andreasmaaan Posted December 17, 2016 Share Posted December 17, 2016 Hi, I'm using software to model a couple of speaker designs that I'm planning to build in the new year, and I'm finding it both interesting and baffling that the modelling software seems to inevitably model a bump in the power response in the midrange of all my 2.5 way designs. This bump seems to be about 2db higher in level than the summed on-axis response. My first thought was that there is some non-linear phase interaction going on between either the two woofers or between the bass-woofer and the tweeter, as this is just below the point at which these two sources interact. However, when I model the response from countless off-axis positions (+90 to -90 degrees on both axes), there is simply no bump, and indeed (as you'd expect from any conventional crossover) there's a significant dip in this range as the listener moves upward along the vertical plain. Is this a mistake of my design software or is there an obvious explanation for this that I'm missing? Cheers, Andreas Link to comment Share on other sites More sharing options...
andreasmaaan Posted December 17, 2016 Author Share Posted December 17, 2016 Update: the same power response bump is there when I remove the tweeter, i.e. if it is the result of an interaction between the drivers' spatial relationship, it is a result of the relationship between the two woofers. Link to comment Share on other sites More sharing options...
VanArn Posted December 17, 2016 Share Posted December 17, 2016 More information is needed to be able to give an informed opinion. You need to indentify the speaker drivers, the crossover cct. and the layout of speakers on the front panel etc. 1 Link to comment Share on other sites More sharing options...
davewantsmoore Posted December 17, 2016 Share Posted December 17, 2016 (edited) Yes... and which software - so we can understand how/what it is calculating. Assuming it is simply the relationship between the two woofers, then I would want to see a amplitude+ phase chart of each driver, and their sum, overlaid. The power response of a LR crossover has a dip in it, the magnitude of which depends on the distance between the two sources .... but this will be for the tweeter/woofer (highpass/lowpass) .... so that's not the effect which you're seeing if it is just the woofer, and the 0.5way woofer. 5 hours ago, andreasmaaan said: This bump seems to be about 2db higher in level than the summed on-axis response. Right, just noticed this. The LR crossover issues is a dip (down) .... yours is the other way. Dunno. More info. Edited December 17, 2016 by davewantsmoore Link to comment Share on other sites More sharing options...
andreasmaaan Posted December 17, 2016 Author Share Posted December 17, 2016 Hi @davewantsmoore and @VanArn, thanks for commenting. The software is Jeff Bagby's passive xover designer and the woofers in this case are Peerless P830884. The crossover is 2nd order LR2 @ 1800hz between midwoofer and tweeter and 1st order at around 400hz on the woofer. The speakers are laid out in a TMM configuration with about 20cm between the centre of each driver. Cheers, Andreas Link to comment Share on other sites More sharing options...
andreasmaaan Posted December 17, 2016 Author Share Posted December 17, 2016 (edited) A screenshot. Note that there is no sign of peaking beyond a decibel or so in this region at any point in the modelled off-axis response, and that there is significant off-axis dipping in this region too. Edited December 17, 2016 by andreasmaaan Edited for better clarity Link to comment Share on other sites More sharing options...
andreasmaaan Posted December 17, 2016 Author Share Posted December 17, 2016 (edited) Actually, here's a screenshot of the same configuration using ideal frequency responses for all drivers (and a modelled baffle step roll-off on the two woofers). So it seems that Jeff Bagby's software is trying to tell me that a bump in the power response in the upper region of the midwoofer (but not one that is caused by anything occurring within a 90 degree vertical or horizontal axis, apparently) is an inevitable outcome of any 2.5 way design. Is that right? Or is this a bug in the software? Edit: here is a screenshot from the off-axis position (within 90 degrees vertical/horizontal) which produces the greatest bump in this reason. It's at -30 degrees vertical, so it's a floor reflection, but still I don't feel that it alone can explain the modelled power response... Edited December 17, 2016 by andreasmaaan Link to comment Share on other sites More sharing options...
VanArn Posted December 17, 2016 Share Posted December 17, 2016 More questions for andreasmaaan, as there is still not enough detail on the components. The H.F. unit from the simulation appears to be a horn or wave guided speaker and that is all I can tell from the 1.8kHz xover freq. and SPL graphs. So, what make and model is it ? Are the driver responses actual measurements made by you or are they imported from another site ? Impedance and phase curves are very important in the design of a passive xover , as basic text book designs do not take into account the actual acoustic slopes of the speakers. Can you supply the values of the xover components and the cct.layout , given by the simulation ? Link to comment Share on other sites More sharing options...
davewantsmoore Posted December 18, 2016 Share Posted December 18, 2016 7 hours ago, andreasmaaan said: So it seems that Jeff Bagby's software is trying to tell me that a bump in the power response in the upper region of the midwoofer (but not one that is caused by anything occurring within a 90 degree vertical or horizontal axis, apparently) is an inevitable outcome of any 2.5 way design. Is that right? Or is this a bug in the software? No, that's not right .... the bump isn't [or shouldn't be] just in the midwoofer, it's part of the summed crossover response. What it is showing you (amongst other things which are partially obscuring this), is that a Linkwitz Riley crossover has a dip in the power response in the crossover region. I've not been a big user of this software, so I can't say if there's other problems with what you're doing that might be causing (some of) what you're seeing. I'd want to see the phase curves of the drivers, etc.... but it's very difficult over a forum. Link to comment Share on other sites More sharing options...
andreasmaaan Posted December 18, 2016 Author Share Posted December 18, 2016 Thanks for the replies everyone At this stage I haven't bought or taken my own measurements of any drivers, I've just been modelling various combinations of drivers from published measurements to try to decide on what to buy and build. The modelling is obviously of limited value for this reason, but it does incorporate modelled in box response and modelled baffle diffraction/step.But really my question is not about this particular design, it's a question about 2.5 way systems in general. That's why I posted the above modelled response using "ideal" drivers, i.e. with perfect frequency response and 0 degrees phase response 20-20,000hz, to show how the software seems to be telling me that a bump in the power response in the midrange seems to an inevitable outcome of the 2.5 way topology. Here's the screenshot again (power response in green):So I'm actually just trying to understand if the bump in the modelled power response of the ideal drivers is based on the topology itself... since it isn't something I can explain through my own understanding of phase interactions between the drivers.Cheers, Andreas Sent from my iPhone using Tapatalk Link to comment Share on other sites More sharing options...
VanArn Posted December 18, 2016 Share Posted December 18, 2016 O.K. Andreas, The 2.5 way speaker concept is workable using bass/ mid drivers of the 5 to 6.5'' variety. Basically they have evolved as a result of the fashion for narrow front panel enclosures. The problem with trying to use an 8'' bass speaker , such as the Peerless 830884 ( which is no longer available ) is that the directivity pattern narrows and rises slightly on axis in the xover region. Any electronic correction will effect the off axis response. The enclosure would be on the large size if two 830884 ,s were employed to their fullest capabilities ( around 130 litre, internal volume ). Link to comment Share on other sites More sharing options...
andreasmaaan Posted December 19, 2016 Author Share Posted December 19, 2016 Hey VanArn, Thanks and yes I understand that. As you picked up in your earlier post the tweeter is horn loaded, and the horn has a similar directivity profile to the 8'' woofer in the crossover region. So using the larger driver is in this case a deliberate move. And yes, the enclosure would be around that size by my calculations. Re: the 830884, it seems to be available here in Germany so I wonder if maybe it's back in production here (or if the store I'm looking at has an outdated website perhaps). I'm still having trouble understanding this power response peak in the modelled response though. The software seems to model it into every 2.5 way speaker no matter what I do. Hence my question about the power response of 2.5 way systems generally - whether this peak is a real phenomenon that is recognised and understood or a glitch in the software? Link to comment Share on other sites More sharing options...
VanArn Posted December 19, 2016 Share Posted December 19, 2016 It is described as the power response and I think, it would be necessary to calculate the acoustic power ( Pa ) ,at a given frequency ,from the formula :- SPL = 112.2 + 10 log (Pa). Its' only use would be to indicate how the power changes for the speaker system at various frequencies to achieve a flat total response i.e. the black 90dB SPL line. Link to comment Share on other sites More sharing options...
davewantsmoore Posted December 19, 2016 Share Posted December 19, 2016 3 hours ago, andreasmaaan said: I'm still having trouble understanding this power response peak in the modelled response though. The software seems to model it into every 2.5 way speaker no matter what I do. Hence my question about the power response of 2.5 way systems generally - whether this peak is a real phenomenon that is recognised and understood or a glitch in the software? Right now, I understand the question (it's about the rise where the red arrow is). Is the question simply "why is the green line higher than the blue line?" .... I'm not sure why you're seeing that. Link to comment Share on other sites More sharing options...
andreasmaaan Posted December 19, 2016 Author Share Posted December 19, 2016 Hi @davewantsmoore, yes that's exactly the question So, if I believe the software, the woofers in any 2.5 way configuration have an inherent peak in the power response in the range between where the bass driver begins to roll off and where the mid-bass driver begins to roll off. And since (according to the software) this peak cannot be accounted for +/- 0-90 degrees off axis (that is far as the software allows me to model), the cause must be in the extreme off-axis response (90 degrees to 270 degrees off axis). Which I guess must be a result of the phase relationship between between the two woofers when one woofer is rolling off at 6db/octave while the other isn't.... or, the software is wrong... Can anyone explain this theoretically? Or explain why the software is wrong? Or if that's not possible, maybe someone who has other software that they use for crossover modelling could check if they get the same results? Cheers, Andreas Link to comment Share on other sites More sharing options...
davewantsmoore Posted December 19, 2016 Share Posted December 19, 2016 2 hours ago, andreasmaaan said: Can anyone explain this theoretically? Or explain why the software is wrong? No, and no. I'm thinking you've done something 'wrong' in the software. I found it easy to make mistakes with this software in the past. I haven't got a computer (Excel) handy which can run it right now. What I would want to do, is take the tweeter out... and see the response for just the two woofers. Link to comment Share on other sites More sharing options...
VanArn Posted December 19, 2016 Share Posted December 19, 2016 (edited) The power response graph is only indicating the acoustic power distribution of the system. It is not the on-axis frequency response. A direct radiator speaker has a low efficiency as there is a mismatch between the mechanical power source and into the air load that is the radiated sound. Basically Pa= V^2.Rma and the radiation resistance (Rma) is a function of the square of the frequency. What this means, is that as the air match between the air and the diaphragm improves as a result of Rma increasing in proportion to the area of the speaker cone; the Pa rises. The limiting factor to this matching improvement is the roll off caused by the moving mass of the cone. This is only part of the answer and a study of a good text book or article on loudspeakers would aid understanding. I do not use software but I do not believe it is faulty, it is more likely a case of too much information that is causing confusion. Edited December 19, 2016 by VanArn 1 Link to comment Share on other sites More sharing options...
davewantsmoore Posted December 19, 2016 Share Posted December 19, 2016 3 hours ago, VanArn said: it is more likely a case of too much information that is causing confusion. Indeed. That's why I said I would want to remove the tweeter... (and the second woofer) ... then carefully add the second woofer, etc. to see where the unexpected result happens (and try to work out why). I am doubtful that the software is actually "faulty".... but I do know it is difficult to drive. Link to comment Share on other sites More sharing options...
andreasmaaan Posted December 19, 2016 Author Share Posted December 19, 2016 Hey @davewantsmoore and @VanArn, thanks for sticking round on this one Dave, I have actually tried that. Basically the power response peak remains when the tweeter disappears. And it's there no matter whether I model the design with the 'ideal' responses I posted last, or with the modelled responses of the drivers as I posted originally. So it seems that it must be to do with the interaction between the two woofers (or any two woofers in this topology). FWIW, here is a screenshot without the tweeter: And here's a shot of the response from +45 degrees vertically off axis, which is the point at which fiddling around with the position of the imaginary mic got me the closest to the effect showing up in the power response, but still IMHO nowhere near close enough to explain it: Van, I'm not sure I understand you correctly, but I think you're relating this to the directivity of the drivers? I do understand what the power response is, and I do more or less understand the principles that affect the directivity of a driver with reference to it's radiating surface and frequency. But the phenomenon that this software seems to be pointing to (the peak in the power response that I marked in the pic) seems to me to be unrelated to this, since it appears to be the result of interaction between two drivers with identical directivity, but non-identical phase response due to differing filter slopes (i.e. a 1st order slope on the lower woofer @ 300hz or so, cascaded into a 2nd order slope over both woofers in the upper midrange). Will be interesting to get to the bottom of this... Cheers, A Link to comment Share on other sites More sharing options...
davewantsmoore Posted December 20, 2016 Share Posted December 20, 2016 14 hours ago, andreasmaaan said: FWIW, here is a screenshot without the tweeter: What about with just the woofer (blue)? Link to comment Share on other sites More sharing options...
davewantsmoore Posted December 20, 2016 Share Posted December 20, 2016 14 hours ago, andreasmaaan said: FWIW, here is a screenshot without the tweeter: I'm not seeing anything which looks "completely wrong" or explainable from the "is the software doing what it is told?!", perspective. Below 100Hz ... You have two correlated woofers (sum = +6dB) 100-1000Hz..... one of the woofers (blue), increases in level by ~4dB ..... however the summed response of the two woofers is falling (at this angle). This implies that the drivers are out of phase (cancelling each other) ... if not, then I don't understand why (perhaps the software is doing something wrong then). Above 1000.... the summed rolloff is tracking the blue woofer (it is a little higher due to the output of the pink woofer!?) Power response .... the power response (ie. all angles) ... is following the blue woofer .... this implies that the (potential) destructive interference between pink and blue, is not happening at all angles. (but green doesn't rise AS HIGH as blue, because there is some interference) Phase charts would help to see the relationship. Link to comment Share on other sites More sharing options...
andreasmaaan Posted December 20, 2016 Author Share Posted December 20, 2016 8 hours ago, davewantsmoore said: What about with just the woofer (blue)? Great question, hadn't thought of that. Here is just the blue woofer and I think this give us an explanation: The software is tracking the blue woofer +3db to model the power response for some reason. I have no idea why it's doing this but it does seem to clear up that it's not to do with the phase relationship between the drivers particularly. What do you think @davewantsmoore? Link to comment Share on other sites More sharing options...
davewantsmoore Posted December 20, 2016 Share Posted December 20, 2016 I don't know why it is +3 Link to comment Share on other sites More sharing options...
andreasmaaan Posted December 20, 2016 Author Share Posted December 20, 2016 Ah well, problem (more or less) solved for now. Cheers everyone 1 Link to comment Share on other sites More sharing options...
Recommended Posts