You went to see Mama Mia?! And it ended unsuccessfully?! Pano, say it ain't so. Say it ain't so. Stick to Vodka and fruit juice. It's a sure thing.
Dan
Dan
Very sound advice Dan. Only - one has to wonder how many "30 somethings" now among us are the result of too much vodka mixed with ABBA.
I think that was the subject of the play, tho I tried hard not to pay attention.
I think that was the subject of the play, tho I tried hard not to pay attention.
Maybe we should set up a list for music that is
adequate for annoyance testing ...
Abba and Kate Bush e.g. would be on my list definitely.
But i must admit, that i use some recordings
to check a speaker's "tolerance" for imbalanced
recordings.
How would some in here e.g. estimate low and mid
bass content on this CD:
YouTube - Joe Cocker Live - Unchain My Heart
Check this out so you can be a trained listener and recognize exactly where problems lie:
Audio Musings by Sean Olive: Harman's "How to Listen" Listener Training Software Now Available as Beta
Listen to learn to listen. Best of all, it's free.
Dan
Audio Musings by Sean Olive: Harman's "How to Listen" Listener Training Software Now Available as Beta
Listen to learn to listen. Best of all, it's free.
Dan
Interesting article, I think I agree. Just don't tell the author. 😱
My edit bolded:
I haven't thought of "tuning" the off-axis, I've gravitated towards a smooth off-axis, but this goes to baffle design and driver layout before doing any crossover work. And as you know, baffle diffraction control. At this point, I may spend more time with the driver tweaks (tweeters if they benefit) and extensive testing with felt until I'm satisfied with that. Then I go for my first crossover.
Some of us do this I think, but haven't thought of it in those terms nor as a specific task.
Extended listening definitely is important, but I think for a bit more than annoyance avoidance.
That's an area that is not natural for me, picking out a specific frequency or narrow area. I'm just not that good, part of the reason I rely so much on measurements. I'll know the general area, but I use measurements to hone in on the specific problem.
As far as a tilt, what I've found is that smaller tweeters such as 19mm domes sound better with more tilt. A 25mm may need only a small tilt. A 28mm may need none. All to my ears and preference, of course. This probably relates to the DI of the tweeters and their power response.
Dave
Thanks Dan - cool! I downloaded it, looking forward to tying it.
(wow, you'd think someone this tech savy would know how to name a file for Internet use, guess not.)
That's funny and true Pano--but anyways. It is very cool and I'm glad they did that for us.
Much thanks to the boys at Harman International.
Dan
Much thanks to the boys at Harman International.
Dan
Here's what my listening position impulse looks like. http://www.diyaudio.com/forums/multi-way/103872-geddes-waveguides-103.html#post2497065
I realize you're talking about pseudo-anechoic measurements, but I would suggest that maybe the best solution is to go outdoors for a true anechoic environment, at least ground plane or half-space using a pit. You won't have the halfspace/freespace baffle transition, but that's a directivity thing.
I think pseudo-anechoic measurements are useful for things like crossover optimization. Usually crossover is high enough that windowing works well. You can gate out the reflections with plenty of room to spare.
As for floor bounce in home hifi, I propose a couple of other solutions. I see that Dave Smith mentioned these kinds of techniques earlier in this thread, but I think they're worth mentioning again. See the following whitepaper:
Pages ~23 onward talk about floor bounce and a couple of solutions. Basically, you have two choices - either keep the sound source within 1/4λ of the boundaries or blend sound sources arrayed so that the self-interference notch from one is "filled in" by the energy from another.
There are competing priorities in that you want the main speakers to be at ear level, but the lower midrange will suffer a deep notch from floor bounce when a point source midrange is used. You can blend it with a woofer though, similar to the way a multisub configuration can be used to smooth the bass in the rooms modal region. I suppose in some sense, the floor and ceiling bounce notches are room modes, just at the high end.
😀
-the silent screams of: "..oh God, make it stop! Please just make it stop!"
..it could have been worse you know. I distinctly remember ABBA playing on tinny ceiling speakers as I was fully reclined - and getting a full bore root-canal.
Last edited:
but if You like to have seriously professional golden ears You have to pay, $$$ for seriously professional package of total 8 CDs + Manual
David Moulton knows how to listen 😎
Moulton Laboratories :: Golden Ears
Not quite clear which post in that thread you're drawing my attention to. Are you referring to the impulse response ?
It's difficult to judge since there's no precise time scale shown in the screen shot, but there look to be at least 3 distinct reflections arriving before ~3ms, with the first at something like ~2ms, which is fairly consistent with what I'd expect from a room measurement taken at some distance.
The maximum gate length you could use at that distance and still get an on axis response that was free of room effects would be to the end of that first flat section, but with such a short gate time you wouldn't get any accuracy below about 1-2Khz.
Otherwise you're measuring some or all of the room response, and then you're back into the whole area of "flat in room steady state response is not right" that's been rehashed dozens of times already in this thread.
Personally I don't think it's possible to measure the free field on axis response in room down to 200Hz with the required accuracy to determine whether you actually do have a flat or -1dB response response in the treble, regardless of any amount of gating.
Outdoors, high up off the ground, with a significantly longer gate time would be the only way for us mere mortals to do it without an anechoic chamber 🙂
Why not? If the reflection free time is long enough and the gating time is set accordingly then you're looking at free field data.
The dual gated measurements are a visual guide to the progress during the procedure. Of course I've looked at it with various gate locations and observed the patterns and trends.
The crossover tweaks were being simmed along the way in Speaker Workshop using power response data taken with a 5ms reflection free time.
The crossover tweaks were being simmed along the way in Speaker Workshop using power response data taken with a 5ms reflection free time.
I agree outdoors would be the only accurate way to measure it other than an anechoic chamber.
The problem with a ground plane pit type measurement though, (apart from the fact that I think AlanB's speakers might be horns, and thus not necessarily have a front baffle that could line up with the edge of a hole in the ground 🙂 ) is it still doesn't answer the question of whether the speaker is anechoic flat on axis.
It's easy to say you "won't have the halfspace/free space baffle transition" and that it's just "a directivity thing", but this transition and the ripple that comes with anything other than a spherical enclosure all contributes to the on axis response, and it's not easily modelled to sufficient accuracy.
It's not just a matter of of adding a 6dB shelf to the measurement, because the diffraction introduces considerable ripple and overshoot dependant on driver placement.
Although software exists to model the effect in more detail than a simple shelf, most have a very simplified physical model which treats the driver and panel as a 2 dimensional plane (depth of driver isn't modelled) and doesn't model the driver directivity accurately either - and to some extent it can't, because you can't just assume the directivity of an ideal piston.
Because increasing driver directivity tends to make baffle diffraction problems less serious such software generally overestimates the severity of the peaks and dips in the response. It's useful for predicting at what frequencies you're likely to see a peak or a dip, but not that useful at accurately predicting the magnitude of the peaks and dips, or whether they will actually occur, once you reach the region where the driver becomes directional...
If you can't accurately predict the shift in response between an in-ground infinite baffle measurement and the baffled free space on-axis response, then your final result will be very inaccurate due to inaccuracies of the diffraction model...
A much better approach would be to put the speaker upright on top of a platform outdoors, as high as reasonably possible, (say 1.5m) which allows you to take a gated measurement with a much longer gate. For example a floor standing speaker with the driver 90cm off the floor would put the driver 2.1m off the ground - if the microphone was 2 metres away that would give you 9.3ms until the first reflection which would go down to 200Hz or so accurately, and you could then splice that to a near-field measurement for the bass if desired.
In a two way design where you only have to worry about a crossover in the tweeter region gated can give good results in room, but trying to measure a 300Hz bass/midrange crossover on a larger speaker isn't going to be successful as not only can you not get a long enough gate time before first reflection, trying to splice a near-field measurement at ~200Hz is very problematic if you have both the woofer and a midrange driver to measure and combine in your near-field measurement. (And possibly a port response too)
There is still enough output from the midrange driver at 200Hz and below that only measuring the near-field response of the woofer (and port) would be inaccurate.
Likewise the gated measurement on a large 3 way will be inaccurate if you are too close to the speaker, due to the different driver sizes not reaching their far-field amplitude balance. (The smaller driver will have exaggerated SPL compared to the larger one when measured too close)
Yep I agree, and described both these approaches a few pages back 🙂
http://www.diyaudio.com/forums/multi-way/172806-flat-not-correct-stereo-system-16.html#post2501620
Technically, the woofer doesn't have to be within 1/4 wavelength of the floor, that only matters for maintaining floor/boundary gain through to higher frequencies. The floor bounce notch is caused by the different distance of the reflected signal from direct, and it's frequency will vary with listening distance even if the woofer stays the same height off the floor.
Thanks for the link to that article BTW.
I don't think floor bounce cancellation is the same as a room node at all. It's a boundary cancellation effect, due to a delayed reflection arriving 180 degrees out of phase.
There are fundamental differences between room modes and boundary cancellation/addition:
* Room modes can only occur when there are opposing surfaces like opposite walls present for a wave to bounce back and forth repeatedly between. Boundary cancellation can occur even if just one wall is present and only one reflection can occur.
* Room modes occur at frequencies dictated by the dimensions of the room, (distance between opposing reflective surfaces) not the speaker placement. Moving a speaker along the axis of a room mode does not change the frequency of the room mode, it only changes the amplitude of it, by controlling how strongly it's excited.
* Boundary cancellations (and reinforcement) on the other hand occur at a frequency which is dependent on the distance of the driver from the boundary. Moving a speaker towards or away from a boundary will change the frequency at which the peak/notch occurs, whilst the change in amplitude is fairly small - gradually reducing with distance.
Although both cause peaks and dips at the listening position they're easily told apart by moving the speaker slightly, as a room mode wont change frequency whereas a boundary cancellation or reinforcement effect will 🙂
Big if. 🙂
No matter how wide or long your room is, if you have a common 2.4 metre (8 foot) high ceiling, the furtherest you could ever get a driver away from the nearest boundary is 1.2 metres.
If your mic is one metre from the driver, at most you can get a window size just under 4.65ms. Unless you have a high vaulted ceiling (or warehouse) and a stand to increase the height of your speaker so the driver is at half ceiling height, this is the best you can hope to do indoors.
This is pretty marginal for going down to 200Hz. Yes, in theory it will, but you'll get a lot of "ripple" and smoothing in the low end response as the information isn't really there, there just aren't enough samples to provide any fine detail at that low frequency.
The problem with gating is that it doesn't just let you measure perfectly down to a certain frequency and then stop abruptly - it starts getting significantly inaccurate for at least an octave above the theoretical minimum cutoff frequency for a given gate time due to the shape of the windowing function and also basic frequency/time uncertainty principles. To get good accuracy you need a longer gate time than you think you do.
Measuring at 1 metre is ok for a single driver but for a 2 way system you'd really want to measure at 2 metres (reducing your maximum gate time to 3.27ms) and for a large floor standing 3 way system it would not be fully integrated at 2 metres, and ideally would need measuring at 3 metres - almost impossible to do indoors in a living room.
Don't get me wrong, gated measurements are a useful tool, but they're not a magic poor-man's anechoic chamber, certainly not in the size of room that most of us have available to use.
Last edited:
So, not horns then 😉
Must have got the wrong idea when I was skimming the other thread you linked to.
A 4.65ms window time on ARTA shows a time/bandwidth limitation of 215Hz. It does depend a little bit on the window function used, and as I mentioned the accuracy in the lowest octave (in this case from 215 to 430Hz) can be considered to be suspect, or at the least very "smoothed".
- Status
- Not open for further replies.