Dynaudio identification help required.

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
New mic has been ordered and shipped. umik-1 should be here mid - late next week...

Going from what I measured with the 2.5k bw3 and the earlier sweeps of the raw drivers.
I believe the 3-4k peak is the tweeter.
It didn't sound too bad. Vocals were a bit foward though.

So I was thinking your idea of a higher ~3.5k xover may be what they want.
Depending on how that measures/sounds, I thought staggering the xover to try and force a flatter response could be called for.

But. Until the mic arrives, it's all just pure guess work.
Well it is for me anyway :D
 
This project can be done in a half an hour of experimenting because there is a measurement system and the LP filter is already in the ball park with 1,8 mH and 10 uF, if you want my opinion. That would create a XO point in the 2 kHz region, as it should be with a large of a woofer as this one. However, freedom of choice is yours. Whatever makes you happy.
 
It's all experimental, IMO. A sloped baffle is, by definition, listening off-axis, so the actual modelling is quite unusual.

643262d1509506818-dynaudio-identification-help-required-bw3-2500hz-theoretical-png


This would be my point of departure. Those values are fairly written in stone for the ideal impedance. The cone breakup is well down at 2.5kHz crossover. You need to add a series resistor to the tweeter to get it to 8R DCR. And the 9.1R/22uF to impedance correct the woofer. Then you add a resistor to the front of the tweeter circuit for level.

I wouldn't fancy 2kHz crossover, because we know this pushes the tweeter hard. And the 3.5kHz will let through some cone breakup. Joachim Gerhard used 2nd order here by the look of things on acoustic BW3 on the Anima: Sonics Anima loudspeaker | Stereophile.com

Impedance isn't very nice, IMO, but an interesting design. You really don't know how this will sound till you build it. Lot of compromises.
 
Last edited:
Sure, they were. Do the same, for the woofer record two versions, one with 10 ms time window, the other with 15 ms. Tweeter time window can be shorter, say 2-3 ms. Phase response of the frd's will be calculated separately in Response Modeler, so don't worry about it or just select minimum phase. On the other hand, phase data of zma's I will use, in a stepped sine mode, frequency resolution 1/24, please.
 
Yay. No one's annoyed me today. Means I can get stuff done :D

So here's the sweeps.

Mic at 1 meter, centred between driver centres, 1/24 smoothing.

Bass driver at 10ms and 15ms with 88200 sample rate.
Tweeter at 3ms, 88200 and 192000 sample rate.

Also a screen shot of the Arta graph window.

I note the tweeter looks smoother with the higher sample rate.

Selecting 'Minimum phase' in the 'smoothed FR' window does nothing ?
 

Attachments

  • 8-11-17 bass 10ms 88200.txt
    14.2 KB · Views: 21
  • 8-11-17 bass 15ms 88200.txt
    14.2 KB · Views: 33
  • 8-11-17 Tweeter 3ms 88200.txt
    14.1 KB · Views: 28
  • 8-11-17 Tweeter 3ms 192000.txt
    14.1 KB · Views: 27
  • 8-11-17 Raw driver Sweeps.jpg
    8-11-17 Raw driver Sweeps.jpg
    282.6 KB · Views: 73
Ok, 88.2 kHz sampling frequency is no good. I'd like you to record the FR plots under following conditions. I can use impedance from before, so don't need these right now.

3 ms time window; 1 m away from baffle; mls mode; 192 kHz sampling fr; mid point between units

woofer.txt
tweeter.txt
woofer+tweeter.txt (both units wired in parallel)


10 ms time window; 1 m away from baffle; mls mode; 192 kHz sampling fr; mid point between units

woofer.txt
tweeter.txt

Make sure the time window captures the impulse from its start, I haven't measured FR with ARTA yet so I want to be sure about that.
 
Started working on a filter, while doing it I thought to ask if you have any passive parts at hand so I can work out a version that you may be able to play as soon as you assemble it. Any caps in the 10 uF vicinity; inductors 1.5-2.0 mH; 0.3 mH, some resistors 2.2-4.7 R ?
 

Attachments

  • Filter.png
    Filter.png
    10.7 KB · Views: 123
Last edited:
I have that xover in place on one speaker.

I've swapped the tweeter resistor from 3.3 to 4.7.
Listening it was a little too bright on axis.

Measures close to ruler flat though.
I'm guessing the dip at 1.1k is baffle interaction ?

Screen shot of sweep with 3.3 / 4.7 / reversed tweeter.

Time to listen and see what their like now.

EDIT, I have no zobel in their at present, only what Lojzek has simmed.......
 

Attachments

  • System xover 1 - 4.7 gate.jpg
    System xover 1 - 4.7 gate.jpg
    204 KB · Views: 122
  • System xover 1 3.3 + 4.7 + Reversed.jpg
    System xover 1 3.3 + 4.7 + Reversed.jpg
    221 KB · Views: 115
  • System xover 1 3.3.txt
    14.2 KB · Views: 24
  • System xover 1 4.7.txt
    14.2 KB · Views: 29
  • System xover 1 4.7 Twtr Reversed.txt
    14.2 KB · Views: 31
Last edited:
Good, you keep up listening to it and we will see after some time how this goes. Zobel you don't need and it would just make spl before XO point to take a dive. We can try a second version of a filter with a 3 kHz XO point, and then you will decide what may suit better. The 1 kHz resonance is a normal thing with these units and often is linked to being caused by cone surround measures.
 
Yes I kinda thought the zobel wasn't needed in this design.
Otherwise you would've shown it :)

I now have both sides running.
Sounds pretty good to me.
I mean ok I can pick faults, but I'm used to my speakers that are just that weee bit more expensive ;)
Main fault (if I can use that term) is the quality of the treble, but I guarantee thats due to the odd mix of caps I'm using for the tweeter at present.

The owner is visiting sometime this weekend to give his verdict.

For the inductors, obviously I should aim for lowest dcr on the bass series 1.8mh.
But is there any guide as to what dcr I should aim for with the tweeter shunt 0.3mh ?
 
I have assumed the woofer inductor to be about 0.5, and the tweeter around 0.3. These values are not critical. I'll post the 3 kHz XO version in case this loudspeaker is going to be driven hard and I would want its tweeter to survive in any case.


edit: 2nd schematic, 1.8 mH(dcr 0.5); 0.82 mH(dcr 0.3); 0.2 mH (dcr 0.2)
 

Attachments

  • 3 kHz XO point version.png
    3 kHz XO point version.png
    11 KB · Views: 106
Last edited:
You are welcome. I am using Bill's "XSim" for quick creation of non-standard filters although Bagby's "PCD" is faster to come up with a solution if the schematic that it is able to simulate will do the job. They are all doing the same thing, crunching numbers.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.