Bringing the first towers I bought up to my current expectations...

[93PGT]

SEE MY 04-APR-09 POST FOR UPDATED INFO
--------------------------------------------------------------
I've been building subs, and measuring TS parameters of speakers for a few years now, and was just recently critically listening to music on the MTM ported (39Hz) tower speakers i have hooked up to my mediaPC. They are well built, mitred MDF, 1/2" mostly, but 1" baffles, fully window-braced and lined with egg-crate foam.

They've always had an annoying upper midrange that i could never smooth with my parametric EQ. But otherwise I've always been pretty happy with these speakers, considering the price I paid. They do sound better (imaging, dispersion) with the lower of the MTM midbass unplugged (not so much the upper... directional IM or lobing maybe?), but this sacrifices balance in the lower mid and bass.

The tweeter is crossed over from best i can tell using 2nd ord LR, at about 3200Hz, (rated 8 ohms, with 3.3uFd mylar cap and from my measurements ~0.81mH 21AWG air core). The woofer is only 1st order, and seems to be using a 1.9mH (again from my impedance @10kHz measurement) FE-core only. The tweeter has a 3.4dB L-Pad, and a thermal varistor for protection.

I can submit impedance curves for trends if it would help, however, I'm not certain the magnitude on them is correct. However, i can say for sure that the 6.5" midbass are Fs=48Hz, and the 1" silk dome is Fs=950Hz.

Even suggestions of a complete redesign of the crossover is fine with me, although I think most of what I already have here can be worked with... let me know your thoughts please!!

[93PGT]

Oops... Series/parallel

I just realised with the midbass in series i completely jacked my crossover point higher... here's a graph of just the midbasses, nearfield, in box, with crossover, and run in normal parallel. (it's a little noisy, but i had to run the test very quietly!!) Peakiness still very visible from 4-6kHz.

[kevinp.]

try adding a parallel notch filter to the circuit, centered at 4500 and notching about 10-12 db, after the crossover and before the driver. There are online calcs to help with the values.

I don't design crossovers, but that's my novice suggestion...

[BobEllis]

...i could only do nearfield as to not disturb those sleeping ...

If all measurements were done nearfield, then you need to measure at 1 meter or so to have useful information above a few hundred Hz. I cannot imagine a tweeter with far field response like that would be listenable.

FRD an ZMA files would be useful as would enclosure dimensions for those that would like to simulate a new crossover for you.

[93PGT]

Speakers are 7.625"W x 41"H x 14.5"D, the flat section of the baffle is 6.5" wide, the internal volume is 1.82 cu ft (accounting for ports, drivers and crossover), there are two 2"ID ports,flared at one end, each 6" long. The drivers are mounted 5.5" on-center due to truncated 4"dia tweeter plate.

My in-room response is a little jagged, but I'd say this is due to the mix of absorbing and reflecting surfaces in a smaller space (this is in a 'study'-type room... desk shelves, carpeting and hardwood, and soft seating area) and induced EMF. But regardless attached in the response curve. I did move the speaker away from any hard obstacle as much as possible, and had it facing soft surfaces for the measurements.

The area I'm looking to smooth out is the 4.5-6K region. From my measurements of other known speakers using this set-up, the roll-off after 10k appears to be my computer's mic-in response. I've achieved almost the same plot from two completely different microphones in identical positions, and even some planar tweeters i measured rated at 5-32kHz +/-3dB rolled off almost as much.

I'll try to create and post a FRD/ZMA fileset soon from the data I've measured so far. Attached graph done at 1m.

[93PGT]

Alright, after sorting out some issues with my microphone input and testing some known samples, I have the FRD and ZMA files for the drivers. Also attached are the current crossover, and my proposed changes.

My goal is to simply add to what is there (ie adding 5uF to the 3.3uF for the tweeter and 10uF across the midbass) to improve the sound without disassembling the solidly built crossover board. My changes appear to smooth the phase and frequency curves by avoiding the midbass breakup, while also considerably flattening the upper midrange impedance curve.

While I have managed to level out the average midrange a bit, I seem to be stuck with a 1.5kHz dip of a few dB, as long as I still use this tweeter.

Advice and suggestions welcome! If you think more serious changes can further improve the sound, I'll definitely consider them. Thanks.