![]() |
Propeller Head Plaza Technical and scientific discussion of amps, cables and other topics. |
Register / Login
|
In Reply to: Re: A correction on one point posted by bwaslo on December 12, 2006 at 05:40:01:
Hi BillAh, I see what you ask.
By “expected†I mean that the slope of the phase reaches 90 degrees times the order, for example, a 2nd order Butterworth filter reaches 180 degrees shift and has a magnitude slope of 12dB/oct.
This part can be referenced with a filter design or spice program, one generates the “theoretically ideal†phase response for that amplitude response.My “reality check†is intended to substitute the speaker, amplifier and microphone with a noiseless, reflection free closed system that has a known magnitude, phase and fixed time delay.
If one uses a BSS366 omnidrive or Behringer ultradrive, (the ones I have used for this test) all one does is use plane Jane Mrs Butterworths filters, nothing fancy, one must mimic a speaker.
This could be implemented with passive parts too if desired (with a time delay added)
The filters only imitate the band limited response of a real speaker but without the noise or extraneous effects.A time delay of say 3 ms is then added to simulate the speaker to mic distance which has to be accounted for to get a real measurement.
What I have seen is that Smaart, MLS as applied in the TEF machine or as in LSPcad (where fwiw, the author also says it doesn’t actually measure acoustic phase), produce a phase error at both ends of the measurement.
As I recall, both the magnitude and phase slopes were incorrect at both the low and high end, showing less lf slope and more hf slope than actual.The TDS process also does not give an exact phase measurement either unless one is aware and thinks ahead. Its potential errors are at the high end.
The band pass filter as used here, actually has 3 different time delays which depend on frequency.
The closed computer model assumes Time = zero.
The ETC one does first with the TEF, identifies the time at / near the hf system cutoff (but not quite at zero time).
To make the displays match (theoretical model of the filters response vs what you measure), one has to set the Time via the ETC for a frequency above the system high cutoff.Like I said, I have not tested every implementation of MLS, just a few.
The discovery actually began with an argument with Sam B (Smaart) on live audio board ages ago. I saw measurements taken of a speaker with the same mic / location / setup done with TDS and Smaart and they (the acoustic phase) were very different.
This prompted me to look for a way to make a closed loop back test that could confirm or deny a measurement systems ability to measure the reality.I am much less concerned about “if†theory says something “is†rather than what one actually measures.
If a measurement system can accurately measure the “simulated†speakers known magnitude and phase response in the closed loop back test, then it works as desired.
Best Regards,Tom
This post is made possible by the generous support of people like you and our sponsors:
Follow Ups
- Re: A correction on one point - tomservo 07:12:46 12/12/06 (1)
- Re: A correction on one point - bwaslo 09:32:05 12/12/06 (0)