John, thank you very much for your explanation. This point is now clearer for me.
But as far as the topic of this thread is concerned, I'm at the beginning. I have one measurement which is assumed to be correct, and one impedance measurement that is also supposed to be correct. But both say different things.
But as far as the topic of this thread is concerned, I'm at the beginning. I have one measurement which is assumed to be correct, and one impedance measurement that is also supposed to be correct. But both say different things.
John
Your explaination seems lacking to me. I can accept the "equivalent to a monopole but at a lower level" part at the lower frequencies for a near field measurement, but as some frequency this effect has to transition to a monpole at normal level and your explaination and plots don't do that. In other words above some fequency a near field measuriment of the speaker in a dipole has to be the same as measuing it in an infinite baffle, but not at low frequencies. This transition is what I suspect is where Ms. Raka's data is getting messed up because she has to assume either one model or the other and neither of them will fit through this trsnsition, making the Q and resonance seem incorrect.
Raka
The resonance measurement with the impedance is the mechanical resonance and is correct no matter what model you use. It is even correct in a vacuum, although in the later case the air load mass is missing. Some of the errors could be from this air load mass being different for a monopole and a dipole. If there is an error it is in the SPL measurement and your assumed model of what it should be.
Your explaination seems lacking to me. I can accept the "equivalent to a monopole but at a lower level" part at the lower frequencies for a near field measurement, but as some frequency this effect has to transition to a monpole at normal level and your explaination and plots don't do that. In other words above some fequency a near field measuriment of the speaker in a dipole has to be the same as measuing it in an infinite baffle, but not at low frequencies. This transition is what I suspect is where Ms. Raka's data is getting messed up because she has to assume either one model or the other and neither of them will fit through this trsnsition, making the Q and resonance seem incorrect.
Raka
The resonance measurement with the impedance is the mechanical resonance and is correct no matter what model you use. It is even correct in a vacuum, although in the later case the air load mass is missing. Some of the errors could be from this air load mass being different for a monopole and a dipole. If there is an error it is in the SPL measurement and your assumed model of what it should be.
An externally hosted image should be here but it was not working when we last tested it.
Above shows the on axis result for 1', 1" and 0.1" at higher frequency. You can see that as the mic distance decreases the response degenerates to 0dB relative to a monopole at all frequencies.
These are for omnidirectional sources so the only considerations are the relative strength of the front and rear sources with respect to the mic position and the phase differences. When the mic is 1' from the front source it is 1.5 feet from the rear so the rear has a strength of 0.666 compared to 1. At low frequency where the phase is close to 180 degrees this sums to 0.333 = -9.54 dB. At the dipole peaks where the sources are in phase they sum to 1.6666 = + 4.43 dB with null (180 degree points) at -9.54dB.
When the mic is moved closer to the front source the relative strength change. For 1" they become 1 and 0.143 for -1.34dB and +1.16dB. For 0.1" they become -.143dB and +0.141dB.
If the mic were at infinity then booth sources would have the same relative strength and the dipole peaks would be +6dB, the nulls would be at -infinity and below the dipole peak the response would roll off at -6dB to infinity.
Obviously consideration of driver directionality changes thing once the directionality enters the picture, but this thread was about low frequency response and I was responding to the "correctness" on my near filed/far field method of measuring /constructing far field dipole response. At low frequency the drivers behave omnidirectionally and the approach is reasonable and correct.
gedlee said:
My asumption is that the impedance measurement is giving two poles for the driver itself (not that the impedance is giving the complete picture of the dipole transfer function). After that, with software, I could apply the opposite zeros to those two poles at the measurement at 1.2m, and identify easier the dipole 6dB trend, correct with the opposite zero, and then identify and correct easily the notch.
Is my asumption correct, or the impedance curve is not isolated from the other effects?
gedlee said:
See K&F, 2nd edition, section 10.6 Acoustic Doublet.
Raka said:
Your thinking is correct, however, you have a choice of using the impedance data or fitting a curve with know poles to the measured SPL data. Remember that T/S parameters are small signal and impedance data will change with signal level.
Also consider other means of equlaizing a gradient woofer system.
I would never use that large of a resistor. I use a .1 ohm series resistor and then pad out the voltage measurement by 10 and I have a one to one relationship to the voltage and current, but scaled by a factor of 10 (which helps the sound card). This value of resistor can simply be left in place for all of the measurements. Leaving 8 ohms in place will seriously change the SPL measurements.
john k... said:
John
Unfortunately those are point sources and not a piston in a baffle. They will be different and could be significantly different in some frequency regions. As I said before the dipole circular piston in a circular baffle has been solved exactly and is in JASA (back in the 50's by some Japanese profs). Gerald Lauchle from Penn State (OUR alma mater!) also did his thesis in this area so you might look up his name. It would be interesting to look that up and compare the answers.
Raka,
Do you have a large, open, test baffle that would allow you to take the measurement with nearly the same Fs but different acoustical conditions as a sanity check?
What are the dimensions of your dipole baffle?
Perhaps your not finding the 0 dB level correctly due to contamination from the rear radiation, where you might get some doublet type of interaction. This curve is not for the near field but it might shed some light on the issue:
http://www.linkwitzlab.com/images/graphics/2pt-src2.gif
Do you have a value for D and fequal?
Interesting question.
Pete B.
Do you have a large, open, test baffle that would allow you to take the measurement with nearly the same Fs but different acoustical conditions as a sanity check?
What are the dimensions of your dipole baffle?
Perhaps your not finding the 0 dB level correctly due to contamination from the rear radiation, where you might get some doublet type of interaction. This curve is not for the near field but it might shed some light on the issue:
http://www.linkwitzlab.com/images/graphics/2pt-src2.gif
Do you have a value for D and fequal?
Interesting question.
Pete B.
gedlee said:
Yes Earl, I said that in my second post: "Obviously consideration of driver directionality changes things once the directionality enters the picture, but this thread was about low frequency response and I was responding to the "correctness" on my near filed/far field method of measuring /constructing far field dipole response. At low frequency the drivers behave omnidirectionally and the approach is reasonable and correct."
I don't want to get hung up in all these details that only obscure the approach I use to measure dipole response. There are two issues. One is whether a near field measurement of a driver on an open baffle is representative of the driver's response. The answer is a simple yes because of no other reason than when the mic is placed close to the source the SPL from that source will swamp any sound from any other source. For equal strength sources it is simple a matter of proximity. The relative level of the more distant source will be 20Log (r1/r2) where r1 is the effective distance to the closer source and r2 the effective distance to the further source. As r1 goes to zero the SPL of the distant source goes to -infinity dB relative to the closer source. The second issue is whether a driver on an open baffle behaves as two omnidirectional (or point) sources at low frequency. Again, for ka nominally less than 1 the answer is yes.
In my measurement approach what happens for Ka >L is already contained in the far field measurement and all I need to do is correctly process the near field data for ka<L and merge it with the far field data. L represents the limiting value of ka for the far field data. Typically the far field data will extend well below ka = 1.
This is really not any different that taking a far field measurement of a speaker which extends into the baffle step region and them merging it with a near field low frequency measurement which has been corrected for the baffle step. Except for the dipole it is not necessary to model the baffle step.
If you want to model a driver on an open, flat baffle there are numerous codes around that do that. The problem is that even though these models consider pistonic sources, with real drivers at higher frequency that what comes off the rear side is not at all what comes off the front side. Too many unique factors involved such as cone angle, basket and motor blockage, etc. Fortunately at low frequency these asymmetries disappear for practical considerations.
john k... said:
Thats fine, but just for the record I was talking about near field effects not directional effects - they are completely different things, and they are quite different for infinite baffle and finite baffle situations. Your approach for the LF is most acurate at VLF and that for your HF the same at HF, hence the transition point is where they are both the least accurate. This is always the problem.
gedlee said:
Well here is a measured result. The near filed measurement is as accurate for the dipole as it would be for a sealed box or infinite baffle. In my mind the limitations of near field measurement directionality are really the same thing both arise form the finite size of a pistonic source compared to a point source. It is not a consequence of the dipole format. I presume we agree on that. Limitations on the far field data are only set by the measurement environment. If there is a sufficiently long reflection free window then the far filed data would be accurate well into the frequency range where the near field data is accurate. Given that, it is a simple matter to construct the low frequency far field data from the near field measurement and merge the two in the region of overlap where both are accurate. In any event, the figure below shows such a construction yielding the full range far field SPL. As you can see the constructed dipole low frequency response overlaps the far field result just about exactly over the octave form 160 to 320 Hz, and could be considered acceptable all the way up to 500 Hz. The maximum different is at 400 Hz and amounts to 0.8 dB. The procedure is more ths sufficient for the highest quality design work.
An externally hosted image should be here but it was not working when we last tested it.
Now I'm off sailing for the day.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.