@ctrlx , the wool i used in the test , the 900gr case:
This is wool that has been in use from 1991 till 2006, as a result the fibres curling making the flake like shape of the pieces of wool when pulled from the pressed block of wool into more curved shaped lumps .
Quite different from lhw ;-)
Obviously this happens especially with woofers, for midranges this curling is a lot less.
This is wool that has been in use from 1991 till 2006, as a result the fibres curling making the flake like shape of the pieces of wool when pulled from the pressed block of wool into more curved shaped lumps .
Quite different from lhw ;-)
Obviously this happens especially with woofers, for midranges this curling is a lot less.
There was a lot of faulty analysis in that ASR thread. Just because you find a problem and measure a problem does not make them the same.
The resonance from the basket is not like that from a cabinet, it is level dependent, but does not neccesarily increase with level like most others. A specific set of circumstances sets it off. You can get a better idea by measuring at a number of different levels, if it goes away at louder SPL's then it is not the cabinet.
I have done that in the original Orion build and it was quite a pain for no real benefit. Decoupling makes the most sense for a pure midrange where there are easier ways to do it than magnet mounting. the LX521 went back to straight mounting but isolated the woofers from the midrange with a physical bridge. My suggestion was to support the magnet with a brace and some rubber or similar compound in between. That will stop the heavy magnet from setting up the mass spring oscillator. Much easier to do and is targeted at the problem.
Last edited:
@fluid , thanks for the feedback, to test at different SPL's will be done. (just did set up a second computer with a BabyFacePro as interface to measure with a mic, next to the setup for measuring impedances.. Regarding the suggestion to support the magnet with a brace and some compound in between, how did that work out? Sounds similar to a CLD type of damping the magnet movements. I recall a paper of KEF i believe where someting similar was done. Could not find that paper though.
I haven't tried it myself yet, but it is on the list to test with my own Purifi drivers. Something like the LS50 was my plan.
Here are some considerations when designing a magnet-support for a driver:
If the driver is to be firmly fixed at the magnet and firmly fixed at the frame, then some method must be employed to control the preload imparted to the magnet by the strut/brace. Generally this would be a force in the up direction applied at the magnet by the strut, and reacted by the circular frame at the baffle. If this preload is uncontrolled (i.e. if we just screw everything down tight), there could be hundreds of pounds of preload reacting through the driver basket.
One way to avoid this is to soft-mount the driver with a flexible connection, either at the baffle or at the magnet.
Soft magnet mounting: If the soft mount is done at the magnet, then the system does not isolate the driver forces from the baffle over a wide frequency range. Only near the natural resonance of the driver frame will the soft magnet mount provide damping. This is the natural resonance where the magnet is the mass, and the driver frame is the spring.
Soft baffle mounting: This method isolates the driver from the baffle. It requires a thoughtfully designed mounting strut that must precisely positions and restrains the driver in three dimensions of axial positioning and two axis of rotation.
As I said, it is difficult to do this correctly.
j.
If the driver is to be firmly fixed at the magnet and firmly fixed at the frame, then some method must be employed to control the preload imparted to the magnet by the strut/brace. Generally this would be a force in the up direction applied at the magnet by the strut, and reacted by the circular frame at the baffle. If this preload is uncontrolled (i.e. if we just screw everything down tight), there could be hundreds of pounds of preload reacting through the driver basket.
One way to avoid this is to soft-mount the driver with a flexible connection, either at the baffle or at the magnet.
Soft magnet mounting: If the soft mount is done at the magnet, then the system does not isolate the driver forces from the baffle over a wide frequency range. Only near the natural resonance of the driver frame will the soft magnet mount provide damping. This is the natural resonance where the magnet is the mass, and the driver frame is the spring.
Soft baffle mounting: This method isolates the driver from the baffle. It requires a thoughtfully designed mounting strut that must precisely positions and restrains the driver in three dimensions of axial positioning and two axis of rotation.
As I said, it is difficult to do this correctly.
j.
Not sure what you mean. What i do is cover the openings in the basket with an fabric netting quite open so the wool cannot get in touch with the cone or spider.
Then when applying the wool, the wool nuggets against the netting are placed loosely, i pull them a bit looser. Then the remaining wool is placed.
One thing to avoid is an open space. It can become quite audible. The wool content has to sit quite firmly in the cavity.
Then when applying the wool, the wool nuggets against the netting are placed loosely, i pull them a bit looser. Then the remaining wool is placed.
One thing to avoid is an open space. It can become quite audible. The wool content has to sit quite firmly in the cavity.
Back from a week away with children and grandchildren, i tested the testbox on resonances. So here the outcome, with no wool in the box:
The setup is Arta Impulse set to record, and knocking on the side or bottom panel or on the woofer cone.
The blue and red cureve is the response of the TIB connected as source the dark purple is the Woofer connected as source.
It is clear that the anomality in the distortion graph is caused by the panel resonances of the testbox, which happen to be also almost overlapping .
A CD shows the very slow decay ;-)
Basically this means a lot of sound radiated by the cabinet itself, not so good :-(
Anyhow the testbox was primarely for testing the amount of wool to use.
But i will do some tests on how to reduce the panel resonances (thus sound radiation), intrigued as i am, after studying the possible approaches.
Basically in case of panel bracing, there are 2 philosophies: Ridgid like f.i. done by Wilson Audio, or some form of CLD like f.i. done by Kef or Gedlee.
Ridgid will push the resonance frequency up, but most likely with reduced amplitude due to damping of the base material.
Kef applies csd between brace and wall, Gedlee between opposite braces.
Regarding the baffle, a similarity: ridgid or with some flex.
Personally i have opted for ridgid, with HPL (pertinax, hartpapier, x-material, etc) layers with a CSD material in between.
I will receive a AKG C411-PP coming week, so i can test on different spots and on test panels, i happen to have some HPL-CSD panels lying around.
The results will be used for the design of the mid-tweeter baffle and cabinet. (and the tes-cabinet ;-))
Also for the woofer cavities it may be useful to do some bracing, and perhaps also the baffle.
Interesting!
The setup is Arta Impulse set to record, and knocking on the side or bottom panel or on the woofer cone.
The blue and red cureve is the response of the TIB connected as source the dark purple is the Woofer connected as source.
It is clear that the anomality in the distortion graph is caused by the panel resonances of the testbox, which happen to be also almost overlapping .
A CD shows the very slow decay ;-)
Basically this means a lot of sound radiated by the cabinet itself, not so good :-(
Anyhow the testbox was primarely for testing the amount of wool to use.
But i will do some tests on how to reduce the panel resonances (thus sound radiation), intrigued as i am, after studying the possible approaches.
Basically in case of panel bracing, there are 2 philosophies: Ridgid like f.i. done by Wilson Audio, or some form of CLD like f.i. done by Kef or Gedlee.
Ridgid will push the resonance frequency up, but most likely with reduced amplitude due to damping of the base material.
Kef applies csd between brace and wall, Gedlee between opposite braces.
Regarding the baffle, a similarity: ridgid or with some flex.
Personally i have opted for ridgid, with HPL (pertinax, hartpapier, x-material, etc) layers with a CSD material in between.
I will receive a AKG C411-PP coming week, so i can test on different spots and on test panels, i happen to have some HPL-CSD panels lying around.
The results will be used for the design of the mid-tweeter baffle and cabinet. (and the tes-cabinet ;-))
Also for the woofer cavities it may be useful to do some bracing, and perhaps also the baffle.
Interesting!
good work. I also use the driver it self as accelerometer (connect the woofer to the sound card mic input) and then tap the box various places. the recorded pulse reveals the resonant modes (including bass tuning). thanks to the reciprocity the measured response reflects how the woofer excites the panel being hit
Thanks @lrisbo , you mentioned it in the spk16 brief i believe, and yes it works nicely, and quite sensitive as well in the z direction.
To isolate the box from the desk i placed it on large but soft silicon grommets, giving the testbox a base resonance of about 5 Hz.
So above that frequency the box is basically isolated from the activeness of the desk/table. Makes quite a difference.
As a note: the testbox baffle and first part of sides has the shape of the real box (~20 degrees slanted), so off axis behaviour (edge diffraction) resembles the real box.
As a result the bigger part of the sides has almost the same physical panel dimensions as top/bottom, so almost the same reso freq, which is also close to the first internal acoutical standing wave mode.
The wool takes care of the inner acoustics, but not so much the physical activity of the box panels.
The real box otoh has an ellipical crossection (phi) so much less an issue. Yet there are always resonances, so studiing possible measures helps in choosing measures to reduce these in the real box, thus improve the clarity and sound.
The AKG c411 pp gives the opportunity to test at defined (marked) spots and then test the effect of the measures, repeatedly. With a flat panel not so needed(enough publications telling where the resonce peaks occur) , but with curved sides and with baffles it is a different story.
To isolate the box from the desk i placed it on large but soft silicon grommets, giving the testbox a base resonance of about 5 Hz.
So above that frequency the box is basically isolated from the activeness of the desk/table. Makes quite a difference.
As a note: the testbox baffle and first part of sides has the shape of the real box (~20 degrees slanted), so off axis behaviour (edge diffraction) resembles the real box.
As a result the bigger part of the sides has almost the same physical panel dimensions as top/bottom, so almost the same reso freq, which is also close to the first internal acoutical standing wave mode.
The wool takes care of the inner acoustics, but not so much the physical activity of the box panels.
The real box otoh has an ellipical crossection (phi) so much less an issue. Yet there are always resonances, so studiing possible measures helps in choosing measures to reduce these in the real box, thus improve the clarity and sound.
The AKG c411 pp gives the opportunity to test at defined (marked) spots and then test the effect of the measures, repeatedly. With a flat panel not so needed(enough publications telling where the resonce peaks occur) , but with curved sides and with baffles it is a different story.
@lrisbo , a question about breaking in the PTT4.0M04-NAC03 driver.
Out of the box it is ~60 hz, spec says 48Hz.
I did one but even after 12 hours of low frequency 15Hz or 25Hz ,amplitude about 6mm p-p , directly after breaking it was ~50 hz but quicly went up a bit. And after a nights rest it measures about 58 hz.
(With the PTT8.0X08-NAB02 it went nicely to 26 Hz and stayed so )
So what is good breaking in for this PTT4.0M04-NAC03 driver?
Out of the box it is ~60 hz, spec says 48Hz.
I did one but even after 12 hours of low frequency 15Hz or 25Hz ,amplitude about 6mm p-p , directly after breaking it was ~50 hz but quicly went up a bit. And after a nights rest it measures about 58 hz.
(With the PTT8.0X08-NAB02 it went nicely to 26 Hz and stayed so )
So what is good breaking in for this PTT4.0M04-NAC03 driver?
thanks. our end of line test database says fs of 47 and 48Hz so pretty much on spec. Can the difference be due to the setup or test signal used? you are welcome to email my colleague Morten at mh@purifi-audio.com to further discuss and resolve the issue. Crossing my fingers
Issue is no more, Kudo's to Purifi-Audio .
It is direct consequence of voltage used to measure the impedance, and the Rsense value used.
Case01: Rsense 49.7 Ohm, via loudspeaker-output out of Scarlett 2i2, open voltage 550 mV
Case02: Rsense 26,7 Ohm, via amplifier out, output Voltage 1V
Morten of Purifi double checked by testing with 0.1V and he too then measured ~ 60Hz.
I do not know how accurate Arta-LIMP is, but as far as i am concerned it is close enough:
Case03 will be with Rsense of 0.1Ohm, and with amplifier output 2,83Volt as proposed by Purifi. Hopefully tomorrow or Friday.
Note: same Rsense of 0.1Ohm also to use for Current Distortion measurement.
At a shop in Germany i ordered "reststucke" rest posts of HPL (pertinax etc) for the midrange/tweeter test box.
Will arrive in 2 weeks or so.
In mean time i will work on the panel resonance reduction in the bass testbox.
Thanks Morten ;-)
It is direct consequence of voltage used to measure the impedance, and the Rsense value used.
Case01: Rsense 49.7 Ohm, via loudspeaker-output out of Scarlett 2i2, open voltage 550 mV
Case02: Rsense 26,7 Ohm, via amplifier out, output Voltage 1V
Morten of Purifi double checked by testing with 0.1V and he too then measured ~ 60Hz.
I do not know how accurate Arta-LIMP is, but as far as i am concerned it is close enough:
Case03 will be with Rsense of 0.1Ohm, and with amplifier output 2,83Volt as proposed by Purifi. Hopefully tomorrow or Friday.
Note: same Rsense of 0.1Ohm also to use for Current Distortion measurement.
At a shop in Germany i ordered "reststucke" rest posts of HPL (pertinax etc) for the midrange/tweeter test box.
Will arrive in 2 weeks or so.
In mean time i will work on the panel resonance reduction in the bass testbox.
Thanks Morten ;-)