15" Drivers, Part 2: Mathcad sims versus SPL graphs

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Having been looking at the following threads, and planning my own OB system (visaton B200 + 15" bass driver), I decided to do some sims, at MJKs suggestion, to compare them with the manufactures datasheets.

These are what inspired me:

http://www.diyaudio.com/forums/showthread.php?s=&threadid=108751

http://www.diyaudio.com/forums/showthread.php?s=&threadid=98956&highlight=

http://www.diyaudio.com/forums/showthread.php?s=&threadid=110452&perpage=30&pagenumber=1

Currently, Vix is having problems integrating his Beta 15 with a B200, and I've been looking into using a Dayton IB15 in its place. MJK's recent OB design article suggests that the Alpha 15 would be a good bet.

To me, the problem seems to be a discrepancy between the simulated results and the measured results. I've run the simulations, and have made some pictures to compare the simulation and SPL graphs. These follow:
 

Attachments

  • one way sims.gif
    one way sims.gif
    44 KB · Views: 357
Those last three graph are for the IB15, Alpha 15 and Gamma15. The following two are for the Acoustic Elegance Lambda Dipole15 and the Knight 15, which are also considerations for OB use. All simulations were done in MJKs single driver OB simulator in Mathcad, changing only the T/S parameters. The baffle is 40" tall by 24" wide.
 

Attachments

  • onewaysims2.gif
    onewaysims2.gif
    40.8 KB · Views: 334
Now, the thing that really stands out in the simulated response is how well the Alpha 15 does, which has already been noted by MJK (although it hasn't been compared to non-Eminence drivers as of yet). The Alpha 15 gets about 6dB more output compared to any other driver at 50 and 100Hz. But the confounding thing is how the trend is reversed for the measured SPL levels - at the Alpha is lowest in output at 2 out of three frequencies. I wish the Lambda and Knight has fr graphs...

The first thing to question is if the SPL graphs were measured the same, they may not have been. I don't know. But the IB15 looks like a clear winner if you consider just the SPL graph - its quite flat to 20Hz, and displays no midrange peaks, which the Alpha has quite bad (it seems Vix was experiencing some problems because of this with his Betas).

So the question is, which do you trust, the simulated responses (I bet MJK does), or the measured SPL graphs? Also, how do you reconcile the very flat IB15 graph versus the more sloppy Alpha 15's?

And the biggest question, in the end, which is going to be a better driver (the speakers I'm putting together will be biamped with a computer XO/time alignment, so things like efficiency and fr are not as important - my goal with the bass is quality, not quantity).
 
cuibono said:
Now, the thing that really stands out in the simulated response is how well the Alpha 15 does, which has already been noted by MJK (although it hasn't been compared to non-Eminence drivers as of yet). The Alpha 15 gets about 6dB more output compared to any other driver at 50 and 100Hz. But the confounding thing is how the trend is reversed for the measured SPL levels - at the Alpha is lowest in output at 2 out of three frequencies. I wish the Lambda and Knight has fr graphs...

The first thing to question is if the SPL graphs were measured the same, they may not have been. I don't know. But the IB15 looks like a clear winner if you consider just the SPL graph - its quite flat to 20Hz, and displays no midrange peaks, which the Alpha has quite bad (it seems Vix was experiencing some problems because of this with his Betas).

So the question is, which do you trust, the simulated responses (I bet MJK does), or the measured SPL graphs? Also, how do you reconcile the very flat IB15 graph versus the more sloppy Alpha 15's?

And the biggest question, in the end, which is going to be a better driver (the speakers I'm putting together will be biamped with a computer XO/time alignment, so things like efficiency and fr are not as important - my goal with the bass is quality, not quantity).

Stimulation is one aspect of speaker design, the other is the measured response. Eg. I bought local generic woofers with reasonable Qts=0.55, the response curve on open baffle is not to expectation, ie it rolls off rather rapidly, hence I redesign and make it mounted on a U frame the response curve improves dramatically. The woofer sounds quite good even though the woofer is very cheap. I can get about 87dB at 50 Hz.

I would opt for the Dayton from the response curve view point. Each woofer has certain sound characteristics, Alpha may not have a better response curve however the tonal quality you may find is better than Dayton. POSSIBLY the Dayton with better response may sound drier. An example is Martin bought JBL driver which is more expensive than Alpha, possibly better response curve, he found the Alpha to sound better, hence recommend it ever since.

Hope this maybe of some assistance. Cheers.....
 
Well, I don't think Dayton or Eminence, when they measured their drivers, were using your proposed baffle to do it on. ;) So of course the simulated results and the IB measurements will be different. That's hardly a discrepancy. You wouldn't expect the drivers to give the same SPL plot as the advertised graph if you stuffed them into a small BR box, right? Well, the same principle doesn't vanish because they're mounted on an OB.

Generally your comparative table shows a small drop-off in SPLs in the LF, which is exactly what you'd expect from a driver mounted on a baffle where the acoustic short-circuit eventually starts causing cancellation. This is before room effects etc. are brought into the equation too, though baffles are somewhat less affected if positioned right as they activate fewer room resonant modes. Clearly, the worksheets are doing their usual superb job, but remember that you have to learn how to interpret the data they give you too. To give one example, if you see a BR giving a ruler-flat anechoic predicted and measured response, you can be certain it's no longer going to be ruler-flat in-room due to room-modes etc.

If you want simple, I'd generally be inclined to go for a high Q driver like the Alpha 15 as you'll be able to use a narrower baffle. The alternative approach, which is more complicated and harder to get right, is to use several drivers with high motor-power & some form of EQ on these or the WR driver they're supporting to balance out the SPLs. Both work, so YMMV.

What you're seeing, & the reason the Alpha is doing well in LF SPLS compared to the other units, is it's high Q. If you haven't read Martin's white paper on the subject, I suggest you look into it: http://www.quarter-wave.com/General/OB_Design.pdf
 
Scottmoose said:
Well, I don't think Dayton or Eminence, when they measured their drivers, were using your proposed baffle to do it on. ;) So of course the simulated results and the IB measurements will be different. That's hardly a discrepancy. You wouldn't expect the drivers to give the same SPL plot as the advertised graph if you stuffed them into a small BR box, right? Well, the same principle doesn't vanish because they're mounted on an OB.

Generally your comparative table shows a small drop-off in SPLs in the LF, which is exactly what you'd expect from a driver mounted on a baffle where the acoustic short-circuit eventually starts causing cancellation. This is before room effects etc. are brought into the equation too, though baffles are somewhat less affected if positioned right as they activate fewer room resonant modes. Clearly, the worksheets are doing their usual superb job, but remember that you have to learn how to interpret the data they give you too. To give one example, if you see a BR giving a ruler-flat anechoic predicted and measured response, you can be certain it's no longer going to be ruler-flat in-room due to room-modes etc.

If you want simple, I'd generally be inclined to go for a high Q driver like the Alpha 15 as you'll be able to use a narrower baffle. The alternative approach, which is more complicated and harder to get right, is to use several drivers with high motor-power & some form of EQ on these or the WR driver they're supporting to balance out the SPLs. Both work, so YMMV.


Hi Scott and cuibono,
The thing is, as I see it (no doubt un-qualified), that Dayton and Gamma hold up against the sim quite nice, Alpha not. Anomalies in the T/S parameters?

Peter
 
I doubt it. Eminence are usually pretty (very) accurate with their specs. It's a low Q unit (underdamped) so it should generally keep the LF up better on an OB than lower Q drivers. Martin's paper which I linked to above discusses this issue, with several examples, including a couple of these units.

So... a demon :devilr:-stration.

This is the Gamma 15 in a sealed cabinet (easiest to quickly knock togethewr -ignore the red plot, we're not interested in that; it's the broken blue dotted line that's important.
 

Attachments

  • gamma15.gif
    gamma15.gif
    5.6 KB · Views: 301
And this is the Alpha 15, same box (not that that matters). Note the blue plot which is the driver's IB response derived from the T/S parameters.

What do we see? The higher Q unit has the peak at Fc & lower mass-corner characteristic of all such comparatively under-damped drivers. This trait is mostly hidden by the smoothing Eminence apply to their published graph, and also by the way they measure them (that's purely an observation, not a critcism -there are many ways of doing both). Using these two graphs, I imagine you can see why the Alpha15 has the superior LF gain over the Gamma15, and the latter, better damped unit, with a higher mass-corner, will require a substantially larger / wider baffle to prop up it's lower frequencies.

Hope that helps
Scott
 

Attachments

  • alpha15.gif
    alpha15.gif
    5.7 KB · Views: 282
Be careful with the Eminence SPL response plots. Someplace I remember reading that these plots were for the drivers mounted in a box. If you look carefully at the plotted response for the Alpha 15A, does the low end roll-off seem consistent with a driver that has an fs~40 Hz and a Qts~1.2?

Try modeling the Alpha 15A driver in one of the free shareware programs using a huge box (essentially an infinite baffle) and compare the results against this plot. Take a look at Figure 4 in my recent OB article to see what the response should look like. Now compare these results to the Eminence plot and decide if it makes any sense. I think it will be obvious that the Eminence plot was made using some kind of enclosure.
 
Agreed. As Martin points out, the LF roll-off of the Alpha15 graph Eminence provide is not consistant with a high Q unit mounted in an infinite baffle, whatever their claims on the data sheet might be. Even old WinISD (bless it) gives the attached result for that driver with a back-chamber of the volume Eminence claim, so that's three completely different pieces of software all giving basically the same result. Eminence's specs are usually good, but their graphs... well, general message guys -take the published SPL graphs all manufacturers provide with a very healthy pinch of salt, and an even healthier dose of scepticism.
 

Attachments

  • alpha15 winisd.gif
    alpha15 winisd.gif
    9.4 KB · Views: 255
Thanks everyone for your replies.

Scottmoose - I wasn't expecting the sims to match the measured SPL graphs, but I was expecting them to trend similarly in their LF roll of. The discrepancy is between the Dayton and the Alpha - the Dayton rolls off in the sim, but not in the measured graph, while the Alpha's sim shows a higher response simulated than measured!

So the take-home message is to not trust the Alpha 15 measured SPL curve. But can anyone explain the Dayton response? Its almost ruler flat from 20Hz up. They must have had it in a special setup, no?

And going back to what Scottmoose said earlier, the Alpha will be easier to get right, and I should probably go that route. What is gained by using a higher power motor - better impulse response? Drier, tighter bass? I'm still worried about that 10dB peak from 1 to 3kHz in the Alphas...

Thanks much

Patrick
 
cuibono said:
Thanks everyone for your replies.

Scottmoose - I wasn't expecting the sims to match the measured SPL graphs, but I was expecting them to trend similarly in their LF roll of. The discrepancy is between the Dayton and the Alpha - the Dayton rolls off in the sim, but not in the measured graph, while the Alpha's sim shows a higher response simulated than measured!

So the take-home message is to not trust the Alpha 15 measured SPL curve. But can anyone explain the Dayton response? Its almost ruler flat from 20Hz up. They must have had it in a special setup, no?

And going back to what Scottmoose said earlier, the Alpha will be easier to get right, and I should probably go that route. What is gained by using a higher power motor - better impulse response? Drier, tighter bass? I'm still worried about that 10dB peak from 1 to 3kHz in the Alphas...

Thanks much

Patrick

You're welcome.

As a general rule, it's the bass end generally that are going to show the greatest differences between drivers, not the least! They have different mass-corners, giving different roll-offs, different resonant frequencies, so of course they'll behave in a completely different fashion. You're not comparing like with like either -the Dayton is a sub driver, not a bass driver. Note that it's Fs is about an octave lower than the other two.

Another general rule: don't trust any SPL graph from a manufacturer. They're trying to sell you something, so they'll always try to make the drivers appear as impressive as possible. They all will use different equipment and methods to measure the unitsm and also apply different levels of smoothing, and almost never say what these are. There is also, like in the case of the Alpha15, potential for daft errors to creep in, like in the case of it's graph (though I repeat, I wouldn't trust a manufacturers graph anyway). Comparisons between different units are very difficult to achieve, and generally mean you have to measure the units and their response yourself to ensure consistancy. Just basic scientific proceedure -only change one variable; but there are a host of unknows and changing ones here.

Yes, the Dayton will roll off more on the baffle than if on / in an infinite baffle -again, this is not a like with like comparison. Overall bass response depends on the enclosure. Different cabinets will give different results. An IB will be different to a BR, both will be different to a TL, these three will be different in their own ways to an MLTL, that will be different to the above and to an ML TQWT, which will be different to the whole lot, just like an aperiodic cabinet will be different to all of them, while an open-baffle will be gleefully jumping up and down and shouting 'I'll give different results from all the aforementioned too!'

In the case of an OB, due to the acoustic short-circuit of the front & rear radiation, the LF response of a driver on an OB will naturally roll off more than it would otherwise. Now, if you have an underdamped, high Q driver like the Alpha15 (forget the manufacturers 'measurement', it's worthless, use the computer generated plots I provided above) what do you see? It peaks in the LF, so will naturally have more gain when positioned on a baffle because its sensitivitiy is naturally higher in this region. That's why high Q units are a good choice for OBs -you can use a narrower baffle.

Theoretically low Q drivers will give the best transient response etc as drivers; but, you choose the unit that's appropriate to what you want to do. You can use a narrower OB with a high-Q unit than you can with a low Q, as you can see (re-run the dayton sim, but triple the baffle width & see what happens).

Re the peak in the Alphas, I'd have hoped you would have crossed over well before then to another unit, so it shouldn't be an issue as they won't be doing anything up there.
 
cuibono,

If I were you I will set one speaker(not stereo) at a time, Dayton and Alpha, listening to the setup(*). pick the one you like and sell the other 2nd hand. You may have to redesign the x-over for each, my hunch is that the Dayton will give a stronger bass(an impression of being punchier favourable to rock music) and the alpha a softer one(maybe more musical?, another hunch). It is personal preference from there and the type of music you listen to.

that way you get the best of both worlds, and don't lose much money, only time.

* you have already one of each for testing I presume.
 
Hi cuibono, I tried to reply yesterday but my ISP wasn't allowing uploads for some reason. I have not got the latest MJK software but that will be remedied shortly. I was going to suggest what Scottmoose and MJK have said and that is looking at the blue line on the sheets. I have done this recently with Alpha 15 and the Pyle Pro PPA 15 which chops is enthusiastic about. The Pyle does not have the hump at resonance as its Qts is about 0.67 I think. Unfortunately it is not as sensitive. I don't know how it behaves higher up in the range, or what it sounds like. Perhaps chops could come in on this. I have to agree with him though that it looks like a promising cheap speaker to sort out ideas with.
jamikl
 
it seems, unfortunately, that the only way to really know the difference between drivers is by trying them out. so much for science:rolleyes: Being a novice, and not really into pumping bass, I'll probably go with the Eminence (and save some bucks too). That Pyle Pro is quite cheap! Cheap enough where trying one out is plausable...

When I get the Alphas, I'll let everyone know how it goes.
 
Well, we've explained the technical differences based on the plots provided. No sim will tell you what it sounds like of course, though you'll get a general idea from the response, impedence curves etc.

Of the drivers you list, the Alpha will have more usable LF gain than any of the other units you refer to, so you're in the positive by selecting that driver over the other units for this application. The entire point & reason the Alpha works well is because its high Q & attendant hump at resonance goes some way to compensating for the natural roll-off of the drivers on the baffle. I'm not quite sure why people are getting the idea that, say, the Dayton unit, will have stronger bass in this application, because it won't, as the theory points out & is clearly indicated by the MathCad SPL graphs.

Eminence's engineering work is usually good -they're well designed and built drivers -they're relatively cheap because Eminence are a big company & churn out a lot. Economies of scale work in your favour. :)
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.