This morning I was doing some work around the house, and noticed that my Summas sounded better than usual. It turns out one of them was disconnected - I was listening in mono.
I would guess they sounded better for a few reasons -
Anyways, just something to consider. A *single* speaker might be an excellent solution for "common" areas where you want a good speaker, but you'll never be anywhere near "the sweet spot."
For instance, what's the point of having stereo when you're working in the garage, or in the yard? (I've long wondered why boom boxes and Bose wave radios are in stereo... it's not necessary IMHO.)
I would guess they sounded better for a few reasons -
- I was doing errands around the house, and wasn't even *close* to the sweet spot. Heck, I wasn't even on the same floor at times.
- These monsters can easily get 20dB louder than your average speaker without breaking a sweat.
- There seemed to be an improvement in intelligibility and clarity, which I would attribute to a lack of comb filtering due to the absence of a second signal.
Anyways, just something to consider. A *single* speaker might be an excellent solution for "common" areas where you want a good speaker, but you'll never be anywhere near "the sweet spot."
For instance, what's the point of having stereo when you're working in the garage, or in the yard? (I've long wondered why boom boxes and Bose wave radios are in stereo... it's not necessary IMHO.)
I listened to everything in mono for about 14 years and in the last 2 years I returned to stereo speakers again. Which is I'm reading DIY looking for ideas, and while not perfect, a lot of Geddes thinking makes a lot of sense.
Summing stereo signals into mono you lose all the flangy, "stereo" effects added in studio. But since most voices are recorded as 2 channel mono, listening in mono gets rid of all the intractable problems (like your head) of trying to get mono out of any 2 stereo speakers. So voices sounded more real to me.
It's always seemed funny to me that every sound in nature is mono and we listen to stereo recordings (Yeah, we do have stereo ears). Location, location, location, is a good mantra for real estate. Location is the last thing I care about listening to music live. The music, the mono sounds, is what matters. But that's just me.
My mono system had 7 drivers (Focal, ATC, Stage Accompany, Volt) in it, lumpy FR was from 30hz to 15khz using 1st order xovers. The "soundstage" had depth, no width.
Wide bandwidth mono recordings sounded best and most real. Most top 40 hits sounded lame because of all the stereo effects missing.
My son and girl friend love listening to movies and music in stereo and I think it is a lot more fun, too. Not as realistic sounding, but more fun.
Sometimes I play only one side to hear a clearer, cleaner sound.
++++Patrick-- Have you ever played a sine wave sweep on the Summa's (from 100hz to 15k or so)?
Did the stereo image always stay focused, centered in one place between the speakers? Or did it wander?
Hey Dr. Geddes, Would the use of a 3-axis accelerometer be possible a better and more sophisticated method of analyzing the vibrational characteristics of an enclosure and then using it to delineate between vibrations transfered through the speaker basket, into the enclosure, and those as a result of vibrations transfered from the cone through the air and back into the walls.
Is the change in speaker enclosure material for your Abbey and Nathan "a" versions a sign that you believe its offering a sonic improvement, or was it purely for ease of painting?
Now that you are using a cast enclosure, are the round-overs on the front made of cast polymer, or are they still routed? Have you increased the round-over now that you can?
Did you ever end up looking at using that BMS driver instead of the B&C (not the coaxial, the one with better high frequency response), and if not would you ever consider offering a tweaked crossover for that tweeter as an option?
Do you think that switching to even steeper or more complex crossovers could offer further improvements in sound, even if they are small (without regard to performance to cost), and will this be worth pursuing in the future. I know you found that switching to a 3rd order for the woofer in the Abbey improved things, and I know Harman seemed to have come to a similar conclusion eventually going with 4th order crossovers in their own designs. The notch filter in the tweeter offers a very steep roll-off at it's low end of the pass-band, but the woofer has nothing like that. But then, does it need a steep rolloff?
While I'm in my move transition and can't listen to my Abbey's I have to keep my self occupied, so please excuse this somewhat pointless brain picking.
Is the change in speaker enclosure material for your Abbey and Nathan "a" versions a sign that you believe its offering a sonic improvement, or was it purely for ease of painting?
Now that you are using a cast enclosure, are the round-overs on the front made of cast polymer, or are they still routed? Have you increased the round-over now that you can?
Did you ever end up looking at using that BMS driver instead of the B&C (not the coaxial, the one with better high frequency response), and if not would you ever consider offering a tweaked crossover for that tweeter as an option?
Do you think that switching to even steeper or more complex crossovers could offer further improvements in sound, even if they are small (without regard to performance to cost), and will this be worth pursuing in the future. I know you found that switching to a 3rd order for the woofer in the Abbey improved things, and I know Harman seemed to have come to a similar conclusion eventually going with 4th order crossovers in their own designs. The notch filter in the tweeter offers a very steep roll-off at it's low end of the pass-band, but the woofer has nothing like that. But then, does it need a steep rolloff?
While I'm in my move transition and can't listen to my Abbey's I have to keep my self occupied, so please excuse this somewhat pointless brain picking.
1) Yes 3 axis is more information than one or two, but what to do with that information. I have not been able to correlate cabinet vibrations with sound at the listening position or even 1 meter away, nor have I been able to detect cabinet vibration in listening (once the most agrerious ones have been dealt with). In general if you can see the effect in the far field response then you can hear it. But if you cannot detect the cabinet vibrations in the far field data then it does not appear to be a problem. There are no cabinet vibrations maesurable in the far field with my speakers. Or none that I have found.
2) the new cabinet material was almost entirely necessitated by the finish. MDF was hard to paint and not stable over time. The new cabinets will be a lot more rigid, but as I said, since I didn't see cabinet vibrations as a problem I don't see this change making much audible difference. The new material is sheet stock, not cast (not by me, but cast when made). So the corners are still routed. But since the cabinets are glued together the routes are perfect and they stay that way. MDF moves and cracks or lines appear over time.
3) I looked at BMS's prices, no need to go beyond that. I could only do all of the work for a different driver if someone paid for that work to be done. I don't see that happening.
4) I did switch to a steeper crossover on the Abbey for the 12a model and it did make quite a bit of difference. I may switch the Nathan as well but thats TBD when I actually get data. Woofers do tend to need steep rolloff because of bleed through above the crossover from cone modes. This can be a real problem.
2) the new cabinet material was almost entirely necessitated by the finish. MDF was hard to paint and not stable over time. The new cabinets will be a lot more rigid, but as I said, since I didn't see cabinet vibrations as a problem I don't see this change making much audible difference. The new material is sheet stock, not cast (not by me, but cast when made). So the corners are still routed. But since the cabinets are glued together the routes are perfect and they stay that way. MDF moves and cracks or lines appear over time.
3) I looked at BMS's prices, no need to go beyond that. I could only do all of the work for a different driver if someone paid for that work to be done. I don't see that happening.
4) I did switch to a steeper crossover on the Abbey for the 12a model and it did make quite a bit of difference. I may switch the Nathan as well but thats TBD when I actually get data. Woofers do tend to need steep rolloff because of bleed through above the crossover from cone modes. This can be a real problem.
Last edited:
An audible one? Or was it the 1-2dB differences that you had illustrated earlier? Did you change it again?
Anand.
Anand
What you have is the latest. All I have is measurement results. As you know I don't put much stock is listening tests.
Matt
It is a better design, but to change what you have gets expensive because you have to change almost all of the values. Only a little bit, but all those new parts gets expensive.
What you have is the latest. All I have is measurement results. As you know I don't put much stock is listening tests.
Matt
It is a better design, but to change what you have gets expensive because you have to change almost all of the values. Only a little bit, but all those new parts gets expensive.
Speaking to point #2 I see exactly what you mean about finishing the mdf Earl, I just completed my first diy subwoofer build yesterday and painting it was a pain in the a$$. I am not happy with the finish at all and will probably end up going with a laminate after all the work of priming and painting. Thank goodness it sounds good. I had a chance to look at the new cabinet material when I listened to Dr Geddes' speakers and it even LOOKS easier to put a finish on than the mdf, so here is my question Earl. Where can I purchase this material and what is the cost as I plan on building at least 1 more sub?
I get mine from Freeman Supply in Aton, Ohio. Check their web site. I bought out what they had in 3/4", but they should get more. They apparantly have a local warehouse, but they only stock what I use at Aton and I bought it all. Let me know as I may be able to part with some of what I have. But its very expensive stuff - > $300 a 4X8' sheet, comparred to MDF at what about $30. And you have to use a special polyurethane glue, nails won't hold, although screws would work. But without glue the joints won't be rigid.
I did some tests on a glued joint and I couldn't break it apart even with a heavy hammer. Its basically like a weld - the joint is stronger than the surounding material. And unlike wood, the glue joint have an identical density of the glue as the board, So it sands perfectly with absolutely no visible joint (except color). This is precisely what the stuff was developed for. It is used to glue together models for making molds out of. The bonds, edges, surfaces, etc. are all flawless. No wood product can do that.
I did some tests on a glued joint and I couldn't break it apart even with a heavy hammer. Its basically like a weld - the joint is stronger than the surounding material. And unlike wood, the glue joint have an identical density of the glue as the board, So it sands perfectly with absolutely no visible joint (except color). This is precisely what the stuff was developed for. It is used to glue together models for making molds out of. The bonds, edges, surfaces, etc. are all flawless. No wood product can do that.
Last edited:
Hey Dr. Geddes, Well I'm now all setup in my new house, with the theater setup in the living room as best I could for now. I have some work to do for sure. The measurements showed a few things that I now have questions about.
First, When I take a measurement with Square Noise Improved MLS method, the response seems to be very detailed, but shows a response that is essentially flat within a +/-5db window. This is the gated response measurement, and the gate point is "automatically" detected. The response shows more detail in the high frequency area than it does in the low frequency area it seems. I see pretty clear signs of comb filtering from 2khz to say 18khz where the response falls off like a brick wall. What I find is that, if I change the subwoofer level, it seems to have virtually no impact on the relative level of the bass to the main speakers. Why would this be? When I switch to a smoothed response instead, it shows the low end response much better, and I can clearly see relative level changes, and EQ change. Again, why would this be? I want to make sure the methods I'm using for setting levels and eq's makes sense, and it seems that the Square Noise gated method isn't giving me good resolution down there (When I make an EQ change I often can't even see the change).
Ok second question, regardless of the method I use to measure, I see signs of excess energy between 300hz and 800hz. It appears to be exacerbated by large response dip centered at 70hz, but gradually extending out to around 150-200hz. Subwoofer placement optimization will likely resolve at least some of that problem, but the excess energy I think is crossover related. I saw this same issue when I would take measurements in my old room. Now that you do have this alternate crossover it makes me think that the new crossover would largely resolve this problem? I know it's expensive, but when you consider what many audiophiles will spend on magic stones, I would consider the cost of replacing the crossovers pretty minimal, with a much more profound impact on the sound. If you don't think its being caused by the crossover design, I'd love to hear what else it might be. I've measured all three of my speakers both in the room and outside, and they all show the same thing.
Ok the last issue is what to do in order to fix these many problems. My thoughts at this point are to fix the room first and foremost. When I get a chance I want to try some different subwoofer arrangements as well to fix as many of those bass problems as I can. The main speakers alone show a 70hz suckout which I believe is caused by the floor bounce. With subwoofers it shows a peak, so I actually have a PEQ centered around that. There is another one at 38hz that I also am removing. That actually gets the peaks below 100hz down, so its a matter of that trough in the response centered around 200hz. Since my subs aren't operating in that range, no amount of eq or moving is gonna impact that. Then there is the excess of energy in the 300-800hz range that I think is best handled in the crossover, but could be done through an external eq as well.
Ok finally, the room reflections. I'm guessing that all of that upper end comb filtering that is so evident is coming from the fact that this room has virtually no absorbing materials. The floors are hardwood, the windows have blinds on them and are centered right at the first sidewall reflection point, and my one and only chair in this room is leather. I plan on putting a large thick wool area rug on the floor with a relatively thick pad underneath. I plan on putting drapes up on each side wall, in front of the blinds. The front wall with the screen will have the 4" thick acoustic panels flanking the screen to the wall, with a curtain surrounding that, at least eventually. I don't think its going to help to try and absorb ceiling reflections, so I"m thinking of using a painted diffraction device above each speaker at the first reflection point, and covering the rear wall with something that diffracts. I probably will have a pair 2'x2' acoustically absorptive panels roughly 2" thick back there as well, but placed up near the projector (I'm hoping it will slightly reduce the fan noise). Bass traps are what I have been using, the tuned traps I built. What else should I consider to deal with these issues?
First, When I take a measurement with Square Noise Improved MLS method, the response seems to be very detailed, but shows a response that is essentially flat within a +/-5db window. This is the gated response measurement, and the gate point is "automatically" detected. The response shows more detail in the high frequency area than it does in the low frequency area it seems. I see pretty clear signs of comb filtering from 2khz to say 18khz where the response falls off like a brick wall. What I find is that, if I change the subwoofer level, it seems to have virtually no impact on the relative level of the bass to the main speakers. Why would this be? When I switch to a smoothed response instead, it shows the low end response much better, and I can clearly see relative level changes, and EQ change. Again, why would this be? I want to make sure the methods I'm using for setting levels and eq's makes sense, and it seems that the Square Noise gated method isn't giving me good resolution down there (When I make an EQ change I often can't even see the change).
Ok second question, regardless of the method I use to measure, I see signs of excess energy between 300hz and 800hz. It appears to be exacerbated by large response dip centered at 70hz, but gradually extending out to around 150-200hz. Subwoofer placement optimization will likely resolve at least some of that problem, but the excess energy I think is crossover related. I saw this same issue when I would take measurements in my old room. Now that you do have this alternate crossover it makes me think that the new crossover would largely resolve this problem? I know it's expensive, but when you consider what many audiophiles will spend on magic stones, I would consider the cost of replacing the crossovers pretty minimal, with a much more profound impact on the sound. If you don't think its being caused by the crossover design, I'd love to hear what else it might be. I've measured all three of my speakers both in the room and outside, and they all show the same thing.
Ok the last issue is what to do in order to fix these many problems. My thoughts at this point are to fix the room first and foremost. When I get a chance I want to try some different subwoofer arrangements as well to fix as many of those bass problems as I can. The main speakers alone show a 70hz suckout which I believe is caused by the floor bounce. With subwoofers it shows a peak, so I actually have a PEQ centered around that. There is another one at 38hz that I also am removing. That actually gets the peaks below 100hz down, so its a matter of that trough in the response centered around 200hz. Since my subs aren't operating in that range, no amount of eq or moving is gonna impact that. Then there is the excess of energy in the 300-800hz range that I think is best handled in the crossover, but could be done through an external eq as well.
Ok finally, the room reflections. I'm guessing that all of that upper end comb filtering that is so evident is coming from the fact that this room has virtually no absorbing materials. The floors are hardwood, the windows have blinds on them and are centered right at the first sidewall reflection point, and my one and only chair in this room is leather. I plan on putting a large thick wool area rug on the floor with a relatively thick pad underneath. I plan on putting drapes up on each side wall, in front of the blinds. The front wall with the screen will have the 4" thick acoustic panels flanking the screen to the wall, with a curtain surrounding that, at least eventually. I don't think its going to help to try and absorb ceiling reflections, so I"m thinking of using a painted diffraction device above each speaker at the first reflection point, and covering the rear wall with something that diffracts. I probably will have a pair 2'x2' acoustically absorptive panels roughly 2" thick back there as well, but placed up near the projector (I'm hoping it will slightly reduce the fan noise). Bass traps are what I have been using, the tuned traps I built. What else should I consider to deal with these issues?
Attachments
Matt
These are complex measurement issues.
First don't use MLS, use log sweep. In you last plot, nothing below 1 kHz is valid and above 1 kHz looks like a reflection in the data.
I would need to see impulse responses first and only then look at frequency response. Most of what you are showing is meaningless.
Its impossible to get windowed data below about 200 Hz in any real room. So when you make changes down below that you likely won't see anything or what you see is not valid anyways. You need to do steady state spatial averaged below 200 Hz. Basically there is no one technique that will work in a real room. It takes a composite of many.
There is no way that you are seeing any effect of the crossover in real room data. The changes to the crossover are far less than what you could resolve in a real room measurement.
There is a lot more that I could say, maybe I should write a book on "real room measurements", but I am out of time.
These are complex measurement issues.
First don't use MLS, use log sweep. In you last plot, nothing below 1 kHz is valid and above 1 kHz looks like a reflection in the data.
I would need to see impulse responses first and only then look at frequency response. Most of what you are showing is meaningless.
Its impossible to get windowed data below about 200 Hz in any real room. So when you make changes down below that you likely won't see anything or what you see is not valid anyways. You need to do steady state spatial averaged below 200 Hz. Basically there is no one technique that will work in a real room. It takes a composite of many.
There is no way that you are seeing any effect of the crossover in real room data. The changes to the crossover are far less than what you could resolve in a real room measurement.
There is a lot more that I could say, maybe I should write a book on "real room measurements", but I am out of time.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.