H91 to H92 Horn Swap - The Results

[toddalin]

The Results

Linda went off to meditation so I fired up some pink noise to see what had transpired. The speaker and microphone were in the same positions as in the prior test (~1/4”). Unforetunately, because the wires for the crossover were run out the port for further testing, this latter test does not have the grill cloth, whereas the prior test did. (Hindsight. )

Again, I dialed the horn and tweeter to about the smoothest response. In this case I also did the horn both in and out of phase. I also did this with the tweeter, but that made little difference and I didn’t photograph those changes. I did note that when the horn was hooked up in phase, I had to turn the tweeter up a hair to get it to balance out as opposed to when the horn was out of phase. I think this could infer that the horn is a little louder when connected in phase with the woofer requiring the tweeter to be elevated to match it.

This first picture shows the horn in phase with the woofer (i.e., +s on reds). Note the 800 Hz point and lower is depressed, but the area to the right is not to bad.



This second shot shows the horn with reversed phase. In this case the 800 Hz band is stronger, but the area to the right is depressed. To me, the first shot (both +s on red) looks to be the preferred phasing based on the surrounding areas of the curve.

(In the past, had I not been using the Behringer, I would have played an 800 Hz tone though the speaker and probably concluded that the preferred phasing was not. )



For contrast, this shot was the before also using the preferred phasing. The H91 actually looks to be smoother between 1K and 3K.

[Beowulf57]

Okay...now to some listening. How does the longer horn sound with music?

[Zilch]

I always use 30 dB range; the fundamental response is more apparent with the additional smoothing.

H91 shows its characteristic midrange emphasis; H92 is flatter there....

[toddalin]

I always use 30 dB range; the fundamental response is more apparent with the additional smoothing.

H91 shows its characteristic midrange emphasis; H92 is flatter there....

Agreed, but the 15 dB range shows the resolution of the areas of interest.

[toddalin]

Okay...now to some listening. How does the longer horn sound with music?

I've listened to Brothers in Arms (original) and it sounds good. (Only one channel is done so far.) It didn't sound bad before. It obviously hasn't filled the hole. I may be able to take the crossover a little lower now with the longer horns and will try adding a little capacitance on the horns.

[Ian Mackenzie]

I am not sure what crossover use are using but the equivalent L300 would be the best bet to sort out the hole.

Getting the phase shift right at that frequency is critical to obtaining a smooth response. Save your time and use the factory engineered LC values.

[4313B]

Greg concurs that his 630 Hz problem is most likely a box problem but can't remember if the L200 was a problem box or not, that was just too long and a thousand models ago.

The 2235H does not have a hole in its response at 630 Hz nor does the proper alnico structure reconed as a 2235. But we already knew that didn't we. The posted FR curve of the 2235H in the EDS stands. The alnico structures will develop a broadband 2 to 3 dB sag but that's just life, double up on the amplifier power.

[Beowulf57]

I've listened to Brothers in Arms (original) and it sounds good. (Only one channel is done so far.) It didn't sound bad before. It obviously hasn't filled the hole. I may be able to take the crossover a little lower now with the longer horns and will try adding a little capacitance on the horns.

Where is your crossover point at present? Also, when you test the results using material that exposes the hole, try reversing the phase of the horns and see which way sounds better to your ears: when I switched from the H91's to the H92's, I found the sound better with the phase reversed (my test was simply listening to dialogue on satellite programming and it was more intelligible in phase for the H91's and out of phase for the H92's).

Also, attached for anyone interested is an .xls file in zip which provides a calculator for room frequency peaks and nulls: just plug in your room dimensions and seating position dimensions and check row 27 for the problem frequencies. Lots of other data as well.

[grumpy]

yeah... not that the EDS needs my concurrence, but I've posted plots from a 4430
system that match them pretty well... even with my level of experience/methods.

Semi-random thoughts:
Would it be a worthwhile test to stuff/block the ports and ignore the response change
down low? (if there's an internal box resonance that's escaping through the ports...
understanding that it could also come through the cone). Certainly simple enough.
Apologies if this has been done already.

Cabinet internal width is pretty close to one wavelength at 630Hz... centered in the
baffle, that would put it at a null. A test septum (bit of a pain) might blow or support
this thought (as might someone with more experience in this area).

Has anyone else w/L200 cabs measured this type of response?

(Beowulf.. good ideas, but I believe the crossover is significantly higher... hence the problem)

[toddalin]

I am not sure what crossover use are using but the equivalent L300 would be the best bet to sort out the hole.

Getting the phase shift right at that frequency is critical to obtaining a smooth response. Save your time and use the factory engineered LC values.

I don't think that my values are that different than the factory values. The 33 mfd and 7 ohm on the Zobal network were subsequently changed to a 20 mfd (like the L300) and 15 ohm to bring the woofer up a little in that area of the frequency band, and the 1 mfd on the tweeter was removed from the tweeter.
I belive that Heather is using something similar, though not identical.



And here is the l300 network and it is very similar:
http://www.jblproservice.com/pdf/Net...%20Network.pdf

I don't believe this to be a phasing issue of the crossover as I noted that I get this same response just connecting the woofer directly to the amp with no crossover.

[Ian Mackenzie]

Try the woofer on an a large baffle

[Beowulf57]

Cabinet internal width is pretty close to one wavelength at 630Hz... centered in the
baffle, that would put it at a null. A test septum (bit of a pain) might blow or support
this thought (as might someone with more experience in this area).

Has anyone else w/L200 cabs measured this type of response?

(Beowulf.. good ideas, but I believe the crossover is significantly higher... hence the problem)

Ahh, yes, my crossover is at 500Hz. Still, calculating the room modes would show whether there is an exacerbation of the dip due to a null at the seating position.

[Zilch]

Do I recall that Grumpy brought over a "Real" 2235H and the problem remained?

I investigated this problem in another thread here. It went away when I pulled back and put the mic on the floor.

Does tipping the box so the baffle is vertical alter the situation?

[Ian Mackenzie]

The problem is these measurements are not gated so what made be heard and measured at the listening position with the whole reverberant field may well be different to this static measurement.

[Beowulf57]

The problem is these measurements are not gated so what made be heard and measured at the listening position with the whole reverberant field may well be different to this static measurement.