JBL 2446J comparison without a horn. The mic's is hard to get to exact same position. Some of the variance comes from that, for example in the extension. I think this is reasonable enough tolarence between the original membranes that are many years old.
Measured directly from the mouth:
10cm away from the mouth:
Measured directly from the mouth:
An externally hosted image should be here but it was not working when we last tested it.
10cm away from the mouth:
An externally hosted image should be here but it was not working when we last tested it.
Both channels ready finally. They are just plain gorgeous. Now standing at 185cm tall with the temporary synergy-stands. I'm in heaven just by looking at them in the spot lights. Gf likes them also😀. Proper listening/measuring will have to wait for tomorrow, it's too late tonight.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Thanks! They are actually my old closed 18" 110 litre infra-subwoofer boxes which I converted to dipoles couple of years ago. 🙂 The propeller looking thing is an inner support.
I will have to think some nicer stands or perhaps cover these in white/red speaker cloth and suff them full of effective acoustic absorbing material to slightly help to damp the longitudinal axis of the room.
Preliminary measurment showed that the response was smoother than the pre-throat, injection port etc. rounding and smoothing. I will try to do some measurements and comparisons in next few hours.
Last edited:
And here are the new measurements.
New horizontal polars are measured the same way as before mic hanging at the mouth in even step between the center and the edge of the coverage. I did not any vertical polars yet.
Here's the last position of the mic:
The polars in 10 even steps (original 500ms time window):
Old polars vs. new polars in 1/3 (10ms window):
in 1/6: (10ms window in left plot, 500ms in right, by mistake)
In 1/12:
0deg measurement comparisons (old on left, new on right):
In same graph (old is the red, new the black plot):
27,5 deg comparison (the edge of horn's coverage, like the mic was in the picture above):
The new graphs look nicer (and not in a small way, Art 😉). All the work did wonders measurement-wise. The measured response is allround smoother including the very edge of the horns coverage. The response at the edge is very smooth compared to before, almost perfect slowly descending slope.
I would also say the effect in heard sound is allround positive, but the leap is not big as they were very good before. Let's say they are more matured, slightly more relaxed, a tad more natural in balance, slightly better at imaging (with mono pink noise the "sound pillar" at the center is now narrower IIRC, for example) etc.
I will do the other horn soon. I hope I got the tolerances right, although human error, when not using a cnc machne, is always possible (and on parts per million/thousand level always present).
New horizontal polars are measured the same way as before mic hanging at the mouth in even step between the center and the edge of the coverage. I did not any vertical polars yet.
Here's the last position of the mic:
An externally hosted image should be here but it was not working when we last tested it.
The polars in 10 even steps (original 500ms time window):
An externally hosted image should be here but it was not working when we last tested it.
Old polars vs. new polars in 1/3 (10ms window):
An externally hosted image should be here but it was not working when we last tested it.
in 1/6: (10ms window in left plot, 500ms in right, by mistake)
An externally hosted image should be here but it was not working when we last tested it.
In 1/12:
An externally hosted image should be here but it was not working when we last tested it.
0deg measurement comparisons (old on left, new on right):
An externally hosted image should be here but it was not working when we last tested it.
In same graph (old is the red, new the black plot):
An externally hosted image should be here but it was not working when we last tested it.
27,5 deg comparison (the edge of horn's coverage, like the mic was in the picture above):
An externally hosted image should be here but it was not working when we last tested it.
The new graphs look nicer (and not in a small way, Art 😉). All the work did wonders measurement-wise. The measured response is allround smoother including the very edge of the horns coverage. The response at the edge is very smooth compared to before, almost perfect slowly descending slope.
I would also say the effect in heard sound is allround positive, but the leap is not big as they were very good before. Let's say they are more matured, slightly more relaxed, a tad more natural in balance, slightly better at imaging (with mono pink noise the "sound pillar" at the center is now narrower IIRC, for example) etc.
I will do the other horn soon. I hope I got the tolerances right, although human error, when not using a cnc machne, is always possible (and on parts per million/thousand level always present).
Last edited:
Here's the horn to horn comparison. The first one I measured is on the left in the pictures. Reasonably similar considering they are not far field measurements and mic location accuracy is approx. +/- 1cm from point to point. The greatest difference is in the HF, left horn seems to beam it couple dB more towards the edge of the horn's flare.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
Your efforts paid off! Looks great. You really smoothed out some of those peaks. Gives me some hope on my tractrix as I have some rather rough edges near the throat area with cardboard and hot melt glue used.
Legis,
Nice improvement above 3.5 kHz. Were the 15" shorted out in the before and after comparison tests?
In the on axis comparison, it appears the old (red) mid band level is about 4-5 dB higher than the new (black) level, were the same drive levels used?
Art
Thanks! Yes they were shorted to the output of the amp (drivers in paralllel). I always just mute the channels I don't need from Nanodigi and leave their amps powered.
The drive level was not the same but it was enough for near field measurement. I did not take otes what drive level I used before so it would have been quessing since I can adjust volume in 4 places (3 in nanodigi and also from the amp).
I have learned the hard way and now keep notes in the notes section of the REW measurement. I generally leave amp and miniDSP maxed out and control level in REW and make note as -22dB FS for example. I find that it is very reproducible from day to day. Just make sure the settings for gain are either fixed or known.
When I got first got miniDSP somehow input gain in miniDSP was set at -10dB and for a month I couldn't figure out why it was playing so low when going through miniDSP - I kept thinking the output DAC was not having enough drive voltage. Operator error. 😀
Your plots look really good. Must sound awesome now.
When I got first got miniDSP somehow input gain in miniDSP was set at -10dB and for a month I couldn't figure out why it was playing so low when going through miniDSP - I kept thinking the output DAC was not having enough drive voltage. Operator error. 😀
Your plots look really good. Must sound awesome now.
Last edited:
I have chosen to keep MiniDSP's inputs at ~ -12dB for safety reasons because I control the master volume with it (via remote) right now. Dangerous amount of sensitivity and explosiveness if someone sits on the remote controller by mistake. 🙂
Before the painting/throat smoothing I liked their sound better with -3dB notch filters at 3kHz and 13,5khz. Now I don't like those (better without), I just use 4dB high shelf filter (0,5Q / 6kHz) for the 2446J's. Without the high shelf they sound a tad on the dark side.
Deltalites are completely without electrical xo'ing and eq'ing, this way they time-align exactly with 2446J that is highpassed 2nd order at 500hz. If I use 500Hz/2nd order LPF for Deltalites I need to put 0,6ms delay for them to time-align with 2446. I have not decided which I like better, with or without electrical xo for Deltas, jury is still out. Difference is not big. I always run into these small decision issues with active system. 🙂
Super tweeters are crossed at 9kHz/8th order Butter. I feel most comfortble with them this way (7-9khz and as steep as possible). Only add a tad of sparkle and fairy dust without adverse effects in feeling of coherency 🙂.
These are quite preliminary settings and chosen by listening. Here's how things look from the listening spot with above mentioned settings:
L & R:
Left channel broken in pieces:
Before the painting/throat smoothing I liked their sound better with -3dB notch filters at 3kHz and 13,5khz. Now I don't like those (better without), I just use 4dB high shelf filter (0,5Q / 6kHz) for the 2446J's. Without the high shelf they sound a tad on the dark side.
Deltalites are completely without electrical xo'ing and eq'ing, this way they time-align exactly with 2446J that is highpassed 2nd order at 500hz. If I use 500Hz/2nd order LPF for Deltalites I need to put 0,6ms delay for them to time-align with 2446. I have not decided which I like better, with or without electrical xo for Deltas, jury is still out. Difference is not big. I always run into these small decision issues with active system. 🙂
Super tweeters are crossed at 9kHz/8th order Butter. I feel most comfortble with them this way (7-9khz and as steep as possible). Only add a tad of sparkle and fairy dust without adverse effects in feeling of coherency 🙂.
These are quite preliminary settings and chosen by listening. Here's how things look from the listening spot with above mentioned settings:
L & R:
An externally hosted image should be here but it was not working when we last tested it.
Left channel broken in pieces:
An externally hosted image should be here but it was not working when we last tested it.
Some interesting measurements I though you might be interested in. I compared two good sounding settings. 1) Deltalites have no low pass/eq 2) Deltalites have 400Hz 1st deg low pass. Electrical delay is the same for both channels in both cases, I don't have to use delay to time align them. 2446J have 2nd order Butter filter at 500hz.
Here's how they look.
Everything is measured mic hanging on the mouth.
1) Without low pass
2) With low pass:
3) Excess group delay without low pass (20ms time window):
4) Excess group delay with low pass filter (20ms time window):
5) Step response comparison (red is the version with low pass)
The synergy horn is quite phase linear and time coherent, although the step response is not the perfect "one step" triangle resembling ultimate time coherency (like some full range electrostatic speakers have for excample).
I'm not sure if such step response can be realized easily in horn systems because of the steep "acoutical crossovers" that horns usually have. Both the response of the comp driver and the mids are cutted quite steep even without electrical xo's. The best time coherency would require use of 1st (or perhaps 2nd order would pass) order filters which include the acoutical filters.
If the mids would go to 2khz, and the comp to 500Hz, and xo'ing at 1kHz (one octace to both directions) one could get slow crossovers slopes realized.
Maybe phase plugging the front chamber would lessen the acoutical low pass filtering of the mids but I doubt much beyond 1kHz is realistical. 2khz might very hard even for very small mids, like 6,5" and smaller.
What do you think of the measurements?
Here's how they look.
Everything is measured mic hanging on the mouth.
1) Without low pass
An externally hosted image should be here but it was not working when we last tested it.
2) With low pass:
An externally hosted image should be here but it was not working when we last tested it.
3) Excess group delay without low pass (20ms time window):
An externally hosted image should be here but it was not working when we last tested it.
4) Excess group delay with low pass filter (20ms time window):
An externally hosted image should be here but it was not working when we last tested it.
5) Step response comparison (red is the version with low pass)
An externally hosted image should be here but it was not working when we last tested it.
The synergy horn is quite phase linear and time coherent, although the step response is not the perfect "one step" triangle resembling ultimate time coherency (like some full range electrostatic speakers have for excample).
I'm not sure if such step response can be realized easily in horn systems because of the steep "acoutical crossovers" that horns usually have. Both the response of the comp driver and the mids are cutted quite steep even without electrical xo's. The best time coherency would require use of 1st (or perhaps 2nd order would pass) order filters which include the acoutical filters.
If the mids would go to 2khz, and the comp to 500Hz, and xo'ing at 1kHz (one octace to both directions) one could get slow crossovers slopes realized.
Maybe phase plugging the front chamber would lessen the acoutical low pass filtering of the mids but I doubt much beyond 1kHz is realistical. 2khz might very hard even for very small mids, like 6,5" and smaller.
What do you think of the measurements?
Last edited:
The near 10dB hole in the response around the acoustical crossover point makes me think the 15" or 2446 need to have their polarity reversed to correct the response.
Danley doesn't use first order slopes.
He hasn't published his xover topology.
LeCleach doesn't use first order slopes.
He *has* published his xover topology.
The solutions that each uses are different, but the ideas are similar. LeCleach has a PDF that's about 30 pages long that describes how he gets (nearly) flat phase with high order slopes.
He hasn't published his xover topology.
LeCleach doesn't use first order slopes.
He *has* published his xover topology.
The solutions that each uses are different, but the ideas are similar. LeCleach has a PDF that's about 30 pages long that describes how he gets (nearly) flat phase with high order slopes.
Seems like when the time-coherency is at it's best, the phase aligment is not. This might have something to do with the unsymmetrical crossover reqion? I will keep trying to twiddle with the higher order slopes with the comp driver to see if I can get it workin'.
By the wqay, when I have measured the response by hanging the mic to the mouth, the HF response is too attenuated because of the mic's position. The mic should be omni but it is not...
Here's the effect of directing the mic to the throat vs. hanging it from the mouth like in the previous pictures:
Here's the phase measured with mic directed to the throat (see the crossover reqion from below):
Excess group delay:
"The most" (this far) time coherent (least excess group delay) crossover is not phase aligned. No summation at all at xo freq.
Synergy horn acts as a coaxial driver, the time aligment between the drivers stays very controlled to all hor/vert directions. Here I measured the step response from min-max angles of the mouth like in the polar measurements by hanging the mic from the mouth:
edit. Also see Danley Synergy's advertisement-measurements (I bet they are accurate, but still not 3rd party made) from here. Phase response seems quite like mine.
- http://www.danleysoundlabs.com/danley/wp-content/uploads/2012/01/The-Tapped-Horn.pdf
- http://www.danleysoundlabs.com/danley/wp-content/uploads/2012/04/SH-50-Spec-Sheet.pdf
Last edited:
Actually now when I see the Danley SH50 measurement in it's own PDF, I see that my synergy has less phase shift between 50Hz and >10khz than SH50. Mine has approx 360 deg, SH50 approx. 540deg (675 minus 135 from the plot). Maybe it's understandable because 2-way vs. 3-way. http://www.danleysoundlabs.com/danley/wp-content/uploads/2012/04/SH-50-Spec-Sheet.pdf
Here 1/48 smoothed version with 15ms window:
Have never seen a step response and/or excess group delay measurement from Danley synergy (or any other synergy for that matter), but those would be interesting to see.
I have always tried for the best time coherency and phase smoothness through each crossover range, but crossover poles add time delay, so you will always see phase rotation using them. My TH/offset 8"/Paraline rig using 24 dB filters shows 720 degrees of phase rotation (as expected for 8 poles) over the pass band, but the crossover points are undetectable, the phase change is smooth through the crossover region.
Since your phase and frequency response through the crossover show a deep dip, while the individual response of the 15"s and the 2446 don't, either the polarity is backwards, or the time alignment off. When you get the polarity correct, it may require a time alignment adjustment.
At any rate, you should be able to get a much smoother response through the crossover region than you have presently, though some EQ (which also corrects phase) may be needed.
As far as the mic being omni, the width of the capsule will affect how "omni" it is, you need around 1/4" width to go out to 20 kHz, and even then the back HF response is "shadowed" by the width of the XLR. Point the mic at the HF source ;^).
Art
Last edited:
Hi Art, do you have some step response and/or excess group delay plots posted here somewhere of some of your designs?
I have always preferred slower xo slopes (1st-2nd order if possible), they always sound quite more natural to my ears.
Some more measurements. The best time coherency and the best phase matching require slight changes to the xo'ing.
Current settings:
2446J: 12dB/oct Butter @ 500hz.
Deltalites: No electrical xo'ing.
(the mic is hanging also in these measurements, so HF's level is slightly more depressed than it is in reality)
1) The more phase aligned setting (7ms delay for both 2446J and Deltalites)
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
2) The more time coherent setting: (6.8ms delay for 2446J and 7ms for Deltalites)
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
To my ear the more time coherent settings sounds allround better than the more phase aligned setting.
For reference, here's my 80cm x 220cm fullrange electrostatic speakers "perfect" step response, phase and excess GD plot. Fullrange 1-way 'stats are inherently very time coherent "flyswatters".
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
- Status
- Not open for further replies.