? Can that Polk sub keep up?

At 50W, you could go with a 150Hz LR2 cross to the sub to help with localization issues and still be OK on excursion.

There you go...you asked for opinions. I hope it didn't seem to harsh. I'm an engineer, and they don't teach how to be nice in engineering school, only

Good luck, please post your progress, and if this thread is still alive when I start mine, I'll post my progress, too.

--Greg

 

[Tubamark]

I've been looking at doing the same thing as you. It kind of surprises me that Zaph hasn't published an applicaiton note covering this...

. . . Also, I think a Twin-T might be too aggressive and too narrow for the 8k breakup node. The more closely matched the components are, the deeper the rejection will be. And I don't think you wan't to count on component mismatches to get the response. You might want to consider some other topology to deal with the 8k peak.

'Sorry to ressurect very old thread, but this addresses what I'm looking to do w/my B3N's. I'd like to impliment at line level between pre and amp -a patchcord with an Altoids tin in the middle of it.

Is it a reasonable approach to introduce intentional mismatches in order to reduce the notch? Either by measuring components and selecting worst matches, or by adding little caps and resistors as needed to skew values.

Assuming the answer is "yes", how far off do the components need to be (I'm thinking a percentage) for the desired effect?
Likewise, how far off is too far?

Alternately, if there is a simpler or just better approach, I'm all ears

Thanks in advance for any replies!

--- Mark