This is a session where some audiophiles just got together, and I brought in a different prototype speaker. I guess many of you will recognise the speakers behind them. After listening to the Thiels and also going through some other equipment others brought in, we finally decided to hook these little babies up.
When they started to play, everyone were surprised how clean the sound was in contrast with the Thiels. The reaction was the same as when a different group listened to some dirty old looking cables I brought to the cable meet. One of the doctors there PMed me later saying he really like that sound. I'm still trying to figure out the best way to manufacture these. The CSD for the tweeter above 10KHz is the one I posted ealier.
The important thing is that we have fun doing these sessions. We share feeling listening to the various systems. Views are exchanged regarding where improvements might be made.
When they started to play, everyone were surprised how clean the sound was in contrast with the Thiels. The reaction was the same as when a different group listened to some dirty old looking cables I brought to the cable meet. One of the doctors there PMed me later saying he really like that sound. I'm still trying to figure out the best way to manufacture these. The CSD for the tweeter above 10KHz is the one I posted ealier.
The important thing is that we have fun doing these sessions. We share feeling listening to the various systems. Views are exchanged regarding where improvements might be made.
SUM,
How close do you calibrate them to? I've been having a hard time getting factory to agree to somthing like 2db. I'm thinking about modifying a driver structure a bit to get better tolerance band.
How close do you calibrate them to? I've been having a hard time getting factory to agree to somthing like 2db. I'm thinking about modifying a driver structure a bit to get better tolerance band.
I wrote it and posted it, read it again and now it is gone. Wazzupwitdat?
We try to never have anything up more than +2dB and a dip of -3dB. This is not always possible. A dip is much more difficult to hear than a peak. It is more important to have frequency balance than absolute flatness. We never use EQ which is lest than 1.5 octaves wide because of the ringing/energy storage problems with high Q filters. This is true except for the case of notch filters used to remove buzz and severe room problems. On our custom install systems of course better flatness is available. The philosophy is this: Remove everything that is bad and what is left should be good. Using DSP this has allowed retuning of existing systems for vastly better ineligibility and performance.
As for the Thiel's and -24dB per octave crossovers- these -24dB/octave filters throw away half the energy one octave above and below the crossover point and result in a "boxy" sound with each driver being very clear but not blending with the other drivers of the system. Probably due to the all pass effect at each crossover point, bad phase. This results in (for a three way) three very clear voices and not much image or blend. A very "flat" sound or as we say here, "DLB, dull, lifeless, and boring." I like music to sound alive and full of life. -24dB per octave always gives a DLB sound.
I was under the impression that Thiels used first order, but I may be mistaken.
How many samples does it take to pick a matched pair? How close do they match?
How many samples does it take to pick a matched pair? How close do they match?
For all you guys who want studies on phase and so on maybe this will interest you.
J.W.Rice Thesis Cover
Using our custom drivers- Midranges fall into for groups and all match within a group except for one group that don't match each other and do not match anything else. These are sold off (car audio). The other three groups match very well within the group.
Tweeters- We test 20 and then match maybe 4 pairs from the 20. Test 20 again (including the unused ones) and maybe get 4 more pairs. This is AFTER all the other testing and modification which weeds out over half of the tweeters. Also of the 20 tested 1 or 2 are no good and rejected and are used for other things (ends of the line array where there is no bandwidth) or attempted to repair. In the end about 1/3 of the original tweeters from the assembly factory make it into the loudspeakers. The rest... well they are still here. Maybe we will get diaphragms made and recycle the magnets. A few tweeters have magnets charged backward, about 1 per 100. Industrial testing has shown out of phase drivers or wiring to be the most common error in loudspeaker manufacturing.
How close do the drivers match? For mids -100-1000 Hz .2dB For tweeters 1200-15000Hz .3dB maximum mismatch at any frequency. Spread- Over 2dB for mids and about 4.5dB for tweeters. The curves for tweeters sometimes looks like a bowl of colorful spaghetti when there are no matches or a lot of defective ones in a test group of 20. Sometimes there are no matches in a group of 20- then we go to lunch and try again later.
For my pair? .1dB for both
hahahaha insane 
J.W.Rice Thesis Cover
Using our custom drivers- Midranges fall into for groups and all match within a group except for one group that don't match each other and do not match anything else. These are sold off (car audio). The other three groups match very well within the group.
Tweeters- We test 20 and then match maybe 4 pairs from the 20. Test 20 again (including the unused ones) and maybe get 4 more pairs. This is AFTER all the other testing and modification which weeds out over half of the tweeters. Also of the 20 tested 1 or 2 are no good and rejected and are used for other things (ends of the line array where there is no bandwidth) or attempted to repair. In the end about 1/3 of the original tweeters from the assembly factory make it into the loudspeakers. The rest... well they are still here. Maybe we will get diaphragms made and recycle the magnets. A few tweeters have magnets charged backward, about 1 per 100. Industrial testing has shown out of phase drivers or wiring to be the most common error in loudspeaker manufacturing.
How close do the drivers match? For mids -100-1000 Hz .2dB For tweeters 1200-15000Hz .3dB maximum mismatch at any frequency. Spread- Over 2dB for mids and about 4.5dB for tweeters. The curves for tweeters sometimes looks like a bowl of colorful spaghetti when there are no matches or a lot of defective ones in a test group of 20. Sometimes there are no matches in a group of 20- then we go to lunch and try again later.
For my pair? .1dB for both
hahahaha insane
I have a few questions about whatever listening tests that were conducted to determine relationship between polar response and preference:
1. Were the speakers all toed in so that the listener was on axis? or were they all facing frontward?
2. Were there a group of people listening together? or did each individual provide preference independently?
1. Were the speakers all toed in so that the listener was on axis? or were they all facing frontward?
2. Were there a group of people listening together? or did each individual provide preference independently?
OK here is a question. What is the graph below telling us? The measurements are of my MTM at 1M on axis and then 15, 30, 45, 60, and 90 deg off axis in the horizontal plane. (those angles being rather approximate as I didn't have a protractor handy, and guestimated) The earlier measurements I posted of just the woofer, I did a little more scientifically).
The first thing that sticks out like a sore thumb to me is that the tweeter response after 4K drops like a stone even at 15deg. Since this tweeter should have better performance than this even at 45deg (only being down 5db at 9K 45deg according to manufacturers specs) I suspect that there is something very wrong with my measurement technique for off axis measurements!
Anyone care to throw out some theories about what I might be doing wrong? and why the response would be so consistent off axis until 4K and then suddenly diverge so much.
Tweeter is a one inch semi horn loaded dome (morel DMS37), woofers are morel MW=144's. Crossover freq is 2nd order bessel (acoustic) at 3Khz.
I probably won't get a chance to do any more measurements for a while, but this was bugging me so thought I'd ask 🙂
Tony.
The first thing that sticks out like a sore thumb to me is that the tweeter response after 4K drops like a stone even at 15deg. Since this tweeter should have better performance than this even at 45deg (only being down 5db at 9K 45deg according to manufacturers specs) I suspect that there is something very wrong with my measurement technique for off axis measurements!
Anyone care to throw out some theories about what I might be doing wrong? and why the response would be so consistent off axis until 4K and then suddenly diverge so much.
Tweeter is a one inch semi horn loaded dome (morel DMS37), woofers are morel MW=144's. Crossover freq is 2nd order bessel (acoustic) at 3Khz.
I probably won't get a chance to do any more measurements for a while, but this was bugging me so thought I'd ask 🙂
Tony.
Attachments
I have the Morel MDT37, which is related to your tweeter. I crossed mine at about 2,700 Hz, but I do not see that magnitude of drop in off-axis performance with mine.
This is my plot at 1 meter, mic lined up with tweeter axis, from 0° to 60° off-axis, in 7.5°*increments:
An externally hosted image should be here but it was not working when we last tested it.
Ignore everything below about 250 Hz due to room reflections.
I would not expect the crossover type to have that much of an impact in off-axis response.
Thanks Loren your plot seems much more like what I would have expected 🙂
I think I've worked out where I went wrong with the measurements off axis (horizontal). I rotated the speaker on its centre axis, where I should have rotated it around an axis level with the front baffle (not the centre of the speaker). Because the tweeters are semi horn loaded the mic wouldn't have been "seeing" the actual dome and the effective angle would have been much greater. Doh. A quick look at the vertical off axis I showed earlier shows the tweeter only being down about 1.5db at 10K on the 15 degrees measurement. That measurement was taken by lowering the mic so wouldn't have had the same issue.
Tony.
I think I've worked out where I went wrong with the measurements off axis (horizontal). I rotated the speaker on its centre axis, where I should have rotated it around an axis level with the front baffle (not the centre of the speaker). Because the tweeters are semi horn loaded the mic wouldn't have been "seeing" the actual dome and the effective angle would have been much greater. Doh. A quick look at the vertical off axis I showed earlier shows the tweeter only being down about 1.5db at 10K on the 15 degrees measurement. That measurement was taken by lowering the mic so wouldn't have had the same issue.
Tony.
Tony, what do you make of the sound? It looks like your crossover is right.
Loren, Are you going to use that tweeter with the AE?
Thanks,
Dan
Loren, Are you going to use that tweeter with the AE?
Thanks,
Dan
Yes. I have them installed now and ran some on-axis plots to see how they worked out with the Audax and Morel and the TD15H seems like a glove fit replacement for the JBL, even with the existing crossover. I am impressed.
Hi Dan,
I think it sounds very good, voice clarity is excellent! Bass will be better once I build my active crossover and cross to my 10" drivers (planning to do so around 200Hz), I may put some line level BSC in as well.
The crossover is ultra simple at the moment. Single cap and a series resistor on the tweeter, only a 4K notch filter on the woofer, which gives a 2nd order bessel acoustic rollof at 3K. I'd like to try a 4th order L/R at 2K, which will hopefully remove the bump between 1k and 2k resulting in an overall smoother response. I'm not sure how much colouration this bump is adding though so may end up being a learning excercise but not result in better sound.
I'll have to do the measurements again properly now I have realised what probably caused the problem 🙂 but it might be a while before the next opportunity arises and I have to work out a way to swivel the speaker whilst keeping the front baffle in the same plane!! My stand isn't really set up for that....
Tony.
I doubt it would make that much of a difference in the amplitude. There might be some difference in edge diffraction influencing the response.
You know Tony, my room is completely modal below 200Hz. BSC may be a good thing or a bad thing if you situation is similar. A separate bass amp with a parametric eq would be better IMO, but cost would go up. I think Dayton just made a new one that might be interesting. I'd say you definitely got the MTM section looking great and to me there some sort of elegance to such a simple crossover. Makes me wish I had hung on to my Morels. Funny how I wanted to buy that same tweeter as well, but bought the MDT 30s and later the 33s instead b/c I was afraid of honk. My Klipsch KG5.5s made me afraid of horns for years to come.
Dan
Dan
Here's some nice measurements. Anyone care to guess what made these?
Polar:
impulse:
AVG of polar response from 11.25 degrees off axis to 90 degrees off axis:
Any thoughts? Personally I wish the pattern was a bit more narrow, but I'm not complaining.
Dan
Polar:
impulse:
AVG of polar response from 11.25 degrees off axis to 90 degrees off axis:
Any thoughts? Personally I wish the pattern was a bit more narrow, but I'm not complaining.
Dan
Last edited:
- Status
- Not open for further replies.