Do I Need A Notch Filter?

[noodle_snacks]

Hi,

My peerless HDS 850439/Seas 27TDFC MTMs are comming along quite nicely, however there is something I don't like about the sound. playing around with a tone generator revealed that sound at about 3-4khz sounded a bit yucky, not sure how to describe what i was hearing.

At first i tried fiddling with the tweeters response etc thinking that maybe the tweeter just needed padding. However then i noticed that the HDS woofer measured with a peak right where i was picking up trouble with the sound.

Do you think maybe a notch filter on the woofer (at 4200hz where the peak is centered) would fix the problem?

 

[DSP_Geek]

It sounds like a good place to start. I had much the same problem in Pro Audio: the folded bass horns I used had a ginourmous peak just above the crossover frequency. EQ on the whole system helped the problem but was far from curing it; about the only things that did work was either shifting the crossover down another half-octave or dedicating a side EQ post-crossover to clean up the bottom end.

The problem with peaks of the sort is not so much the gain anomaly, but the time-domain response being smeared by the peak, which is only partially compensated by the time-doamin response of pre-xover EQ.


Good luck,
Francois.

 

[noodle_snacks]

Actually helped quite alot, the woofer follows the target 4th order rolloff perfectly now and as a whole the speakers sound much better. As a whole tweeter just needs a bit of padding now i think.

 

[DSP_Geek]

I'm not so sure the tweeter needs padding; its response above 5 kHz seems fairly even with the woofer below 1 k. It does look as if the frequency response is down 1 or 2 dB between 1.5 k and 5 k, which will take away some of the definition in the vocal range and also make the treble instruments seem a bit edgy in comparison. I'd suggest tweaking the crossover to fill that spot in.


Cheers,
Francois.

 

[noodle_snacks]

It does seem to me that the tweeters response would be close to the woofers if it was padded down maybe half a db or so.

Attached in the predicted response, removing the woofer peak actually increased the output in that 1.5-5k range you mentioned (meaning the peak was probably out of phase with the tweeter). There is definately a subjective improvement in vocals and treble instruments do sound less "harsh" than before. What components would you add or modify to optionally tweak the response in this range?

I am curious as to what might be causing the peak in the tweeters response at 2khz with a dip either side (at about 2.5khz and 1.5khz) could it be the effects of the square edged baffle (210mm wide and a centered tweeter)? I seem to recall reading somewhere that strategically placed foam blocks, or compressed wool felt either side of the tweeter could help smooth tweeter response, is this worth trying?

 

[Paul W]

That does look like diffraction so I think you are on the right track. I recently did a 27TDFC, but with asymetric placement and roundovers modeled using Paul Verdon's BDS simulator. (Google for PVConsultants). That one was extremely smooth.

Felt might help but, especially on small baffles, roundovers and asymetric placement are more predictable.

An externally hosted image should be here but it was not working when we last tested it.

 

[DSP_Geek]

It does seem to me that the tweeters response would be close to the woofers if it was padded down maybe half a db or so.

Attached in the predicted response, removing the woofer peak actually increased the output in that 1.5-5k range you mentioned (meaning the peak was probably out of phase with the tweeter).
[...]

I am curious as to what might be causing the peak in the tweeters response at 2khz with a dip either side (at about 2.5khz and 1.5khz)
[...]

Ah, ok. I thought you meant padding it down 1 or 2 dB, which would've been way too much. I'm a bit surprised notching the woofer had such an effect at 3k, but you're probably right about this being out of phase. As for the peaks, Paul W is most likely right: they look like diffraction to me also, and I second his suggestions.


Francois.

 

[noodle_snacks]

In future projects i will definately be offseting the tweeters. However too much work has gone into the construction and veneering of my current boxes. I will play around with some foam and compressed wool felt. and post my results as soon as it stops raining and i can get outside to get some accurate measurements done.

 

[DSP_Geek]

I've gotten good results from cutting notches into the foam so it looks like triangles pointing at the tweeter. That way you avoid setting up diffraction from the foam itself.


Cheers,
Francois.

 

[DSP_Geek]

By the way, Paul, how did you get that gorgeous finish on your cabinet? I'm green with envy.

 

[Paul W]

The finish is really easy. Rustoleum Hammered Silver applied over Zinsser BIN primer using a thin foam roller. The only trick is that you need to work fast because the paint sets up very quickly...2nd coat in an hour. Extremely good air circulation and/or an effective respirator.

 

[DSP_Geek]

Thanks much. I'll give it a shot. How long does the primer need to dry?

 

[noodle_snacks]

Just a final note on the crossover, the cap i had in parallel with the tweeter to lower the response of the tweeter a bit was doing strange things to response in the 3-5khz range, i will experiment tomorrow, now that the cap is gone the woofer notch filter may not be needed. I am fiddling with various tweeter attenuation methods now.

 

[Paul W]

I don't remember primer dry time, but it is also fast...primer plus two final coats in one day. Just follow the times/directions on the cans and be careful with the fumes.