After your measurements and listenings, is this still a question ?
For me, many signs seem to validate the fact that averaging is closer to perception.
There should be no difference with enough averaging, but MMM is quicker.
Enough measurements would mean about 30 independant measurements to keep within +-1dB at 300Hz (with 95% confidence), that takes time...
Last edited:
It is - and is one of the biggest problems we face. If a room rings for 3 seconds at 400Hz, what can you do? You can pull it out of the mix to not excite the room, but then the direct sound is missing 400Hz. It becomes a subjective balance of what works best. It will never be good, but we do what we can to make it suck less.
Agree, very much so. In my recent tries at this the mic was kept within the space just above my chair. That's were I sit. Making the space as small as my head led to slightly different results, but only slight. I should check that again.
You seem to be implying a narrow band of very low damping that does not show up as a strong steady state response peak If this interpretation is right, then do you have any examples of this? To me it violates the physics.
Then its not really what Dave was referring to in
Or am I missing something.
If the response "rings on" then it has to have a strong effect on the steady state response. You can't have a situation where something "rings on" but does not increase the steady state measurement. This thread is about steady state measurements so I have to assume that's what we are talking about.
Or am I missing something.
If the response "rings on" then it has to have a strong effect on the steady state response. You can't have a situation where something "rings on" but does not increase the steady state measurement. This thread is about steady state measurements so I have to assume that's what we are talking about.
It may violate physics but it doesn't violate electronics.
I assume you are thinking in terms of simple systems with a single path through serial blocks. Clearly with any system with multiple parallel blocks you can have a very high Q on a parallel path at any gain relative to the main path. Researchers refer to resonance dilution and have done interesting tests where a flat response path is paralleled by a second order variable Q, variable gain bandpass.
This is a good way to look at cabinet resonances: Q can be quite high but the output can be relatively low. The indication in the steady state response will be subtle, possibly overlooked, but the resonance can ring on long enough to be quite obvious. Rooms can react in the same way, so yes, it is possible to have high Q resonances that aren't obvious in the steady state response.
This is also why we use waterfall plots to see submerged resonances.
David S.
Out of curiosity, could you post the ETC and waterfall too, please. Also, FR with both speakers playing the sweep in mono.
Thanks
Whilst the MMM technique does not capture phase, and Jean Luc has said it is not suitable for designing crossovers, I am going to go out on a limb, and give it a try.
I've been struggling to get measurements good enough to design my active crossover (at 200Hz) to my satisfaction. It seems to me that I should be able to take measurements of the Woofer cabinets with MMM, the main speaker cabinets with MMM, and then BOTH running with MMM, and use Jeff bagby's method of extracting minimum phase from the measurements and then working out the offset in PCD by tweaking until the combined response of the separate woofer and main measurements matches the measurement of both playing.
My active crossover can be derived from a passive correction network so all I then would do is model in PCD to get the response that I want.
Does this seem plausible or am I missing something? I think I will have to give it a shot. I've somewhat stalled since I have not had measurements that I have been 100% happy with.
Tony.
I've been struggling to get measurements good enough to design my active crossover (at 200Hz) to my satisfaction. It seems to me that I should be able to take measurements of the Woofer cabinets with MMM, the main speaker cabinets with MMM, and then BOTH running with MMM, and use Jeff bagby's method of extracting minimum phase from the measurements and then working out the offset in PCD by tweaking until the combined response of the separate woofer and main measurements matches the measurement of both playing.
My active crossover can be derived from a passive correction network so all I then would do is model in PCD to get the response that I want.
Does this seem plausible or am I missing something? I think I will have to give it a shot. I've somewhat stalled since I have not had measurements that I have been 100% happy with.
Tony.
It should be fine to get the minimun phase response from MMM measurement on each driver, because a driver with its filter is mostly min phase and MMM gets rid of most of the room excess phase. Two or more drivers with crossovers are not anymore min phase near crossover frequency so you'll have to tweak a bit to get the best offset.
I never did it this way. Let us know...
Note he's talking about a 200Hz xover there.
edit: P.S. if PCD is the software of choice, it already has active xover simulation on the right side of the spreadsheet - no need to take the passive transfer functions or anything like that. Also, I'm not withholding advice to be coy - I've never had to try such a thing with home measurements in that freq. range (I'm assuming we're talking a custom analog active working with some driver rolloff in the xover range, not DSP?).
edit: P.S. if PCD is the software of choice, it already has active xover simulation on the right side of the spreadsheet - no need to take the passive transfer functions or anything like that. Also, I'm not withholding advice to be coy - I've never had to try such a thing with home measurements in that freq. range (I'm assuming we're talking a custom analog active working with some driver rolloff in the xover range, not DSP?).
Last edited:
How would you completely eliminate those ? Anechoic, yes,... but in room ?
Why would you totally eliminate those reflections ? For R&D purposes, filter study, etc...it certainly helps.
But for in-room measurements ? There is still a debate on what is really perceived : mostly the direct sound for mid/high frequencies, but you know better than anybody, that directivity/power response/room is also part of what is perceived. so you have to keep "some room" into the measurement.
I've not heard yet of a measurement method that totally corresponds to perception, the ideal objectiv measurement of subjectiv quality.
Isn't that what Harman did in their studies?
Audio Musings by Sean Olive: Part 3 - Relationship between Loudspeaker Measurements and Listener Preferences
OK, I missed the 200 Hz crossover point. Yes, in a typical room it would not be possible to eliminate the room at that frequency and using MMM might be very useful in this situation.
The room enters the picture in a couple of ways; the early reflections interact with imaging and one needs to know what the direct sound is in the primary frequency range of imaging, i.e. > 700 Hz. So you would not want any room reflections here if you wanted to setup your loudspeakers because it is the direct sound that matters most. In the modal region, below typically 200 Hz. the room dominate the perception and you must have in-room measurements and EQ for these frequencies. The region >200 Hz and < 700 Hz is not really very well documented or understood, but it is my opinion that we are pretty tolerant of things in this region and so the use of direct sound or room sound is probably not critical.
I absolutely do believe that EQing a sound system > 700 Hz based on in-room measurements is absolutely wrong however. The speakers dominate in this region and the speakers anechoic response needs to be right, not the room response. You can "fix" bad speakers in a room with EQ in this frequency region, but you cannot improve on speakers that have a near flat direct sound and near flat Directivty Index with EQ in this region.
I think that the Harman approach works pretty well and it does not use any in-room measurements, although it did use in-room measurements to find what it calls the "listener window". However this window could easily be assumed just on logical grounds.
The room enters the picture in a couple of ways; the early reflections interact with imaging and one needs to know what the direct sound is in the primary frequency range of imaging, i.e. > 700 Hz. So you would not want any room reflections here if you wanted to setup your loudspeakers because it is the direct sound that matters most. In the modal region, below typically 200 Hz. the room dominate the perception and you must have in-room measurements and EQ for these frequencies. The region >200 Hz and < 700 Hz is not really very well documented or understood, but it is my opinion that we are pretty tolerant of things in this region and so the use of direct sound or room sound is probably not critical.
I absolutely do believe that EQing a sound system > 700 Hz based on in-room measurements is absolutely wrong however. The speakers dominate in this region and the speakers anechoic response needs to be right, not the room response. You can "fix" bad speakers in a room with EQ in this frequency region, but you cannot improve on speakers that have a near flat direct sound and near flat Directivty Index with EQ in this region.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.