Have I designed first order? Absolutely, a 3-way, one that I liked rather well actually. I had several crossovers for that system. Your assumption that I don't like first order is certainly wrong. At our annual gathering the first order was preferred by the others as well, somewhat to my surprise. It was a rather complicated crossover being a 3-way first order, something of the way Dunlavy did his I would guess, his were not simple. I think that I did a good job on that one given the group's response.
There wasn't much of fact in the statement "these combinations are rare". To what should I pay attention? It says absolute nothing with regard to the acoustic responses.
I am not setting myself up as some expert, I have some experience, certainly enough to ask what I believe to be pertinent questions. If something I've said is shown to be wrong, fine. I had to make a few assumptions at first because there wasn't much to go on. But with all said, I still see no relation to what Dunlavy did. He certainly never ran any drivers full range to my knowledge. Are you aware of any that he did?
So the tweeter Fc is set at 6000Hz. That may start as first order, but it will turn to third at the knee of the driver raw curve before the baffle diffraction influence is taken into account, that was my point. Diffraction will, of course, be much less of an issue with a modest horn loaded tweeter such as the MDT-37. A couple of measurements would be much easier to discuss, one would assume that they were made. It seems that the tweeter is the closer one in this case.
An Fc of 6K might also be an issue with regard to center-to-center spacing, having a wavelength of may 2.25". The spacing is easily going to be more than a full wavelength at Fc. That bears some attention as well. In that area, though, I've never been one that put as much emphasis on that as so many others do.
To me it would seem that the full range midwoofer is more problematic. If the driver examples that you used matched the Morel anechoic measurements (smoothing unknown if used as it was in some of the tweeter responses), the MW144 does not have a corresponding -3db point at 6k, it's more like 10K. Looks like a bit of overlap to me, but again, a set of measurements could put that to rest, at least on the design axis. The MW144 is still at 0db relative at 6K, so the interaction with the tweeter according to those numbers would not achieve a BW1 target. Maybe the MW144 drivers that you used did not match the Morel supplied curves that I've seen. I'm looking at a catalog from 2003 and the MLSSA anechoic response that they supplied. That one is in fact relatively flat to 8K, hence my suspicion that the integration would not be as desired.
If you could show the two measurements, that would be most helpful. We have not discussed the baffle step loss, of course, that's another issue that generally must be taken into account in a proper crossover. Some drivers have an inherent downward sloping passband response that coupled with the right baffle will provide a flat response when coupled with the step, but that MW144 shows flat down to 50Hz as seen in the Morel measurements, anechoic. Put that driver on a baffle with no crossover and there's no getting around a baffle step loss with a peak above the loss area. How have you handled that with no crossover?
How they interact (Marry as you call it) is immaterial when analyzing each driver. They ultimately combine, but they do nothing one to the other (except for tweeter diffraction due to the woofer in many cases). One has to cross each one to get a specific target summed acoustic response. Both have to be taken into account of course, but that's pretty much a given, wouldn't you say? It's all part and parcel of design.
It is not necessary for one to hear a system to discuss the technical aspects such as crossover slopes and targets. I certainly never even implied that it would sound poor. If one designs by ear, yeah, you've got to listen. But I don't design by trial-and-error listening. I set a goal, match it with measured driver highpass and lowpass, then listen and evaluate. If a change is required, I do that by setting and achieving a change to the target that I expect to result in the change I desire. But in the end it is all verified with measurements.
Dave
There wasn't much of fact in the statement "these combinations are rare". To what should I pay attention? It says absolute nothing with regard to the acoustic responses.
I am not setting myself up as some expert, I have some experience, certainly enough to ask what I believe to be pertinent questions. If something I've said is shown to be wrong, fine. I had to make a few assumptions at first because there wasn't much to go on. But with all said, I still see no relation to what Dunlavy did. He certainly never ran any drivers full range to my knowledge. Are you aware of any that he did?
So the tweeter Fc is set at 6000Hz. That may start as first order, but it will turn to third at the knee of the driver raw curve before the baffle diffraction influence is taken into account, that was my point. Diffraction will, of course, be much less of an issue with a modest horn loaded tweeter such as the MDT-37. A couple of measurements would be much easier to discuss, one would assume that they were made. It seems that the tweeter is the closer one in this case.
An Fc of 6K might also be an issue with regard to center-to-center spacing, having a wavelength of may 2.25". The spacing is easily going to be more than a full wavelength at Fc. That bears some attention as well. In that area, though, I've never been one that put as much emphasis on that as so many others do.
To me it would seem that the full range midwoofer is more problematic. If the driver examples that you used matched the Morel anechoic measurements (smoothing unknown if used as it was in some of the tweeter responses), the MW144 does not have a corresponding -3db point at 6k, it's more like 10K. Looks like a bit of overlap to me, but again, a set of measurements could put that to rest, at least on the design axis. The MW144 is still at 0db relative at 6K, so the interaction with the tweeter according to those numbers would not achieve a BW1 target. Maybe the MW144 drivers that you used did not match the Morel supplied curves that I've seen. I'm looking at a catalog from 2003 and the MLSSA anechoic response that they supplied. That one is in fact relatively flat to 8K, hence my suspicion that the integration would not be as desired.
If you could show the two measurements, that would be most helpful. We have not discussed the baffle step loss, of course, that's another issue that generally must be taken into account in a proper crossover. Some drivers have an inherent downward sloping passband response that coupled with the right baffle will provide a flat response when coupled with the step, but that MW144 shows flat down to 50Hz as seen in the Morel measurements, anechoic. Put that driver on a baffle with no crossover and there's no getting around a baffle step loss with a peak above the loss area. How have you handled that with no crossover?
How they interact (Marry as you call it) is immaterial when analyzing each driver. They ultimately combine, but they do nothing one to the other (except for tweeter diffraction due to the woofer in many cases). One has to cross each one to get a specific target summed acoustic response. Both have to be taken into account of course, but that's pretty much a given, wouldn't you say? It's all part and parcel of design.
It is not necessary for one to hear a system to discuss the technical aspects such as crossover slopes and targets. I certainly never even implied that it would sound poor. If one designs by ear, yeah, you've got to listen. But I don't design by trial-and-error listening. I set a goal, match it with measured driver highpass and lowpass, then listen and evaluate. If a change is required, I do that by setting and achieving a change to the target that I expect to result in the change I desire. But in the end it is all verified with measurements.
Dave
Sounds good to me! Perhaps you would like to put your views forward or ask questions. I have 'things I always do" when designing. Taking parts away from a designed xover is one of those things. Initially I had a 0.22 inductor on the MW144/142. When I took it off, the system measured better and sounded better. If I used an MW168 the inductor stayed. The MW144 has brilliant response out to 8K. It loves the DMS37. It should do, I had a hand in it's design. At the time I was the Aus distributor for Morel. When Morel was set to design the DMS 37 I sent them my research on time alignment which was based on the MDT30/32. They saw the light and lined up the acoustic centres to be spot on. It was when I was playing with the first development samples that I discovered the incrediable relationship that this combination has. Tis quite rare, but god it sounds incredible/
Terry
Let me interject - have any of you not seen a speaker that measures superbly, yet sounds terrible?
I also have to disagree with Dave if he means to imply that the portion of drivers (mids/woofers + tweeters) which will function well as a system together with simple first order electrical crossovers (or, the "just a cap on the tweeter) is not an extremely small fraction of them?
I have seen VERY few designs like Zaph's Waveguide TMM in proportion to the vast majority of designs with higher order electrical crossovers (not counting bad sounding K-mart speakers and the like).
I have to admit to not have seen very many of the crossovers of older well respected commercial designs (by "older" I mean from the previous few decades of the last millenium).
I also have to disagree with Dave if he means to imply that the portion of drivers (mids/woofers + tweeters) which will function well as a system together with simple first order electrical crossovers (or, the "just a cap on the tweeter) is not an extremely small fraction of them?
I have seen VERY few designs like Zaph's Waveguide TMM in proportion to the vast majority of designs with higher order electrical crossovers (not counting bad sounding K-mart speakers and the like).
I have to admit to not have seen very many of the crossovers of older well respected commercial designs (by "older" I mean from the previous few decades of the last millenium).
I remember the first xover I designed for the JV60. Measured like a dream come true. Then I took it out to the listening room and plugged it in, sat back, 
. It was woeful. 
I started from scratch again. Terry
What was the design axis? Distance? Baffle dimensions? It's rather difficult to see what the situation is with a single point summed response and no other data as to conditions.
The 2k dip should I suppose be credited to baffle diffraction of the woofer and likewise the mile peaking at 1K.
There's a fairly broad dip centered at 4K, the source of which can't be discerned in seeing only a summed response. Given that the woofer is supposed to be flat in that area and that the tweeter is claimed to have had a true first order response in the crossover area, something is amiss to cause a significant dip such as that. It's sure not what I would expect from the two drivers' curves shown by Morel, but then manufacturers have a habit of showing optimistic curves. I would suspect phase issues.
The measurement stops at 500Hz. Baffle step is typically just at the relative 0db at this point. That output is trending down and will be dropping below that point by a full 6db, so the measurement is missing a significant part of the response. The indication would be a rather lean bass, but maybe that was considered desirable.
In the end it still does not appear to meet a first order summed target unless the woofer is somehow rolling off earlier than in Morel's curves. The phase influence of the tweeter rolloff may be part of the reason when such a high crossover is used.
It would be interesting to see the polar response given the rather wide CTC with a 6K crossover.
Dave
Last edited:
I'm not sure what I said that left that impression, but it's not my position, far from it.
Dave
I thought that +/- 2dB was quite good. It was +/-2.5dB with raw data and +/-2dB with 0.1 Octave smoothing. That response is taken at 1 metre. with less than 1 watt of power. Speaker was 92dB at 2.83V. It was a point source using 2 MW144 and the DMS37 tweeter. Sorry but that's the only graph I have of that system.
Terry
Maybe the sensitivity was different back then, but that would be a surprise. The Morel info I have shows a sensitivity of 1W/1m as 88db. The MLSSA curve would be about 87db and I suspect is 2.83v. Neither one gets to 92db. If at 1m on the tweeter axis, a full range midwoofer will likely have some off-axis droop, these aren't truly point source drivers. That would help explain the 4K dip as well as relative phase in that area.
The DMS-37 would be in the 92db range with just a cap, but I don't see how the combination does 92db given the data that Morel provides. In any case, a full range MW144 on a baffle is going to act as all dynamic drivers do on a baffle, it's going to operate into 4-pi below baffle step so the area below 500Hz is going to lose 6db. That was one key point I've been trying to make. The measurement supports that this in fact occurring, even though we only see the area above 500Hz that will be peaking due to baffle diffraction. Maybe lean bass wasn't an issue, that falls into the area of preference. It will be the case, however.
There does still seem to be some destructive interference in the overlap region. If the woofer is flat to 8K (0db relative) and the tweeter is crossed to be -3db at 6K, there is not going to be proper summation as a BW1, especially with two drivers with different sensitivity ratings. It may be relatively flat due to phase issues, the 4K dip indicates something is still a bit amiss, but then we're also still limited to a single point. I actually would have expected the area above 6K to show some interference (call it non-optimal summation), but it may be that the off-axis droop (assuming that the measurement is on the tweeter's axis) is responsible for preventing peaking that would have to occur at 6K up if the responses were as shown by Morel. The gradual rolloff of the MW144 is certainly better than most to 10K, though there will likely be a much quicker rolloff at relatively low off-axis angles. I can't say if it's like the Dynaudios I used to use.
In the end it is a relatively smooth on-axis response, but the baffle step loss and the expected uneven power response would give me pause. I'd need to see individual responses to include off-axis (or sims) to get a better feel for the actual integration. I would have an issue with a driver that is becoming directional well below the crossover point. Morel's curves show only one off-axis response, I suspect it's 30 degrees, they don't even say. It starts to drop at 4K. Personally I wouldn't take that driver so high, especially given the ability of the MDT-37 to go as low as it can. I would consider as high as 4K and that doesn't address the CTC spacing issue. Very small changes vertically are guaranteed to see quickly varying destructive inteference due to path length difference where the spacing is at least 2x the wavelength at Fc (6K).
But again, to each his own.
Dave
Last edited:
Seems I missed the point of there being 2 midwoofers, certainly explains the sensitivity question I had. I guess I wasn't fully awake this morning. Two drivers that are spec'ed at 87db will easily get to 92db, but that's only 2-pi. I'm still looking for feedback on the baffle step since two drivers full-range that are flat anechoically as measured by Morel are going to lose 6db starting somewhat around the 500Hz point that is not shown in the measurement. It doesn't matter whether there's one or two, there will be a 6db step loss.
The issue that concerns the 4K broad dip still remains unanswered as well, since the claims were that this is a true first order crossover. Were that the case, there should be a peak at the stated Fc of 6K given that the MW144 is flat (not down 3db) at 6K.
Dave
Dave, If you can design and make all these decisions based on the manufacturers data then we design much differently. Me, I go through the process of measuring drivers, feeding real world data into LEAP, etc etc.
Manufacturers data from Morel is generally very good, but not that good.
Terry
I'm not trying to design, I'm trying to get answers. Believe me I fully understand the issues, this is more to clarify for others who may not and accept commentary at face value.
As far as design, I do the same as you in that respect. I've used LMS, MLSSA, LAUD and SoundEasy as well as CALSOD and other tools, still use several in fact. My questions can be answered without making assumptions of what or how I do it or implications that I do otherwise. The point is, there is a distinct lack of information as I've indicated and I can ask questions, must in fact, since there is insufficient data. This relates to the basic topic of running a midwoofer full range and the limitations that can and do occur that needs a full accounting if one is going to recommend running drivers such as these full range. Interested DIYers should know all of the details. You offered your commentary about the claims, I didn't solicit it, so I'm trying to get answers.
There seems to be a reluctance on your part to respond directly to pertinent questions, that's why I'm repeating them. I questioned the accuracy of claims of first order response and have indicated why I think that there is an issue with the crossover region as well as the lack of data on the step, the latter of much more significance I would say. I don't understand the reticence to answer that question, it occurs in all baffle mounted direct radiators. The former point requires individual driver responses to assess. I am not one to accept statements without support, hence my queries. Since the claim was that it is first order similar to Dunlavy's work, it must follow that the response be BW1. I see reasons to question that and have stated where and why. That's it.
Dave
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.