Well, all of that ends up influencing and being displayed in .... the frequency response
To make it even more clear: changing any of the factors you mentioned will change frequency response.
To be even more precise: you do NOT hear by themselves, but the resulting ... frequency response ....
That is not quite true. The frequency response (or more precisely the magnitude response or power spectrum as being discussed here) is simply a mathematical transformation of some measure (in this case sound pressure). This particular transformation removes phase information and disperses the harmonic (distortion) components throughout its extent - hence they are not observable. We should remember that the "frequency response" is simply an analytic tool that is extremely useful within certain bounds. Our hearing and our perception demand extending discussion beyond those bounds if we are to characterise completely the performance of a driver or complete loudspeaker.
soundbloke;5699844 Our hearing and our perception demand extending discussion beyond those bounds if we are [B said:
What prevents? Now it is not the 50s of the last century - there are many good measuring programs, and there is a fantasy based on knowledge.
The amplitude measurement is the tip of the iceberg, followed by the measurement of the phase response, the group delay, the impulse response, the rate of increase of the signal, measurement and analysis of distortion in frequency and input power, impedance and its phase response, etc.
What's the purpose of this thread? Isn't it one of those pointless discussion about flattnes of maximum torque of the engine influence on how well driver can drive a car in the corners?
Being more specific. Isn't it a meritless to discuss about general importance on end result of one property of one component which is isolated from all the system and isolated from its working environment?
Flat axial frequency response is only one meaningless line on one meaningless graph. In order to make the system sounding good there is a need for something called systemic design, where designer need to balance many many aspects like said frequency response in accordance of predicted working environment of end product.
Being more specific. Isn't it a meritless to discuss about general importance on end result of one property of one component which is isolated from all the system and isolated from its working environment?
Flat axial frequency response is only one meaningless line on one meaningless graph. In order to make the system sounding good there is a need for something called systemic design, where designer need to balance many many aspects like said frequency response in accordance of predicted working environment of end product.
Well, all of that ends up influencing and being displayed in .... the frequency response
To make it even more clear: changing any of the factors you mentioned will change frequency response.
To be even more precise: you do NOT hear by themselves, but the resulting ... frequency response ....
Indeed. But I would like to add one thing. And that is what we can change and what we cant change.
Correct me if I'm wrong. But I have had great succes, by accepting that if I correct my drivers individually by my DSP, according to a fixed gated measurement. Then I have a bigger chance of succes, because I do not try to correct something a filter cant correct.
I understand from Earl Geddes, that a filter is in 1 dimension, and the sound is in 3D, as soon as it leaves the speakers and get mixed with the reflections. So in 1 dimension, the "distortion" is linear and I can correct it with my DSP - like breakup notes in hardcone drivers. If I add the "room", the problem is in 3D and I cant correct it by filter - I then need to change drivers, cabinet, position and so on.
Since a room is almost never without reflections, I see little point in ignoring the off-axis measurements. And I always measure off-axis to find out if I have resonanses or problems with the reflected sound.
Is anything I write misunderstood or unlogically for DIY speaker builders?
Last edited:
I'm having a bit of a revelation here... Using a DAC where it is possible to introduce EQ biquads in a almost lossless manner, I have tuned my FR to quite tight tolerances, measured in my listening position. I do this by making a circular motion (40 cm dia.) with the mic pointing in the same direction as my nose playing pink noise in both channels. I use REW RTA with 4x averaging. I have chosen the old B&K target curve as my goal. As I'm approaching +/-1dB over most of the range I'm beginning to get reproduction that really outperform anything I have had before. The system was well performing before but now it's really magic. If you have a good start I would recommend to try to make it "flat" - it appears to effect a lot of aspects of the sound - all it seems
So for me "flat" means maybe "even" and it seem to be a requirement for accurate reproduction - but it's most probably not the only requirement.
//
PS. The three graphs are from the one and same measurement. Still, a little more energy en the lower region is needed
i have seened that b&k curve before but have yet not figured out if the response is measured for a stereo pair one at a time or both at the same time?
Yes, since drivers are minimum phase a great deal can be achieved by using filters to flatten the response within the passband.
Absolutely nothing prevents further analysis. The thread starting by asking if a flat frequency response is all that matters in driver measurements: It is not. Clearly conclusions from the use of any measure are only of relevance where that measure is properly understood.
Last edited:
No, the principles are long established, just often forgotten or overlooked. Simplicity is great, but the ease of measuring and FFT processing provides ample chance to misappropriate reason. Ruler flat frequency responses, for example, whilst an admirable target, should (IMHO) always be treated with caution when encountered in reality
The frequency response will tell us it's phase response and impulse response, what other measurements matter?
do you guys only listen to fixed frequency sine waves at 2.83V with your speakers, all the day?!
no, i dont, the music in my world are complex signal of mixed frequencies.
there is no way for an FR plot to reveal the content/amount of perceived sound distortion under such complicated real life condition.
if FR is the final answer what all you need is just a simple SPL meter, whats the point for the industry to develop all that crazy expensive audio analyzers? could they be idiots?
no, i dont, the music in my world are complex signal of mixed frequencies.
there is no way for an FR plot to reveal the content/amount of perceived sound distortion under such complicated real life condition.
if FR is the final answer what all you need is just a simple SPL meter, whats the point for the industry to develop all that crazy expensive audio analyzers? could they be idiots?
Hi Scott, frequency response alone can't tell you phase or impulse.
For instance, you can have flat freq response with well constructed crossovers of different orders, and they will have different phase traces and impulses.
edit: If you are referring to an individual driver only, pls ignore
Last edited:
The complex frequency response is not what is referred to by the "flat response" at the begiining of the thread: That comment applied only to the magnitude response. But even for a single driver (which as you point out is likely to be minimum phase), the complex frequency response is still insufficient to describe its audible performance. Distortion evident as harmonics will be dispersed throughout the measurement and the assumption of linearity is generally not sufficient. Distortion may also not vary simply as a function of power output. Those looking for a single characteristic of "sound quality" need look a lot deeper into the information their measurements contain: It is not an impossible task, but a flat magnitude response is certainly not where the task ends.
- Status
- Not open for further replies.