Now you know why some people like PA drivers used for Hifi purposes (OK this is not the only reason, but one of them) !
There is another ugly effect when soft suspensions are used: Some reflex tunnels have asymmetric flow resistance. This leads to some "rectification" of the air pressure inside the box with a subsequent signal dependant offset of the voice-coil.
Regards
Charles
There is another ugly effect when soft suspensions are used: Some reflex tunnels have asymmetric flow resistance. This leads to some "rectification" of the air pressure inside the box with a subsequent signal dependant offset of the voice-coil.
Regards
Charles
Same model. I just plugged in the altered BL and Cms values corresponding to the excursion needed to get the desired SPL at 40Hz.
These altered Bl and Cms values came from where? Were cancelling non-linearities accounted for? Was the excursion within the Xmax on the spec sheet?
Would you be so kind and explain what values at what
excursion have you included in the simulation to get the
second graph, bumpy one.
Your simulation probably means one should not try to
win Formula 1 Grand Prix with a medium powered family
car.
I used 7mm to make it more evident - so outside declared Xmax, but within the max mechanical excursion.
According to my back-of-the envelope calculations, this is the excursion that would be required to produce realistic levels (80dB mean with 18dB crest factor) at a 3m listening distance.
The BL decreases from 6 to ~3, and the Cms (=1/Kms) also goes down by a factor of 2 (see: http://techtalk.parts-express.com/showthread.php?227563-Scanspeak-18w-4434g00-Klippel-Results)
Using the published Xmax = 4.5mm you would get a less severe, but still considerable, parameter drift (e.g. BL ~4.7), and the following response:
An externally hosted image should be here but it was not working when we last tested it.
A bit less horrendous, but still not pretty.
Exactly my point
Also, this is why people mistakenly think the whole Bass-Reflex concept is flawed (when the muck they hear is really due to the inadequate drivers used).
Last edited:
I had looked at the link earlier but could not see anything that looked like an average Bl and Cms values over a large displacement. I am not familiar with Klippel output and so they might well be there. But at present I do not know where your values are coming from and hence whether they are reasonable for the point you are making.
I think Marco is making the interesting point that reflex, which relies on the motor and surround for linearity, may go horribly, horribly wrong at high power. Ferrites saturate, suspensions start to bounce. It's a bit worrying.
Compare that with acoustic suspension and wadding. Air compresses fairly predictably for the restoring force.
An externally hosted image should be here but it was not working when we last tested it.
But our main criticism of reflex remains that it doesn't have any slam or timing. I really don't know how transmission line fits into that. I have listened to big IMF transmission lines, and they weren't good party speakers IMO, but maybe that is more a criticism of bextrene cones and steep filters.
Exactly my point
Also, this is why people mistakenly think the whole
Bass-Reflex concept is flawed (when the muck they
hear is really due to the inadequate drivers used).
Thanks Marco for the response. I was playing with similar
numbers and it appears Qts has to rise approx. twice the
original value to start looking like your first bumpy graph.
I used 7mm to make it more evident - so outside declared Xmax, but within the max mechanical excursion.
According to my back-of-the envelope calculations, this is the excursion that would be required to produce realistic levels (80dB mean with 18dB crest factor) at a 3m listening distance.
The BL decreases from 6 to ~3, and the Cms (=1/Kms) also goes down by a factor of 2 (see: Scanspeak 18w/4434g00 Klippel Results)
Using the published Xmax = 4.5mm you would get a less severe, but still considerable, parameter drift (e.g. BL ~4.7), and the following response:
An externally hosted image should be here but it was not working when we last tested it.
A bit less horrendous, but still not pretty.
Exactly my point
Also, this is why people mistakenly think the whole Bass-Reflex concept is flawed (when the muck they hear is really due to the inadequate drivers used).
If you had even more of an extended bass shelf (EBS) shaped response to begin with, this hump is not a problem and it actually fills in and extends the flat area of the passband nicely. This was mentioned in an earlier post of mine in this thread.
Not true.
Cms decreases with excursion (the suspension becomes stiffer and stiffer), and Fs is inversely proportional to Sqrt(Cms), so it increases.
Likewise, BL decreases with excursion (as the voice coil leaves the gap), and Qes is inversely proportional to BL^2, so it too increases.
Last edited:
Possibly, in some cases it can be factored in, yes.
However, you still get difference response curves at different listening levels, AND of course inevitable power compression (decreased BL = lower electroacustical conversion efficiency) as the level is increased.
As already mentioned I prefer closed box for listening at home. But I own both types of speakers and I have also heard many good spekers of both principles.
I think the problem arises when someone wants to build as cheaply as possible. The EBS responses mentioned by Charlie can't be had with underdimensioned box sizes.
Often the temptation is big to dimension on the small side and thinking that nothing than just adding some type of "hole in the box" will get some magic improvement. Then there are the ones who believe one can use 6 inch drivers for disco levels in huge living rooms.
Engineering is always a balancing act between good and bad properties of certain construction principles and then chosing the one that is best fitted for a given situation.
Regards
Charles
I think the problem arises when someone wants to build as cheaply as possible. The EBS responses mentioned by Charlie can't be had with underdimensioned box sizes.
Often the temptation is big to dimension on the small side and thinking that nothing than just adding some type of "hole in the box" will get some magic improvement. Then there are the ones who believe one can use 6 inch drivers for disco levels in huge living rooms.
Engineering is always a balancing act between good and bad properties of certain construction principles and then chosing the one that is best fitted for a given situation.
Regards
Charles
I always wanted to believe that it was the changing of distance from the back of the driver to the ports opening on the face of the cabinet. The ports next to the speaker had a shorter distance to travel as opposed to the ports at the very bottom. But that's just one of my many speculations as to why they sounded the best.
It is the horribly, horribly that I am trying to check. It doesn't look like we will know how Marco determined average Cms and Bl values for a given extension from the non-linear Klippel data. My guess is that the second curve is not a reasonable representation of the driver driven well beyond its linear range. Which in turn is not particularly useful anyway. What would be more useful is what the curve looks like at xmax from spec and/or the Klippel data. This could be taken from a measurement of how the TS parameters vary with voltage and/or heating up. There is some data on the web on the topic. Alternatively one could integrate the Klippel curves assuming some non-linear model to get equivalent average parameters for a linear model for a given extension.
What is trying to be shown is useful information but it would be rather more so if it was reasonably correct.
hi Dave, I think you have that round the wrong way, a vented box would be smaller than a closed box. For example a scanspeak18W8535 in a sealed enclosure 60 litres and ported 101 litres. That me be the reason for the port, to allow the driver the space to breath, but as you say probably not ideal.
I think speakerdave got that exactly right. Most reflex boxes are evidently undersized.
Let's do your 18W8535: Vas 71.2 L, Qts 0.38.
Closed box @ Qtc 0.7 = Vas / 2.39 = 29.7 L
Reflex = 71.2 L (I don't have to think about that with a Qts of 0.38!)
But Troels Gravesen gives a driver like that a mere 33L in a floorstander with a longish port! He'd probably do even less with a standmounter.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.