Issues with OT12 frequency response

[mlkras]

I started off building Bill’s horn sub speakers, T24 and T39 and was mightily impressed. I have gone on to build a number of OT12 speakers and I have not been nearly as impressed with these.

What initially attracted me to the design was the low build cost, efficient design, the capability of playing very loud and nice flat response curve. The speakers were easy to build and pretty low cost but just have never sounded right to my ears. Although they do play extremely loud, even at low volumes voice, male or female, through the speakers sounded pretty unnatural, irregular and combed. This to me is the acid test of speaker performance and extremely important to me as I regularly work with vocal groups, choruses and small jazz ensembles. I think they sound fine for loud rock and roll and would be killer in a club setting or with electric instruments only.

These speakers were built exactly to plans with 3012lf, throat fillers and Eminence 2005 drivers on the Goldwood horns.

These results were confirmed when I tested the speakers with SMAART in a large open area, about 80 feet from any vertical reflecting surface. I found the same response unevenness through the midrange with peaks and troughs of over 15 db. The also exhibited an almost 20 db dip at around 1150. The issue then seemed to be with the basic cabinet. I disconnected the HF section and the box response was the same.

Working with other experienced sound types who have been watching my building we tried to equalize out the dips and peaks but it was immediately clear that these were not equalizable frequencies. They absorbed a lot of boost or cut without much impact. Maybe I just don’t know how to equalize these frequencies out. We also tested a number of commercially made speakers and a number of them were not so hot either.

I then scoured the forum for any insight into the issue, particulary the post in the EQ section ‘My various SPL charting sessions’. The poster, WB, shared a SMAART graph that was identical to what we found. In the thread I found this post by Bill:

“I've confirmed that dips in the 800-1200 Hz range are products of horn loading. They can be addressed with some changes in the horn, but at the cost of overall sensitivity, so it's not worth it.”

Engineering is always the balancing of many design goals. I do not think that sensitivity is the best primary goal for speaker design. I would gladly give up a bunch of sensitivity for a more even frequency response pattern.

So has anyone seen these frequency aberrations? Been able to do anything about them? Bill?

[David Carter]

I am not nearly as knowledgeable about sound reinforcement as you or many of the folks here, so I can't really contribute much to this discussion, but I do want to double-check one thing you said in your post. You said they were built with the 3012LF? The 3012LF is a sub driver. The 3012HO is the one that is intended for tops like the OT12's. This was discussed recently when someone else either had built or was about to build OT12's with the 3012LF.

[mlkras]

Good catch Dave. This is what happens when you buy T39 speakers at the same time as OT12 speakers. I will and get a 3012HO and see if there is a difference. I will bet that there will be some difference. The specs for the two speakers are not the same. I do not know how sensitive this design is to varying speaker characteristics.

I only read the plans a hundred times!

[David Carter]

I sure hope that fixes your issues. If it does, I don't think you'll have too hard of a time selling the 3012LF's if you don't need them. I just built a pair of T39's with them, and they're awesome!

[WB]

The also exhibited an almost 20 db dip at around 1150.

This dip can't be removed with EQ. (It's also very narrow)


I found my unequalized OT12 to be efficient but rough in response. I used the Auto EQ feature on the DEQ2496 to even out the response. It made a huge difference in sound quality.

Swapping out the LF driver and replacing it with the HO is the proper thing to do, but the response issues will remain.

[mlkras]

That is what I suspected. Assuming that your OT12 had the correct driver (am I embarassed) the curve that you published is almost identical to the frequency response curve that we got. You would expect a coherence blip with a trough that deep.

If there are simple changes to the horn, or even not so simple changes that can smooth out the horn then I would be very interested.

What I am always trying to get is a louder but transparent amplification of the performance so the PA should disappear in the process, even at semi war volume. Everything has to be so smooth for the ladies who sing jazz.

[Ron K]

Here's an O-Top12 after EQ. with the Delta Pro 12A driver with the 6-Piezo straight array. Note the coherency issues where you are seeing your problems. However it shows to be not completely unrepairable for me.



I also replaced the Piezos with Selenium 220Ti drivers and Dayton Elliptical horns for the bottom cabs. I used the Goldwoods for the tops in the stacks,soon to be replaced with the DIY horn when I get some time. I moved to a bi-amp and removed the LPF on the bass driver.I also built the necked version with the throat fillers.

I was really able to smooth out the box with the Selenium configuration.

I dont know if all that makes all the difference but I really cant complain about the sound of the boxes at all. They get loud and can be very detailed even at distance.They do require a good bit of EQ.

Here's the DeltaPro12A without EQ in the O-top12 single config. No horn or piezo or filter.



Post up your SMAART Screenys so we can have a look-see!

[Bill Fitzmaurice]

The source of the dip is complicated, suffice it to say that minimizing it will result in about a 3dB loss of sensitivity in the 250-800 Hz range, which isn't worth it. If this was a hi-fi speaker that wasn't intended to be used with DSP it would be a problem, but it isn't. Of course, using the right driver will make a big difference too. BTW, high resolution plots of all midrange and HF horns will show similar results, it's in their nature. The best way to address EQing is to RTA using the same ISO points as those of your EQ. If your RTA is done by your EQ then that's automatic.

[mlkras]

I would gladly give up 3 db of efficiency for a flatter response. Is there any straight forward modification of the box that would help in this regard?

I would also be happy to purchase an alternate version of the plans. Anyting left over from your development work? The whole write up would not be necessary, just the physical plans of a prototype that was tested and rejected.

I've got bunches of switching amps waiting to get warm.

This forum has been very helpful on my question, thanks to all who have participated.

[gdougherty]

I'm looking at possibly starting some DR280's in the future. Is there a similar dip in the 280 horn, or does the curve and length help take care of that?

[gdougherty]

I would gladly give up 3 db of efficiency for a flatter response. Is there any straight forward modification of the box that would help in this regard?

I would also be happy to purchase an alternate version of the plans. Anyting left over from your development work? The whole write up would not be necessary, just the physical plans of a prototype that was tested and rejected.

I've got bunches of switching amps waiting to get warm.

This forum has been very helpful on my question, thanks to all who have participated.

250-800 takes more power than 1.1K. I'd rather save 3db of power requirements in the lower range and boost an EQ to smooth out 1.1K to a smaller dip.

[Bill Fitzmaurice]

250-800 takes more power than 1.1K.

A lot more. With every octave you go down power requirements double, and that doesn't take equal loudness into consideration, which requires even more power for lower frequencies to be heard at equal levels.

[WB]

I'm looking at possibly starting some DR280's in the future. Is there a similar dip in the 280 horn, or does the curve and length help take care of that?

I think the latest DR's with the throat fillers "should" have the dip. I didn't find it on the DR250 I just built though.

250-800 takes more power than 1.1K. I'd rather save 3db of power requirements in the lower range and boost an EQ to smooth out 1.1K to a smaller dip.

The 1.1K dip can't be EQ'd out, otherwise there'd be not issue. I think it's a cancellation at that frequency because of the throat filler. The wave front from the center of the driver is several inches closer to the horn than the outside rim of the speaker. At that frequency there's a cancellation where they meet. Bill would know better than I. I'm sure he could explain the real reason better as well.

[Bill Fitzmaurice]

I'm looking at possibly starting some DR280's in the future. Is there a similar dip in the 280 horn, or does the curve and length help take care of that?

I think the latest DR's with the throat fillers "should" have the dip. I didn't find it on the DR250 I just built though.

250-800 takes more power than 1.1K. I'd rather save 3db of power requirements in the lower range and boost an EQ to smooth out 1.1K to a smaller dip.

The 1.1K dip can't be EQ'd out, otherwise there'd be not issue. I think it's a cancellation at that frequency because of the throat filler. The wave front from the center of the driver is several inches closer to the horn than the outside rim of the speaker. At that frequency there's a cancellation where they meet. Bill would know better than I. I'm sure he could explain the real reason better as well.

It is a phase related dip, but has nothing to do with the filler, which improves response. All horns have phase dips, related to the differentials in pathway distances. The DR geometry minimizes those differentials, so the dips aren't as deep. But while narrow bandwidth dips look bad on a chart they aren't particularly audible. The OPs problem isn't that dip, it's the 5dB or so of broadband response droop above 250 Hz caused by using the wrong driver.

[WB]

It is a phase related dip, but has nothing to do with the filler, which improves response. All horns have phase dips, related to the differentials in pathway distances. The DR geometry minimizes those differentials, so the dips aren't as deep. But while narrow bandwidth dips look bad on a chart they aren't particularly audible. The OPs problem isn't that dip, it's the 5dB or so of broadband response droop above 250 Hz caused by using the wrong driver.

If you bi-amp the 12s and tweets, can you use your DSP to tweak delay a tick and minimize the phase dip?

I doubt it. The dip seems to be 1150 Hz for 12" (OT12) and 1250 Hz for 10" (Wedge10). I suppose if the high frequency was able to be crossed over below those frequencies problem solved.

The DR geometry minimizes those differentials, so the dips aren't as deep.

I can't find that deep dip at all with my DR250. Just a mild dip at 1400 Hz.