Welcome! Need support, you got it. Or share your ideas and experiences.
Return to Planar Speaker Asylum
32.212.76.154
In Reply to: RE: Using REW with Maggies posted by computerman on October 10, 2015 at 09:00:13
One factor is that line source planars are hard to measure because not only do they interact strongly with the room, the frequency response changes as a function of distance -- so you'll get very different results if you try to isolate the speakers from the room with a 1 meter gated measurement, particularly a huge and non-existent at the listening position rise in the bass, but also some artifacts higher up. Ultimately, the only measurements that matter are what reach your ears in your room. (That's true of any speaker, but truer of planars!)
However, there are apparently arguments for measurements of the drivers if you're doing an active crossover, particularly DSP. You'll get a much more accurate measurement of on-axis amplitude and phase response if you measure outside. (You won't be able to measure power response outside unless you're prepared to take a lot of readings from different angles.) This can help with crossover tweaking and I believe it's the recommended procedure for the DEQX. You could use this curve to create a complementary frequency/phase curve, then convolve the impulse for driver correction. Not only would this regularize amplitude and phase response, it would help match the units for a more precise stereo image (some smaller Maggies have asymmetrical resonances in the woofers, I'd leave those alone). After that, you'd do room correction using an in-room measurement, making the measurements within your listening area (not in a larger area, as is sometimes recommended). That's where you'll correct for power response and room effects.
Necessary? Practical? I honestly don't know -- DEQX does it that way (or at least did last I checked), but there are lots of successful room EQ boxes/programs and not all of them do.
Follow Ups:
As you know, Siegfried Linkwitz commented on your first point regarding inaccuracies of nearfield measurements of planar speakers in the original Stereophile review of the Magnepan 3.6s.You have an interesting theoretical idea, but I think the key, as you said, is measurements at the listening position. Everybody has their own preferences, but I don't find the 3.6's on-axis to yield the best sonic results, especially if you can use room treatment and get the speakers far enough away from the front wall (68" is the minimum distance for a 10mS delay of direct to reflected sound).
My measurements were with tweets in and both speakers toed-in only 3.25", not pointed at the listener. It helps the speakers disappear, provides a big sound stage, and gives the ideal amount of high frequency roll off to sound natural. To get a solid centered image with this placement, you must place absorption some distance symmetrically behind, and somewhat to the outside of the panels. This absorbs some of the increase in reflected sound. Again, I'm sure others have their own approach, but this measures well and sounds great.
For a great airy, coherent soundstage, get both speakers the same exact distance to the listening position (get within 1/10" with a laser distance meter). You can actually hear perfectly timed reverberation from the sounds coming from one channel in the other speaker.
Edits: 10/10/15
Yes, in fact, that discussion of the measurement issue in Stereophile was when I first became aware of it.
I've noticed the importance of precise distance to the speakers but was never able to adjust so precisely, I think your suggestion of a laser distance meter is an excellent one and I plan to try it myself. Though since I'll be using a digital crossover I should be able to adjust the time delay instead!
I've also noticed that in some setups, putting the speakers parallel to the front wall improves the rendering of space. But my sense is that optimal speaker orientation will always be somewhat idiosyncratic because of different driver and crossover characteristics, room interactions. and listener preferences. In practice, you're going to hear not just the crossover lobes in front, but from the early reflections, as well as the overall power response. Also, you may (or may not) want to optimize the dipole nulls at the first reflection points, lateral or a compromise between lateral and rear. So what you hear is going to be very room, setup, and speaker dependent.
Both my 1-D's and my MMG's only sounded right on axis because the tweeters were beamy, whereas the true ribbons in the 3.6's have better HF dispersion and should give you more leeway. So with the 3.6's, I think you'd be primarily concerned with the direct and reflected crossover lobes. With a digital crossover you can alter the timing of the drivers and that would allow you to put the dipole null where you want it while independently optimizing lobing. With the IVa's, I should also be able to keep the woofers oriented for minimal excitation of lateral modes while toeing in the MT panels as appropriate. The digital crossover should allow me to eliminate the time delay in split configuration and also get the angle of the XO lobes right. I think diffusers would be another way to make lobing more manageable, since they delay the reflections which allows for more randomization before they arrive at your ears.
Finally, re the original question, I'd have to say that I agree that what happens in the room is ultimately most important. But I gather that there's an advantage to measuring and correcting the speakers first separately, provided that they're measured at the correct distance for line sources, since this is the procedure that DEQX and some speaker/room correction crossover software companies recommend. I was trying to figure out why that might be so but I'm not at all confident that my speculations were correct.
to use an REW group delay plot to adjust the DSP filters in DEQX or mini-DSP?
Sure, check out Rephase for a free example:
http://www.minidsp.com/applications/advanced-tools/rephase-fir-tool
Note that you'll have to use an FIR-capable Mini DSP.
DEQX can also do phase linearization, though I assume they use their own analysis software, not sure what the story is with importing filters.
With a convolver, you can also use a correction filter on a PC or Mac.
Thanks for the links. Very helpful. After a few hours of reading I now understand that FIR filters can correct phase but not group delay. I need a little more education. In any event I have confirmed that miniDSP with add-ons has much more functionality than what I am using.
Keep in mind that to get to the FIR filter level, you need to go up-line at Mini.
I tried reading the WIKI on FIR filters and my brain turned to MUSH after the first sentence.
Too much is never enough
http://www.prosoundweb.com/article/fir-ward_thinking_examining_finite_impulse_response_filtering_in_sound_rein/av/P1/
It would appear to be accurate.
This is a complex subject. I'm sure the math is helpful but most of it eludes me. Just because someone is an expert doesn't mean they're capable of explaining something.
I've TAUGHT technical stuff. Some people are seemingly incapable of absorbing detailed, scientific information. That's one reason I have trouble reading a lot of 'Advertising Babble' about technical stuff.
I'd LOVE an explanation of FIR filters. But even THAN? I don't know that I'd want or NEED to try to get the gear to impliment.
Too much is never enough
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: