Xmax question just a check really. Some makers specify a +/- figure, some peak to peak. Dayton just give a number. Is this +/- ? I've read that mechanical limits are said to be twice this figure? A post mentioned that they had asked Dayton.
Acoustic origin. An frd file states 51.73 uSec behind the baffle. My simulator needs a distance. Any thoughts? What do I do? It's critical in respect to looking at phase.
Acoustic origin. An frd file states 51.73 uSec behind the baffle. My simulator needs a distance. Any thoughts? What do I do? It's critical in respect to looking at phase.
I assume so based on this posted dimension*. Just a rule of thumb (ROT), need Klippel measurements or factory Xmech spec for accuracy.
As a 'day one' acceptable vertical alignment point is where the magnet bolts to the frame, though of course one can accurately measure time delay nowadays, which apparently has already been done for us.
Sound travels at ~34400 cm/sec and a uSec apparently is 0.000001/sec (10^-6), so basically it's the back of the baffle, making me question this spec and/or the baffle's thickness.
* 18" design for enormous air movement Over 2" of total cone travel! 2" = 50.8 mm, so its 22 mm Xmax seems reasonable
As a 'day one' acceptable vertical alignment point is where the magnet bolts to the frame, though of course one can accurately measure time delay nowadays, which apparently has already been done for us.
Sound travels at ~34400 cm/sec and a uSec apparently is 0.000001/sec (10^-6), so basically it's the back of the baffle, making me question this spec and/or the baffle's thickness.
* 18" design for enormous air movement Over 2" of total cone travel! 2" = 50.8 mm, so its 22 mm Xmax seems reasonable
Xmax is one way, an Xmax of 22mm would (should..) be +/-22mm, 44mm peak to peak excursion (travel).
Xmech, the mechanical limit may be anything from slightly more than Xmax to around double.
In the case of the UM18-22 18" Ultimax GM mentioned, the Xmech is only slightly more than the Xmax- 44mm peak to peak compared to "Over 2" of total cone travel!" or more than 50.8mm peak to peak, just over 25.4mm Xmech.
As a comparison, the B&C 18SW115 has an Xmax of 14mm (28mm peak to peak), and 60 mm Xmech, peak-to-peak excursion before damage.
Thanks. The driver I am interested in states 6mm 🙂 if that was +/- 3mm things would be rather different. 12mm p to p is competitive with certain other smaller drivers. SEAS for instance.
The acoustic origin. I'm using Visaton's Boxsim for simulation. It needs acoustic phase plus the usual. When one of their drivers are loaded it sets this to a distance behind the baffle, Seems logical for a cone speaker. I did see a comment by some on that they use some fraction of the cone depth related sum. Spec'ing as Dayton have time must relate to the speed of sound. (I'd hope). Standard air, no such thing really has a speed of 343m/sec. Given various parameters about the air the figure derived could be standardised to that. It works out at 17.7mm. This is a 7"speaker and that doesn't tie in well with what Boxsim loads on a 6 1/2 which is 46mm. Or sizes around that. A tweeter loads ~9mm. The actual statement in the Dayton frd file is
Acoustic origin is 51.73useconds behind the mounting plane.
The acoustic origin. I'm using Visaton's Boxsim for simulation. It needs acoustic phase plus the usual. When one of their drivers are loaded it sets this to a distance behind the baffle, Seems logical for a cone speaker. I did see a comment by some on that they use some fraction of the cone depth related sum. Spec'ing as Dayton have time must relate to the speed of sound. (I'd hope). Standard air, no such thing really has a speed of 343m/sec. Given various parameters about the air the figure derived could be standardised to that. It works out at 17.7mm. This is a 7"speaker and that doesn't tie in well with what Boxsim loads on a 6 1/2 which is 46mm. Or sizes around that. A tweeter loads ~9mm. The actual statement in the Dayton frd file is
Acoustic origin is 51.73useconds behind the mounting plane.
One thing I can do in Boxsim is load a Visaton chassis and change the "acoustic origin" and see what happens. Taking one simulation that is pretty sorted out and uses 45mm changing that to 17. This results in a phase shift of ~90 degrees in the xover region from something that was near perfect.
All rather disturbing. At the youtube end of things and Dayton it's look how easy it is. Download the data files, use Xsim etc and design your speaker and xovers. Without the origin info the results look like they are likely to be useless. Also on this particular Dayton it may be the only one that states an origin in it's frd file. The 7" ref model. No signs of anything in a couple of other chassis I have their data on.
Boxsim. There are loads of project files available and some also show measured results that are close to the simulation. 😉 Aimed at selling Visaton chassis though. Visaton is a bit strange on Xmax spec's. What gets loaded in Boxsim doesn't match data sheets. One sheet spec's 10% distortion levels. The others - pass. Might be mechanical.
I suppose the solution is to use spl plots and TS param and do a quick and dirty simulation on just the chassis to get an idea of what has to be dealt with and what the xmax figure means at the bass end. Then buy and measure the data that is needed for xover design. In my case assuming REW can determine the acoustic origin. Then get the parts, build and see what the room effects are.
If I lived in the USA I would be asking Dayton why they do not provide usable acoustic data. Mind you they do provide more data than others but one aspect appears to be useless. Boxsim - get the lot but only with Visaton drivers.
Well, it's not particularly useful in itself anyway, since if you're going to simulate a speaker (any) using supplied factory data rather than measuring it yourself, you'll likely be better off junking the supplied phase information, splicing in the simulated on-baffle response, and making an MP derivation from that, as the existing values will probably go to hell in a handbasket as soon as you start factoring everything else in. The acoustic centre of a driver on the Z-axis varies depending on design, but a reasonable approximation / starting point approximation for, say, an average 6 1/2in midbass tend to be something like 19mm - 25mm (3/4in - 1in) assuming the driver is on the same flat baffle without tilt as a typical / 'standard' type of tweeter. At the end of the day, you'll always be better off measuring it / the phase relationship between the drivers yourself though. No substitute for 'hard data' in the design situation.
It doesn't matter whether you live in the US or not -anybody can ask Dayton / Partsexpress anything they like, wherever they live, although you're probably more likely to get an answer on PE's own Techtalk site as Dayton are in effect their house brand.
Re Xmax, there's no real consensus about what it even is, let alone a single method for deriving a figure for it (as I often say, a while ago I tried to do a count of the number of different methods for generating an Xmax figure. I gave up at 12, all of which give different results if applied to the same driver & aren't especially informative). Usually though, it's given as 1-way, unless it's specifically stated as being point-to-point.
It doesn't matter whether you live in the US or not -anybody can ask Dayton / Partsexpress anything they like, wherever they live, although you're probably more likely to get an answer on PE's own Techtalk site as Dayton are in effect their house brand.
Re Xmax, there's no real consensus about what it even is, let alone a single method for deriving a figure for it (as I often say, a while ago I tried to do a count of the number of different methods for generating an Xmax figure. I gave up at 12, all of which give different results if applied to the same driver & aren't especially informative). Usually though, it's given as 1-way, unless it's specifically stated as being point-to-point.
Last edited:
The driver that I changed was a 6 1/2" bass mid. The acoustic offset is loaded as 45mm. Tweeter I chose 6mm. 🙁 This driver does not have enough xmax for the bass level I'd like to reach. That based on the value the chassis data load gives.
As I mention some of the project files that are available from a database that Visaton maintain show good agreement between simulation and final measurements but yes this is a tricky area. One of Boxsim's load options is as measured in the box as designed. The others are pretty specific concerning how the data is measured. Actually as an engineer I wonder why all speaker chassis makers don't work to some standard or the other.
I have found one decent pdf on finding acoustic offsets
https://app.box.com/shared/ouxjjsx0m8bs00cil5iq
I suppose the same thing could be done with 2 Dayton supplied FRD files.
but I'd rather find a way of doing it individually with eg REW. Must also run under Linux.
Yes, that's the late and much-missed Jeff Bagby's method for establishing a relative offset between drivers using single-channel measurement gear.
There seems to be a bit of a confusion here between the relative acoustic centres (the front-rear Z axis), and the relative offsets between the drivers, which also includes offsets in the horizontal X and vertical Y axis, at a given microphone / listening position. There is no way driver manufacturers can supply that in regular driver datasheets / data-sets for obvious reasons -they don't know your baffle and box design, they don't know what other drivers a given unit will be partnered with, and they don't know what offsets will exist between them.
If you want to simulate something, you'll need to use your desired design package or individual software to add all the anticipated baffle / box effects into the basic data with the drivers at their intended locations on the baffle, extract minimum phase, then in your crossover design package specify the relative acoustic offset between them. The Z axis is usually the hardest to establish as you've little frame of reference available, but most drive units of a given size fall into a fairly narrow range, and this approach is always an approximation.
As far as measuring the on-baffle responses of the drivers in your own design (preferable) goes, if single channel, you can use Jeff's method in your desired measurement software, or invest in a dual channel measurement setup, which will give you the relative phases, and you can then use this data in your desired crossover design software, with no need to manually specify any relative offsets (since you already have the correct values). Boxsim will, I suspect (it's a while since i last used it) be doing both of these, depending on what data you specify. If it's an existing Visaton speaker design, which they've measured, then it will likely be using the actual on-baffle data they took of those built speakers; if it's your own speaker design using Visaton units or any others you add, then it will probably be doing more or less as I described above for simulated designs.
There seems to be a bit of a confusion here between the relative acoustic centres (the front-rear Z axis), and the relative offsets between the drivers, which also includes offsets in the horizontal X and vertical Y axis, at a given microphone / listening position. There is no way driver manufacturers can supply that in regular driver datasheets / data-sets for obvious reasons -they don't know your baffle and box design, they don't know what other drivers a given unit will be partnered with, and they don't know what offsets will exist between them.
If you want to simulate something, you'll need to use your desired design package or individual software to add all the anticipated baffle / box effects into the basic data with the drivers at their intended locations on the baffle, extract minimum phase, then in your crossover design package specify the relative acoustic offset between them. The Z axis is usually the hardest to establish as you've little frame of reference available, but most drive units of a given size fall into a fairly narrow range, and this approach is always an approximation.
As far as measuring the on-baffle responses of the drivers in your own design (preferable) goes, if single channel, you can use Jeff's method in your desired measurement software, or invest in a dual channel measurement setup, which will give you the relative phases, and you can then use this data in your desired crossover design software, with no need to manually specify any relative offsets (since you already have the correct values). Boxsim will, I suspect (it's a while since i last used it) be doing both of these, depending on what data you specify. If it's an existing Visaton speaker design, which they've measured, then it will likely be using the actual on-baffle data they took of those built speakers; if it's your own speaker design using Visaton units or any others you add, then it will probably be doing more or less as I described above for simulated designs.
Last edited:
You also need to be a bit careful regarding any "fudge-factors". Most manufacturers use the figure derived directly from coil-height minus gap-height. But some, e.g. Faitalpro, apply a fudge factor to this figure (maybe related to the distortion resulting from excursion).
Right, which makes comparions of Xmax between models almost impossible, since as noted there as multiple derivations, and it's not always stated which were used. A handful of derivations (I'm not saying I agree with any of them, and I won't waste time giving critiques -just stating what some use -there are plenty more but I won't write them all out):
Xmax = Xmech
Xmax = 1-way Xmech / 1.41
Xmax = Absolute value of gap height - VC winding height / 2
Xmax = (Absolute value of gap height - VC winding height / 2) + xyz scaling factor
Xmax = 10% THD
Xmax = 5% rise in HD3
Xmax = excursion required to hit 95dB (or some other arbitary figure)
And so on. Personally, I'd say that the 2nd & 3rd are about the most consistent methods if you're going to compare between units but since it's a non-standard value we're never going to get any consensus. The first of those two I suppose at least tells you something of value, which is more than can be said for the others, since it's the point at which you'd need to double voltage input to reach the mechanical limit. But we're to a point clutching at straws here since Xmax isn't (can't really be) a very informative value whatever method is used to generate a figure for it. 😉
Xmax = Xmech
Xmax = 1-way Xmech / 1.41
Xmax = Absolute value of gap height - VC winding height / 2
Xmax = (Absolute value of gap height - VC winding height / 2) + xyz scaling factor
Xmax = 10% THD
Xmax = 5% rise in HD3
Xmax = excursion required to hit 95dB (or some other arbitary figure)
And so on. Personally, I'd say that the 2nd & 3rd are about the most consistent methods if you're going to compare between units but since it's a non-standard value we're never going to get any consensus. The first of those two I suppose at least tells you something of value, which is more than can be said for the others, since it's the point at which you'd need to double voltage input to reach the mechanical limit. But we're to a point clutching at straws here since Xmax isn't (can't really be) a very informative value whatever method is used to generate a figure for it. 😉
I'm not too concerned about X-Y aspect just the Z offset. The speakers are for our lounge so there isn't a specific listening position and polar response becomes the main aspect which can't be perfect at the HF end as tweeters beam more as F goes up. One old design intended for concert halls used several of them but doubt if I will try that. 😉 But who knows at some point.
I'd ideally like to use individual speaker monitoring to sort out relative phase. Using 2 speakers at the same time and aligning them - suggests the offset can be determined just in the mid frequency range area avoiding bass measurement problems. 😉 mid may not be the correct word but that sort of idea about actually doing it.
Boxsim can play with X-Y offsets and other factors but I'll go through those one at a time. It includes baffle effects which may be why they use baffle offsets rather than relative phase between speakers. Boxsim simulation accuracy. I'd be prepared to build one to see providing it can do what I want. That actually is compact and an F6 of 40Hz that I would ideally like to be 30 but is unlikely to be possible in ideal box sizes. I chose 20L. I do know what this means in terms of listening to a very wide selection of music, On the other hand we also watch streamed films. Worrying, the effects channel just gets dumped in. Way too loud really. We also use an AV receiver that as usual has a poor selection of woofer xover F's as they expect rather poor front speakers due to size. So I can't use that to get rid of deep bass. We don't use surround sound.
So I work on 6 1/2" Visaton chassis. It can exceed 100dB and more past 100Hz. ~95 down to 30Hz making me think more xmax would be a good idea. Looked at the 8" version and no gain as they use a smaller xmax. It's been educational anyway and I can now look at peak suppression. The driver has a bad one at 5kHz., Get rid of that and I may be able to xover at >2.5k. On the other hand a simple 2nd order xover greatly reduces it's effects. Working on something I may not build may sound silly but it's educational.
That particular one relates to some organisation's standards. They probably have others that relate to speakers. Standards seem to be a rather weak area in the speaker manufacturing world. As an engineer I find that more than disturbing.
Boxsim tries. Data load options are infinite baffle. Probably derived, DIN baffle / 200L, Visaton test box small or long side and box that is being simulated. So far I haven't found out what a Visaton test box is.
Fair enough providing you are accounting for them, since they can & usually do make a significant difference with rising frequency unless you're running a coax, so if you're concerned about phase they have to be factored in.
Yes, that's what Jeff points out.
As noted, the relative phase between the speakers at a given point is determined by the total offset of the drivers in all three planes, not the the difference between the acoustic centres (Z axis) alone. The baffle offsets on the X, Y & Z axis (the latter including the front-rear shift in acoustic centres & any tilt / steps present) to a given point in space determine the relative phase between the drive units. You don't use 'baffle offsets rather than relative phase' -the former sets the latter. When data which does not include the measured relative phase of the drivers taken on the baffle is used in simulation software like Boxsim, VituixCAD, PCD, WinPCD etc., (e.g. if you're starting with IB data which you then modify in the software to create a 'simulated measurement', and extract a minimum phase value from this for each driver) then you can use their X, Y & Z axis offsets to set the relative locations of the units in these planes from a virtual microphone / listening position to provide approximately correct phase relationships / offsets.
Loudspeaker drive units are the only things I know of that have an Xmax value. Setting Xmax at the deflection where 10% THD is reached is one method used by manufacturers for providing a figure for it. It's about as useful as all the others used by different manufacturers (i.e. 'not very').
It isn't ideal, but there have never been any universal standards in this field at all, which most of us have regretted for decades. Since we're not going to get it changed (as that would require every single manufacturer in the world to sign up to a given set of standards, which many wouldn't even agree with), all we can do is ensure as far as possible that we compare like with like ourselves when doing our own design work.
Last edited:
Breaking things down a little for clarity: manufacturer IB data contains no phase information useful for designing a multiway loudspeaker, since its response (and phase) will change as soon as baffle effects come into play, and you need accurate data to establish the relative phasing between units mounted at different points, and at different distances from, the designed microphone / listening position.
Assuming you have design software available (as is the case), then you have three broad methods open to you for designing.
1/ Simulation only. You take the IB response data for each drive unit, post-process it to splice in baffle effects (diffraction, step-loss etc.), and relevant box load, then perform a minimum phase extraction on this 'simulated measurement'. You then use this data in your crossover design software in which you'll add the design microphone / listening distance and the acoustic offsets between the drivers. Some packages like PCD, WinPCD, VitixCAD, Boxsim etc. allow you to enter this as individual X, Y and Z axis dimensions to the reference plane & distance. The first two are obvious, the Z axis is basically an approximation (for 6in - 7in units, 19mm - 25mm is usually a reasonable starting point) Others, like XSim, require you to calculate the summed offset yourself, so you'll need to establish the X & Y offset with basic trigonometry, then add the approximation for Z to that for a total offset value to the reference plane / postion.
2/ Single channel measurements. Assuming the microphone is at the same location (the reference plane / distance) for the individal on-baffle driver measurements, you can use Jeff's / similar method to establish the X, Y and Z axis offsets by taking another measurement with both drivers playing together in the same polarity. Using software like PCD which allows you to set offsets on each axis independently, you'll set the microphone distance you used, the horizontal and vertical offsets to the reference plane, and then adjust the Z axis offset for the larger unit / units until the simulated system response matches the measured response of both units playing together. You now have the Z axis offset value which you can use in combination with the X and Y values for designing your crossover.
3/ Dual channel measurements. These should preserve the relative time-of-flight, so no adjustments should be required if you set things up properly. Your individual driver responses at the reference position contain the phase data you need, so you should not need to manually enter any offset data at all in your crossover design software as it already includes everything necessary for that axis.
Assuming you have design software available (as is the case), then you have three broad methods open to you for designing.
1/ Simulation only. You take the IB response data for each drive unit, post-process it to splice in baffle effects (diffraction, step-loss etc.), and relevant box load, then perform a minimum phase extraction on this 'simulated measurement'. You then use this data in your crossover design software in which you'll add the design microphone / listening distance and the acoustic offsets between the drivers. Some packages like PCD, WinPCD, VitixCAD, Boxsim etc. allow you to enter this as individual X, Y and Z axis dimensions to the reference plane & distance. The first two are obvious, the Z axis is basically an approximation (for 6in - 7in units, 19mm - 25mm is usually a reasonable starting point) Others, like XSim, require you to calculate the summed offset yourself, so you'll need to establish the X & Y offset with basic trigonometry, then add the approximation for Z to that for a total offset value to the reference plane / postion.
2/ Single channel measurements. Assuming the microphone is at the same location (the reference plane / distance) for the individal on-baffle driver measurements, you can use Jeff's / similar method to establish the X, Y and Z axis offsets by taking another measurement with both drivers playing together in the same polarity. Using software like PCD which allows you to set offsets on each axis independently, you'll set the microphone distance you used, the horizontal and vertical offsets to the reference plane, and then adjust the Z axis offset for the larger unit / units until the simulated system response matches the measured response of both units playing together. You now have the Z axis offset value which you can use in combination with the X and Y values for designing your crossover.
3/ Dual channel measurements. These should preserve the relative time-of-flight, so no adjustments should be required if you set things up properly. Your individual driver responses at the reference position contain the phase data you need, so you should not need to manually enter any offset data at all in your crossover design software as it already includes everything necessary for that axis.
Normally we design speakers to an axis. Often tweeter. Your voice coil offset (Z axis) will be Larger than the 17.7mm quoted, since you have to allow for a distance down from your microphone (Y axis) as shown.
dual channel measurement, so that we record actual measured phase means all this is irrelevant (all values are zero)
If you only have a single channel measurement system (e.g. no loopback - USB mic) - then you can measure both drivers in parallel, and "line up" the Z offset on the woofer until the individual curves sum to the parallel measurement
I feel that is a good reason for simulation. Currently I have a certain baffle size and shape. It's based on a fairly ancient speaker, an AR-6a as I wanted to see if I could duplicate what it could achieve in a 15L sealed cabinet, I can't due to the 8" drivers that are available. Go to 20L vented and I can get close. When it comes to build I might decide to change the profile of the baffle to reduce the depth. Then comes the question of using manufacturers data or measured. Going on what I have noticed using the manufacturers data shows differences in behaviour that may even just depend on cone material and cabinet volume. This figures in a decision to buy. Also do I want to build a number of rough cabinets with different proportions to see the effect they have via monitoring each one?
Xmax. No simulator I am aware of takes no notice of it. Same with the people that use them.
Thanks for the other comments, I'd already concluded that the 17.7mm is likely to be way out. Interestingly this particular speaker is the only Dayton I am aware of that contains this style of what appears to be an offset in it's FRD on axis file.
Thanks for the sketch. Clearer than comments. In order to be able to measure calibrated SPL with various weighting I went for Dayton's USB mic. I also have a Behringer for if that type turns out to be needed. Bought long ago.
I'll scan the web to see if there are way of doing it on a single speaker/ Probably find nothing.
And you are defining Xmax as?
The majority of reasonable box modelling software have a data entry for it (however you define it, keeping in mind there is no standard definition), and excursion modelling by which you can see, by adjusting power input, whether you remain within this limit, or not, as the case may be.
Missed that part. When you say 'single speaker' do you mean one complete speaker (as in box, multiple drive units etc.), or are you meaning 'one drive unit'?
The answer in both cases though 'yes, you can use a Dayton USB measurement microphone or a Behringer measurement microphone' to do either.
Just for clarity -when designing a loudspeaker, you'll typically only measure one box and the drivers in it. When we talk about 'dual channel measurement' in the context of speaker design, we're usually referring to a particular type of measurement equipment and method, not 'two complete separate loudspeaker enclosures (e.g. a left & right stereo pair)'.
The SPL measurements will be calibrated to whatever input level you use providing you calibrate the microphone / microphone. I'm not sure what you're referring to re 'various weighting' though (?).
The USB mic's come calibrated for spl. The others need a calibrator to set them up correctly.
REW allows the weighting to be changed. Eg one I can recollect is A, there are others. Not that I know that this feature is needed but it's a sort financial decision. Given the usual mic some one might buy a calibrator or a calibrated sound level meter. REW allows no weighting. Buy a meter that will do this sort of thing and is calibrated gets rather expensive.
Against this I have seen comments - avoid a USB mic. This may be down to latency problems which the analogue audio route through PC's also have. USB spec's keep improving so earlier views may be dated. Transfer rates have got faster and faster suggesting latency is reduced as well.
Which Xmax. The one Boxsim loads that is intended to be used. Also Xmech or the excursion limit that they state in their data sheets. That to see what SPL levels could turn out to be a disaster and also consider max drive wattage.
😉 I should have said single driver / chassis. Looks like there is no way of doing that. I can theorise a way with a moving microphone while measuring it's distance from a baffle but various factors indicate it wont be accurate enough. A simple cheap laser measurement device can measure the distance to the baffle to within 2mm.