Good point, these were some images I could quickly find to try and save a few thousand words, good images for high passes eluded me, I tried flipping them in Paint but it seemed more confusing
Attachments
How far away was the mic for this measurement?
Can you re-post that pic without the SMOOTHING So we can see the RAW response
You spoke of the "bigger picture".... I think you would like to visit burst decay and avoid the same area where GD gets too high, which is also the area with excess decay
Last edited:
I just spent several minutes trying to figure out why I thought Fc was for sealed boxes and not conical expansion only to realize that it is ambiguous lol
Regarding your last statement, for domestic use we aren't going anywhere near peak power ratings...so playing below fs on a horn is applicable and dictated by excursion in which the horns loading will help alleviated...I know you know this already, just keeping it in the picture
Last edited:
I thought the connection to time to peak energy, was exactly that...
Remember I posted the graphs next to each other and the trend lines where almost identical?
If there is truly a connection to the amplitude envelope, as I suspect and read about....I just realized that I should be able to record it and display the signal as such....I will try.
Fluid
Unfortunately your pictures do only tell the truth when it comes to amplitude- and phase- response.
The group delay has the same lowpass-like shape for the high- and lowpass. This is because group delay is defined as the negative derivative of the phase response. And the phase response of high- and lowpass filters have the same kind of shape that is falling with rising frequency and asymptotally reaching a finite value at zero Hz and infinity. With the only difference that the phase response of a lowpass is zero at zero Hz and -90 degrees per order for frequencies approaching infinity while the highpass has +90 degrees of phase shift per order towards zero Hz falling towards zero at infinity.
If the group delay of a high pass was like the one you showed then the design of multiway speakers with simple analog crossovers that can reproduce square waves would be a walk in the park.
Regards
Charles
The point in time when the peak energy is leaving your speaker is not determined by the group-delay alone but also by the spectral content of the input test signal.
When you have a peaking group-delay but no content in the frequency range of the peak then that GD peak is meaningless for the outcome.
Same for group delay rising towards lower frequency. A speaker with a very low cutoff frequency will have higher group delay around its cutoff than one with a higher cutoff frequency. But the one with the lower cutoff frequency will most likely have less group-delay in the lower midrange.
Regards
Charles
Yes that was the point I was trying to make when I said they they were confusing when flipped. I should have left the flipped group delay out. I have been able to make my own graph now to show the high pass group delay correctly for a BW filter of 24 and 48dB to make it easier to see the group delay peaks. The main point being that group delay peaks around the cutoff -3dB point of a filter.
Attachments
For horns with a clearly defined cutoff (such as an exponential or a Le Cléac'h horn), the generally accepted rule of thumb is 1.5 times the cutoff frequency. Paul Klipsch chose a value of 1.5 when designing his K-400 midrange horn (as used in the Klipschorn). The K-400 horn has a cutoff frequency of 270 Hz, and as the name suggests, is designed to be crossed-over at 400 Hz.
It would seem that Jean-Michel perhaps used a more conservative factor of 2 when designing his horns.
Hi camplo,
About 3m away.
Here's no smoothing.
And NO, i don't give a damn about burst decay, or area of excess decay, or etc etc.
I can see performance deteriorates the lower it goes. But we all know and expect that, right..
So I build a set of xovers through the range of measurement deterioration and go listen. That's what i give a damn about
Everything I predicted has come true so far. Based off of electrical impedance to start.
Haha...i always specify if it's an in-room measurement, to let reader beware.
Default is 3m outdoors, off my back deck which opens out to empty space nicely.
True enough, if you ever feel like the bother...a direct energy measurement can be taken without having to move your horn from its current location....just put the mic barely an inch away from the horn mouth, dead center and fire away =)
It gives you a clearer view of what your horn/driver is going.
It gives you a clearer view of what your horn/driver is going.
Measurements for horns should be taken in far-field conditions. So should any speaker, really. Taking near field measurements is only appropriate when they are manually convolved with a majority of far-field data.
Group delay is also derived from phase data, which is influenced by relative distance, reflections and the like.
All of the ‘mic’ type ABEC/Akabak observation types default to a far-field determination which is then compensated to produce the response at the designated measurement point. So it calculates the response at 10 metres, then adjusts the gain via inverse distance law to give the response at 0.1 metres, if that’s where you set the point position.
There’s a good reason for that. It’s also why anechoic rooms are qualified via inverse distance traversal methods, plotting the deviation from theoretical results for different angles and positions of source and receiver.
A good rule of thumb is to measure at 3 or 4x the diagonal distance of the baffle for the device under test.
Group delay is also derived from phase data, which is influenced by relative distance, reflections and the like.
All of the ‘mic’ type ABEC/Akabak observation types default to a far-field determination which is then compensated to produce the response at the designated measurement point. So it calculates the response at 10 metres, then adjusts the gain via inverse distance law to give the response at 0.1 metres, if that’s where you set the point position.
There’s a good reason for that. It’s also why anechoic rooms are qualified via inverse distance traversal methods, plotting the deviation from theoretical results for different angles and positions of source and receiver.
A good rule of thumb is to measure at 3 or 4x the diagonal distance of the baffle for the device under test.
Last edited:
