Large Coil Simulation Accuracy Issue and Adjustment - AVS Forum | Home Theater Discussions And Reviews
Baselworld is only a few weeks away. Getting the latest news is easy, Click Here for info on how to join the Watchuseek.com newsletter list. Follow our team for updates featuring event coverage, new product unveilings, watch industry news & more!



Forum Jump: 
 10Likes
 
Thread Tools
Old 08-24-2015, 08:14 AM - Thread Starter
Advanced Member
 
diy speaker guy's Avatar
 
Join Date: Jan 2011
Posts: 854
Mentioned: 25 Post(s)
Tagged: 0 Thread(s)
Quoted: 265 Post(s)
Liked: 234
Large Coil Simulation Accuracy Issue and Adjustment

Without adjustment, Hornresp and other programs don't accurately simulate drivers with very large voice coils. With a minor adjustment most of the variation can be eliminated.

Previously thought to be an issue caused by high inductance, it is now suggested that high inductance and simulation inaccuracy are both symptoms of this larger issue of very large voice coils.

This adjustment is now included in the newest version of Hornresp, just check the "Large Voice Coil" checkbox and the adjustment is made automatically. There's also a formula you can use in other simulators if you don't use Hornresp.

There's also a paper that I wrote on the issue, how to tell which drivers are affected, a description of the process leading up to the formulation of the new adjustment method, and 30 examples of unadjusted and adjusted sims compared to real professional measurements. The sample size is 22 drivers, some measured in multiple enclosures. The simulation accuracy of the new adjustment method is very good, even better than the original method.

This is the link to the download page for the paper - https://sites.google.com/site/amateu...and-adjustment

It's a long paper but it's mostly pictures, it doesn't take long to read.

Special thanks to the contributors, LTD02, David McBean and Josh Ricci.

I can answer any questions on this topic but I have no internet at home at this time. I'll try to get online once a day if I can.



Image courtesy of LTD02.
LTD02, eXa and 3ll3d00d like this.

Last edited by diy speaker guy; 08-24-2015 at 09:37 AM.
diy speaker guy is offline  
Sponsored Links
Advertisement
 
Old 08-24-2015, 09:34 AM
AVS Addicted Member
 
Stereodude's Avatar
 
Join Date: Jan 2002
Location: Detroit Metro Area
Posts: 13,149
Mentioned: 23 Post(s)
Tagged: 0 Thread(s)
Quoted: 2552 Post(s)
Liked: 1687
Your link to the paper doesn't work.

I'd like to refrain from making too many comments until I've read the paper, but why is this a function of voice coil size? There are a variety of pro audio woofers / subwoofers with large to very large voice coils that behave in line with their simulations.

Edit: Working link

https://sites.google.com/site/amateu...and-adjustment

Last edited by Stereodude; 08-24-2015 at 09:38 AM.
Stereodude is online now  
Old 08-24-2015, 09:37 AM - Thread Starter
Advanced Member
 
diy speaker guy's Avatar
 
Join Date: Jan 2011
Posts: 854
Mentioned: 25 Post(s)
Tagged: 0 Thread(s)
Quoted: 265 Post(s)
Liked: 234
https://sites.google.com/site/amateu...and-adjustment

Larger coils need more adjustment.

I suspect the pro drivers are using larger diameter coils for cooling and smaller gauge wire compared to the very high xmax car audio type drivers which typically have 3 inch diameter voice coils and larger wire (smaller gauge) and much longer coils, in other words way more wire. I'm not a driver designer though.

Anyway, there are 22 example drivers and my adjustment worked well on all of them to provide a more accurate sim, and the adjustment method is scaled based on normalized inductance because that's what worked.

Last edited by diy speaker guy; 08-24-2015 at 09:46 AM.
diy speaker guy is offline  
Old 08-24-2015, 10:51 AM
AVS Addicted Member
 
LTD02's Avatar
 
Join Date: Oct 2003
Posts: 19,887
Mentioned: 277 Post(s)
Tagged: 1 Thread(s)
Quoted: 2032 Post(s)
Liked: 2260
nice work dsg and congrats on bringing the paper to conclusion. lots of work involved.


definitely an issue to be aware of when modeling these types of drivers, particularly in horns, but also in ported, and to a less extent even in sealed where model accuracy matters.

Listen. It's All Good.
LTD02 is online now  
Old 08-24-2015, 10:58 AM - Thread Starter
Advanced Member
 
diy speaker guy's Avatar
 
Join Date: Jan 2011
Posts: 854
Mentioned: 25 Post(s)
Tagged: 0 Thread(s)
Quoted: 265 Post(s)
Liked: 234
Quote:
Originally Posted by LTD02 View Post
nice work dsg and congrats on bringing the paper to conclusion. lots of work involved.


definitely an issue to be aware of when modeling these types of drivers, particularly in horns, but also in ported, and to a less extent even in sealed where model accuracy matters.
Thanks again to you for your very important contributions. The adjustment method and the write up are both better because of your involvement.
diy speaker guy is offline  
Old 08-24-2015, 11:11 AM
AVS Addicted Member
 
Stereodude's Avatar
 
Join Date: Jan 2002
Location: Detroit Metro Area
Posts: 13,149
Mentioned: 23 Post(s)
Tagged: 0 Thread(s)
Quoted: 2552 Post(s)
Liked: 1687
While I appreciate your effort and the work you put into it, I have to question the validity of the conclusion when you discarded all the "outlier" cases where the T/S model actually matches reality. IMHO, if your correction formula and underlying premise is correct, it should work for any large voice coil, not just the 22 cases you cherry picked from data-bass.
Stereodude is online now  
Old 08-24-2015, 11:40 AM - Thread Starter
Advanced Member
 
diy speaker guy's Avatar
 
Join Date: Jan 2011
Posts: 854
Mentioned: 25 Post(s)
Tagged: 0 Thread(s)
Quoted: 265 Post(s)
Liked: 234
Quote:
Originally Posted by Stereodude View Post
While I appreciate your effort and the work you put into it, I have to question the validity of the conclusion when you discarded all the "outlier" cases where the T/S model actually matches reality. IMHO, if your correction formula and underlying premise is correct, it should work for any large voice coil, not just the 22 cases you cherry picked from data-bass.
The type of drivers studied are considerably different in just about every way than "normal" or pro drivers. Pro drivers max out at around 14mm xmax and usually won't move past 20mm. This type of very large excursion driver that was studied has a coil much larger (at the very least it's much longer and quite often physically larger in all ways except diameter) and different than a pro driver, very likely usually using a larger wire (smaller gauge) too and can usually move twice as far as "normal" drivers.

Anyway, results don't lie. All the pro drivers on data-bass sim accurately, all the large excursion drivers don't. The proposed adjustment method works 100 percent of the time to make these drivers sim accurately in any enclosure.

I didn't cherry pick drivers to study. I used every single driver of this type that was on data-bass.com that had measurements (frequency response and t/s parameters) enclosure details (so I could sim it) and was powered by Ricci's external amp (to make sure no response curve shaping was applied by the amp).

When I say "driver of this type" I think it's pretty easy to tell which drivers fall into this category. Most of them you can tell just by looking at them, if they have higher than average xmax and power handling they probably belong in this category, and if they have very high normalized inductance they probably belong in this category. There are pretty good reasons why pro drivers are not in this category and most of the time you can tell just by looking which category a driver belongs in. It's well beyond the scope of this study to investigate coil physical properties and dimensions (that info can be hard to find for a lot of drivers), but it's still fairly easy to tell which category any given driver is in without this info.

Last edited by diy speaker guy; 08-24-2015 at 12:01 PM.
diy speaker guy is offline  
Old 08-24-2015, 12:47 PM
AVS Addicted Member
 
Stereodude's Avatar
 
Join Date: Jan 2002
Location: Detroit Metro Area
Posts: 13,149
Mentioned: 23 Post(s)
Tagged: 0 Thread(s)
Quoted: 2552 Post(s)
Liked: 1687
Quote:
Originally Posted by diy speaker guy View Post
I didn't cherry pick drivers to study. I used every single driver of this type that was on data-bass.com that had measurements (frequency response and t/s parameters) enclosure details (so I could sim it) and was powered by Ricci's external amp (to make sure no response curve shaping was applied by the amp).
Sure you did. You've made an arbitrary and somewhat fluid and flexible distinction in your classification of the drivers that allows you to claim that all the drivers that meet said criteria are affected.

Lets try this... Are these drivers "large voice coil" drivers that need correction?

Acoustic Elegance AV15-H
Infinity 1260w
Alpine SWR-1522D

Why or why not?
Stereodude is online now  
Old 08-24-2015, 02:25 PM - Thread Starter
Advanced Member
 
diy speaker guy's Avatar
 
Join Date: Jan 2011
Posts: 854
Mentioned: 25 Post(s)
Tagged: 0 Thread(s)
Quoted: 265 Post(s)
Liked: 234
I'm not a driver designer and I've never done an in depth study into specific drivers and the parts you can't see that make up these drivers. I don't have full voice coil info on any of the drivers in the study or any other drivers. But yet clearly there are some drivers that sim accurately and some that don't. And from the drivers that data-bass measured I can tell which are which just by looking at a picture of them. And I believe the term "inductance hump" was around long before I was. I don't know much about any of the 3 drivers you would like me to comment on. At least one of them is probably a borderline case in which a very small amount of adjustment might help.

So let's try this a different way. In your opinion, what is it that makes some drivers not sim accurately? Initially I thought it was just inductance, due to the fact that all high inductance drivers don't sim well and the term "inductance hump" is pretty persuasive. But I discovered there are some very low inductance drivers that don't sim well, but they do have large displacement capability. If you can tell me exactly what causes these drivers to not sim accurately you will be included as a contributor to paper and the paper will be corrected. But for now I have no reason to believe it's anything other than the voice coil or motor. For awhile, after finding out it wasn't purely inductance causing this, I was calling the adjustment the large coil/motor issue, but I shortened it recently. Maybe I shouldn't have? It really can't be caused by anything other than the voice coil, the motor or the suspension (spider and surround) and I'm pretty sure it's not the suspension. It's not the basket, it's not the cone and it's not the vc former. And that covers all the parts I think. Nothing other than VC and motor seem relevant.

Regardless, once you identify drivers that don't sim accurately, this adjustment works 100 percent of the time to fix the sim. And most of the time I can tell which drivers aren't going to sim accurately just by looking at a picture of them. The fact that I can go through all the drivers on data-bass and know just by looking which category the drivers are in has to be a pretty strong clue that there are categories and something about the appearance of the drivers indicates which category a driver will fall into. And the visual cues that identify these drivers all point to large stroke. On the other hand, as I made very clear in the paper there are borderline cases and some drivers that are not so easy to judge, and you picked three of them to challenge me with. Do a sealed box measurement, show me the response and I'll tell you which category they are in. And that applies to all drivers that can't be evaluated by any other means.

BUT I'm pretty sure if I had details on voice coil wire type, wire gauge, voice coil geometry (length, diameter and number of layers) that info would be all you need to know. At this time I can't prove that since I don't have that info for any of the drivers. And since I have no proof I could be wrong, but I seriously doubt it's anything other than the voice coil or the motor causing this. Maybe one day I'll try to find voice coil and motor info on all the sample drivers (and a bunch of "normal" drivers to compare to), but the point of all this was to fix the sims, not to pinpoint the cause of the problem. I'm pretty sure though, if you do find this info you will find that the drivers with the largest coils have the largest inductance and need the most correction to get an accurate sim.

Last edited by diy speaker guy; 08-24-2015 at 02:38 PM.
diy speaker guy is offline  
Old 08-24-2015, 03:09 PM
AVS Addicted Member
 
LTD02's Avatar
 
Join Date: Oct 2003
Posts: 19,887
Mentioned: 277 Post(s)
Tagged: 1 Thread(s)
Quoted: 2032 Post(s)
Liked: 2260
Quote:
Originally Posted by Stereodude View Post
Lets try this... Are these drivers "large voice coil" drivers that need correction?
Acoustic Elegance AV15-H
Infinity 1260w
Alpine SWR-1522D
Why or why not?

unlikely. - copper sleeve on the pole short circuits the issue
minor. - 300 watt, 2" coil 1.3" tall. not a very big coil.
most likely. - 4 layer coil, 2.6" diam x 2" tall coil, 750 watts rms 2kw peak.


haven't seen the alpine measured close mic (or outdoors). would be curious if you have a sweep on that one to see what it look like.

Listen. It's All Good.
LTD02 is online now  
Old 08-24-2015, 05:24 PM
AVS Addicted Member
 
Stereodude's Avatar
 
Join Date: Jan 2002
Location: Detroit Metro Area
Posts: 13,149
Mentioned: 23 Post(s)
Tagged: 0 Thread(s)
Quoted: 2552 Post(s)
Liked: 1687
Quote:
Originally Posted by diy speaker guy View Post
So let's try this a different way. In your opinion, what is it that makes some drivers not sim accurately? Initially I thought it was just inductance, due to the fact that all high inductance drivers don't sim well and the term "inductance hump" is pretty persuasive. But I discovered there are some very low inductance drivers that don't sim well, but they do have large displacement capability. If you can tell me exactly what causes these drivers to not sim accurately you will be included as a contributor to paper and the paper will be corrected.
In the absence of contrary evidence I still think it's an inductance issue. Some of the driver's you've characterized have low Le numbers (at 1kHz). That doesn't mean they have low inductance between 10-100Hz. I will readily admit it's a hard nut to crack. We don't have LR2 or LR4 models for most (all?) of the drivers. We don't have a way to use LR4 models (that I know of) which REW can generate. We don't have a way to generate LR2 models which Unibox and WinISD can use (I'm pretty sure the latest WinISD supports LR2 inductance models). JohnM has indicated he can add the LR2 model to REW, but no idea on the ETA. It hasn't been in any of the subsequent versions released. It would be interesting if David McBean could add support for LR2 and LR4 inductance models to Hornresp. Then the LR4 models we can generate with REW could be used to on some test cases to see how close the LR2 and/or LR4 inductance models get to predicting actual behavior. If the better inductance models fail to predict response accurately then we can definitively say there's more than just an inductance issue, but I just don't think we're yet.

Quote:
...On the other hand, as I made very clear in the paper there are borderline cases and some drivers that are not so easy to judge, and you picked three of them to challenge me with. Do a sealed box measurement, show me the response and I'll tell you which category they are in. And that applies to all drivers that can't be evaluated by any other means.
That was sort of my point. If I have to measure them in a box to tell if they're affected it sort of defeats the purpose. At that point I already have bought the driver... I'd like to know before buying the driver. It seems like you've made a reactive correction. Once you know a driver doesn't match the model you can correct for it reasonably well, but you don't have a definitive empirical way to determine if it will match the model ahead of time. At best it seems like you're making speculative generalizations prior to seeing a measurement based on characteristics the drivers that vary seem to have.
Stereodude is online now  
Old 08-24-2015, 05:35 PM
AVS Addicted Member
 
Stereodude's Avatar
 
Join Date: Jan 2002
Location: Detroit Metro Area
Posts: 13,149
Mentioned: 23 Post(s)
Tagged: 0 Thread(s)
Quoted: 2552 Post(s)
Liked: 1687
Quote:
Originally Posted by LTD02 View Post
unlikely. - copper sleeve on the pole short circuits the issue
AFAIK, the AV15-H has a 2.5" coil. It has a 1kW RMS power rating, and has 23mm of xmax. Per the general guidelines in the paper it doesn't seem like a border case or a no.

On the other hand, the datasheet for the Infinity 1260w seems to show it does a bit of a "inductance hump" not reflected in the T/S model of that box despite the 2" coil, fairly low power, and modest xmax.

35.4L sealed box:
Attached Thumbnails
Click image for larger version

Name:	Infinity 1260w.png
Views:	342
Size:	14.2 KB
ID:	905402  
Stereodude is online now  
Old 08-24-2015, 06:08 PM - Thread Starter
Advanced Member
 
diy speaker guy's Avatar
 
Join Date: Jan 2011
Posts: 854
Mentioned: 25 Post(s)
Tagged: 0 Thread(s)
Quoted: 265 Post(s)
Liked: 234
Quote:
Originally Posted by Stereodude View Post
In the absence of contrary evidence I still think it's an inductance issue. Some of the driver's you've characterized have low Le numbers (at 1kHz). That doesn't mean they have low inductance between 10-100Hz. I will readily admit it's a hard nut to crack. We don't have LR2 or LR4 models for most (all?) of the drivers. We don't have a way to use LR4 models (that I know of) which REW can generate. We don't have a way to generate LR2 models which Unibox and WinISD can use (I'm pretty sure the latest WinISD supports LR2 inductance models). JohnM has indicated he can add the LR2 model to REW, but no idea on the ETA. It hasn't been in any of the subsequent versions released. It would be interesting if David McBean could add support for LR2 and LR4 inductance models to Hornresp. Then the LR4 models we can generate with REW could be used to on some test cases to see how close the LR2 and/or LR4 inductance models get to predicting actual behavior. If the better inductance models fail to predict response accurately then we can definitively say there's more than just an inductance issue, but I just don't think we're yet.
The Dayton UM has normalized inductance (Le/Re) of .36. Dayton HO has .4, MTX 9515 has .4. These are remarkably low, yet all 3 have this issue. Maybe you are right, maybe they do have incredibly high inductance at 10 or 100 hz despite their low inductance at 1khz. I don't have any of these drivers so I can't test that theory.

Please believe that I don't want to spread misinformation and I just want to get this right. At this point, with the remarkably low 1khz normalized inductance that these 3 drivers have I can't characterize this as a high inductance issue. I would welcome any evidence anyone has and if this turns out to be the case I will definitely change the paper. At this point I still attribute this to the large coil and maybe the coil's interaction with the large motor.

Anyway, maybe 8 out of 10 drivers can be reliably judged based on appearance, xmax and power rating, and normalized inductance. If there's any question a quick measurement will sort the rest out. And again, it's way beyond the scope of what I was doing to pinpoint the cause of the issue, this was all about fixing the sim. Maybe I overstepped with the assumption that this is not an inductance issue but the information I have now (Le/Re for the 3 drivers mentioned) suggests it's not.

I think McBean has already been asked about complex inductance and said no. He's willing to do changes and upgrades but some things would require massive changes to his programming and too much time.

Quote:
That was sort of my point. If I have to measure them in a box to tell if they're affected it sort of defeats the purpose. At that point I already have bought the driver... I'd like to know before buying the driver. It seems like you've made a reactive correction. Once you know a driver doesn't match the model you can correct for it reasonably well, but you don't have a definitive empirical way to determine if it will match the model ahead of time. At best it seems like you're making speculative generalizations prior to seeing a measurement based on characteristics the drivers that vary seem to have.
You have a good point, I don't have a definitive way to determine definitively which drivers are affected except measuring frequency response, and the borderline cases are especially hard to judge without that FR measurement. But if people start paying attention to this we can start keeping track and have a database. Maybe I'll have a look around and see if any of these drivers have impedance curves on their published data sheets. Remarkably my internet seems to be working tonight.

The sealed box FR measurement doesn't really defeat the purpose, unless you just want to make a sealed box in which case it definitely does. But if you wanted to make a 125 cubic foot horn for example, if you are able to quickly model the driver in a sealed box and figure out what sim adjustment to make to make sure it will sim accurately in that 125 cubic horn before building it, I think that's a small price to pay.
diy speaker guy is offline  
Old 08-24-2015, 06:26 PM - Thread Starter
Advanced Member
 
diy speaker guy's Avatar
 
Join Date: Jan 2011
Posts: 854
Mentioned: 25 Post(s)
Tagged: 0 Thread(s)
Quoted: 265 Post(s)
Liked: 234
Quote:
Originally Posted by Stereodude View Post
AFAIK, the AV15-H has a 2.5" coil. It has a 1kW RMS power rating, and has 23mm of xmax. Per the general guidelines in the paper it doesn't seem like a border case or a no.

On the other hand, the datasheet for the Infinity 1260w seems to show it does a bit of a "inductance hump" not reflected in the T/S model of that box despite the 2" coil, fairly low power, and modest xmax.

35.4L sealed box:
The TC Pro 5100 is an even more extreme example of a driver that is not affected by this issue. It's got rising response up to 100 hz (no inductance hump) in it's small sealed box as measured by data-bass. It's Le/Re is .71, a bit higher but very close to the CSS SDX 15 which shows a massive inductance hump in it's measurement even though it's a smaller driver with a presumably smaller coil and a tiny bit less normalized inductance. I didn't sim or study the Pro 5100, maybe it's measurement is a bit more high qtc than a sim would indicate in which case it might be a bit affected by this issue but just by looking at the measurement it doesn't look like it.

So I don't really have an answer for the exceptions except to say that in the paper I did say there were exceptions to every rule and that I really don't know what's in these drivers - maybe it's shorting rings or extreme motor design that cause exceptions like this. The TC 5100 does have a very small picture of the VC provided by Ricci on the site but other than that I don't know anything about it's coil. Maybe it is shorting rings and the neo motor that keeps this driver in line. Maybe it's an issue with motor design and gap width. These are all things that can be investigated further, but for now I think it's a large coil issue. Maybe expensive drivers like the TC Pro 5100 can avoid the issue but it's the only driver with that much displacement that I've seen without the inductance hump in it's sealed box measurement.

Last edited by diy speaker guy; 08-24-2015 at 06:31 PM.
diy speaker guy is offline  
Old 08-24-2015, 06:57 PM
AVS Addicted Member
 
LTD02's Avatar
 
Join Date: Oct 2003
Posts: 19,887
Mentioned: 277 Post(s)
Tagged: 1 Thread(s)
Quoted: 2032 Post(s)
Liked: 2260
someone was kind enough to take a sweep of a 1260. i forget who.


i don't recall the exact setup. for some reason I'm thinking this was one of the small sealed cube builds. there is some rounding in the response. i think if you plug the numbers in, it will work pretty well, but the amount of the correction is minor. nothing like the re-xxx.


Attached Thumbnails
Click image for larger version

Name:	1.png
Views:	352
Size:	188.8 KB
ID:	905602  

Listen. It's All Good.
LTD02 is online now  
Old 08-24-2015, 07:01 PM
AVS Addicted Member
 
LTD02's Avatar
 
Join Date: Oct 2003
Posts: 19,887
Mentioned: 277 Post(s)
Tagged: 1 Thread(s)
Quoted: 2032 Post(s)
Liked: 2260
@Stereodude
"AFAIK, the AV15-H has a 2.5" coil. It has a 1kW RMS power rating, and has 23mm of xmax. Per the general guidelines in the paper it doesn't seem like a border case or a no."


the pole has a full copper sleeve. that makes it a real oddball. it should still work ok in the model because the le/re adjustment will be next to nothing. as a result the adjusted bl will pretty much just be the bl and the flat t/s response will pretty much translate into a flat measured response (which it does).

Listen. It's All Good.
LTD02 is online now  
Old 08-25-2015, 11:50 AM
AVS Special Member
 
A9X-308's Avatar
 
Join Date: Mar 2008
Location: Australia
Posts: 6,997
Mentioned: 12 Post(s)
Tagged: 0 Thread(s)
Quoted: 962 Post(s)
Liked: 683
Quote:
Originally Posted by diy speaker guy View Post
Pro drivers max out at around 14mm xmax and usually won't move past 20mm.
Incorrect. The Xmax/Xmech ratio in pro drivers is typically much larger.
A9X-308 is offline  
Old 08-26-2015, 06:43 AM - Thread Starter
Advanced Member
 
diy speaker guy's Avatar
 
Join Date: Jan 2011
Posts: 854
Mentioned: 25 Post(s)
Tagged: 0 Thread(s)
Quoted: 265 Post(s)
Liked: 234
Quote:
Originally Posted by A9X-308 View Post
Incorrect. The Xmax/Xmech ratio in pro drivers is typically much larger.
Published Xmech or Xlim is typically much larger than 20 mm, that is correct, but there's a big technicality here. Published Xmech is usually reported as the distance the cone must travel before it hits the back plate and causes physical damage. That doesn't mean the cone can actually travel that far, in fact it usually can't. The suspension often limits the excursion to around 20 mm even if published Xmech is much higher than 20 mm. And even if the suspension doesn't limit excursion, typically the coil will be right out of the gap long before it gets to Xmech. That's a danger situation and you really shouldn't go that far.

If you look at the B&C 21sw152-4 spec sheet it proudly states 60 mm peak to peak (30 mm one way) excursion before damage. But if you look at Ricci's comments on data-bass, this driver can't actually move anywhere near that far. Copied and pasted from data-bass, bold added by me -

At about 18 to 20mm of excursion in one direction the cone will simply not go any further despite further increases in power.
This phenomenon may be that the suspension is at its maximum position or there may possibly be some magnetic braking built into the coil design as aluminum shorting circuits on the former above the coil but the coil is completely beyond the gap at 21mm anyways. Alternatively the driver could also simply be running out of useable force at that displacement level. Whichever the reason, we simple could not get the driver to develop further excursion even in free air at very high voltage levels at 10 cycles.

This is fairly typical, but this doesn't really matter anyway. It's pretty clear that the ultra high displacement drivers are very different than pro drivers and that's the point I was trying to make.
diy speaker guy is offline  
Old 08-26-2015, 08:43 AM
AVS Special Member
 
rhodesj's Avatar
 
Join Date: Apr 2005
Posts: 1,641
Mentioned: 31 Post(s)
Tagged: 0 Thread(s)
Quoted: 750 Post(s)
Liked: 422
Quote:
Originally Posted by LTD02 View Post
someone was kind enough to take a sweep of a 1260. i forget who.


i don't recall the exact setup. for some reason I'm thinking this was one of the small sealed cube builds. there is some rounding in the response. i think if you plug the numbers in, it will work pretty well, but the amount of the correction is minor. nothing like the re-xxx.
I remember. It was a 38.6L sealed box. And the resulting TS paramters from those sweeps:

But even with those TS paramters the model still needed a little correction.
LTD02 and eXa like this.
rhodesj is online now  
Old 08-26-2015, 10:13 AM
AVS Addicted Member
 
LTD02's Avatar
 
Join Date: Oct 2003
Posts: 19,887
Mentioned: 277 Post(s)
Tagged: 1 Thread(s)
Quoted: 2032 Post(s)
Liked: 2260
@rhodesj
thanks man.

Listen. It's All Good.
LTD02 is online now  
Old 08-26-2015, 11:47 AM
AVS Special Member
 
A9X-308's Avatar
 
Join Date: Mar 2008
Location: Australia
Posts: 6,997
Mentioned: 12 Post(s)
Tagged: 0 Thread(s)
Quoted: 962 Post(s)
Liked: 683
Quote:
Originally Posted by diy speaker guy View Post
Published Xmech or Xlim is typically much larger than 20 mm, that is correct, but there's a big technicality here. Published Xmech is usually reported as the distance the cone must travel before it hits the back plate and causes physical damage. That doesn't mean the cone can actually travel that far, in fact it usually can't.
Bollocks. I'm not referring to your theoretical experience, but my actual experience of watching drivers excurse that far under fault conditions, and still test fine post show and remain in service.
Bill Fitzmaurice likes this.
A9X-308 is offline  
Old 08-26-2015, 03:17 PM - Thread Starter
Advanced Member
 
diy speaker guy's Avatar
 
Join Date: Jan 2011
Posts: 854
Mentioned: 25 Post(s)
Tagged: 0 Thread(s)
Quoted: 265 Post(s)
Liked: 234
That's fine with me if you want to run your pro drivers so hard the coil comes right out of the gap. That doesn't make pro drivers the same as high displacement drivers.

The B&C 21sw152-4 is right out of the gap at 21 mm according to Ricci's actual experience. Most of the drivers in the study I did are only 2/3 of the way to their published xmax at 21 mm. They are considerably different than pro drivers.

Not sure what point you are trying to make.
diy speaker guy is offline  
Old 08-26-2015, 04:12 PM
AVS Special Member
 
A9X-308's Avatar
 
Join Date: Mar 2008
Location: Australia
Posts: 6,997
Mentioned: 12 Post(s)
Tagged: 0 Thread(s)
Quoted: 962 Post(s)
Liked: 683
Quote:
Originally Posted by diy speaker guy View Post
That's fine with me if you want to run your pro drivers so hard the coil comes right out of the gap.
Your reading and comprehension skills are also poor. I said under fault conditions: note the underlined word. Please point to where I said I wanted to do it.

Quote:
Originally Posted by diy speaker guy View Post
Not sure what point you are trying to make.
That you were incorrect in your global assertion a few posts back.
A9X-308 is offline  
Old 08-27-2015, 07:36 AM - Thread Starter
Advanced Member
 
diy speaker guy's Avatar
 
Join Date: Jan 2011
Posts: 854
Mentioned: 25 Post(s)
Tagged: 0 Thread(s)
Quoted: 265 Post(s)
Liked: 234
Quote:
Pro drivers max out at around 14mm xmax and usually won't move past 20mm.
That's the statement that you originally disagree with. When the word "usually" is used it's not a global assertion. The following post from me to clarify uses the word "usually" and "typically" twice each. I went out of my way to NOT make global assertions.

As far as running the coil right out of the gap, that does not guarantee driver failure. Even if it gets caught up and won't go back in the gap itself you might be able to push it back into the gap. I'll definitely give you that, I never disagreed with that or stated that was not true. But it's not an ideal situation even if it's a fault or accident and even if it doesn't result in driver failure.

This is the 3rd or 4th time I'm saying that pro drivers and high displacement drivers are very different, which is the only important point here. USUALLY pro drivers will be right out of the gap at excursion levels a high displacement driver will still be comfortably well within published xmax. Clearly there are exceptions but this is USUALLY the case.

Hopefully we can just drop this, pro drivers and high displacement drivers are clearly very different. I'm not saying one is better than the other or that pro drivers can't be driven well past their published xmax, they are just very different. I'm definitely not trying to say your experiences with running the coil out of the gap are untrue or even uncommon but there's a big difference between running a pro driver into fault conditions and what I was talking about.

Last edited by diy speaker guy; 08-27-2015 at 07:42 AM.
diy speaker guy is offline  
Old 08-29-2015, 06:32 PM - Thread Starter
Advanced Member
 
diy speaker guy's Avatar
 
Join Date: Jan 2011
Posts: 854
Mentioned: 25 Post(s)
Tagged: 0 Thread(s)
Quoted: 265 Post(s)
Liked: 234
@Stereodude - I know you were not happy with my "large coil" theory. @ everyone ... inductive reactance?

Here's a post from DJIM on diyaudio -

Quote:
The problem here is in the inductive reactance (XL), which opposes the change in current and becomes the dominant force here. You can increase or lower this force by increasing or lowering the force field (BL). If you look for driver indications: high Le and/or high Mms (for its size) in relation to a low EBP value.

Cheers,
Djim

Link to post - http://www.diyaudio.com/forums/subwo...ml#post4434166

Link to theory page on inductive reactance - http://www.learnabout-electronics.or...eactance61.php

This sounds reasonable but I don't have the math skills to verify or discredit this theory. My eyes go a bit hazy when talk turns to reactance, phasor sums, vectors and imaginary numbers. Is inductive reactance the problem? If so I'd be happy to update my paper to reflect this. If it is inductive reactance I think this still points to a large coil issue but at least it would be a better defined issue.

Last edited by diy speaker guy; 08-29-2015 at 06:39 PM.
diy speaker guy is offline  
Old 08-29-2015, 08:13 PM
AVS Addicted Member
 
Stereodude's Avatar
 
Join Date: Jan 2002
Location: Detroit Metro Area
Posts: 13,149
Mentioned: 23 Post(s)
Tagged: 0 Thread(s)
Quoted: 2552 Post(s)
Liked: 1687
That would jive with the better inductance models. You've got additional inductance and a resistance elements which are the two pieces of the phasor.
Stereodude is online now  
Old 06-30-2016, 03:27 PM
AVS Special Member
 
chrapladm's Avatar
 
Join Date: Mar 2010
Location: Adelaide, South Australia
Posts: 3,908
Mentioned: 9 Post(s)
Tagged: 0 Thread(s)
Quoted: 645 Post(s)
Liked: 237
Because I dont want to ruin some else's thread I will ask here.
Quote:
Originally Posted by diy speaker guy View Post
That's been there since the paper was released, McBean released the "Large Coil" tweak in Hornresp simultaneously with the release of the paper. You and McBean were instrumental in ironing out some of the finer points of the generic tweak, and you were the one that suggested tweaking Bl instead of Re, so thanks again for that.

Hornresp uses the generic tweak. The curve fit tweak should be a bit more accurate, looking at the spreadsheet the curve fit value for adjusted Bl on the UXL is 16.7. IIRC I sent you a copy of the spreadsheet that shows all the curve fit Bl adjustments for each tested driver as I was putting the paper together. (All the curve fit adjustments are only applicable when using data-bass measured t/s.) The curve fit adjustments should be a bit more accurate, there's only so much accuracy a generic tweak can have - but the generic tweak is really easy so that's what Hornresp uses. Also the generic tweak can be used with any driver, while the curve fit tweak needs to be tested for each driver individually and so far the data pool is only 22 drivers IIRC.

A couple of people commented that the UM and TC drivers didn't need the tweak. While it's true that they aren't extremely inaccurate if you don't use the tweak, you will get more accurate results if you do use the tweak. Both of these were tested in the paper and in both cases the simulated vs measured results for both were substantially more accurate when tweaked. Even the Dayton HO and MTX 9515 with incredibly low normalized inductance still benefited in accuracy when tweaked. All drivers tested in the paper simulated more accurately when tweaked compared to the measurements.


DIY SG what is the generic way of adjusting BL for all drivers?

I only ask because I was wondering how can we tell the difference of inductance between one design verse another?

If FI and Sundown use different VC's they must be different in their inductance. SO I wasnt sure if the HR large coil adjustment was just a general one. My simulated response shapes for my drivers I wanted to use have gone to crap now that I used the new adjustment. SO I am curious to know more about what exactly causes this inductance to see if its possible to have a new VC made for an existing motor for much less inductance.

Thanks again

Last edited by chrapladm; 06-30-2016 at 03:38 PM.
chrapladm is online now  
Old 06-30-2016, 03:55 PM - Thread Starter
Advanced Member
 
diy speaker guy's Avatar
 
Join Date: Jan 2011
Posts: 854
Mentioned: 25 Post(s)
Tagged: 0 Thread(s)
Quoted: 265 Post(s)
Liked: 234
Quote:
Originally Posted by chrapladm View Post
Because I dont want to ruin some else's thread I will ask here.

DIY SG what is the generic way of adjusting BL for all drivers?
The generic method involves adjusting the value of Bl. The formula for determining the new value of Bl to use in your sim is included in the paper linked in post 1. If you are using Hornresp you can let the program do the work for you, just use the Loudspeaker Wizard and check the "Large Coil" checkbox. Hornresp uses the same formula that's in the paper.

The generic formula is a scaled tweak based on normalized Le (Le/Re), the higher the normalized Le the lower the tweaked Bl value will be. Drivers like the Dayton HO don't need much Bl adjustment because the normalized Le is very low, drivers like some of the Fi products have absurdly high normalized Le so they need a sizable Bl adjustment to get a more accurate sim.

Quote:
I only ask because I was wondering how can we tell the difference of inductance between one design verse another?

If FI and Sundown use different VC's they must be different in their inductance. SO I wasnt sure if the HR large coil adjustment was just a general one. My simulated response shapes for my drivers I wanted to use have gone to crap now that I used the new adjustment. SO I am curious to know more about what exactly causes this inductance to see if its possible to have a new VC made for an existing motor for much less inductance.

Thanks again
Read the paper and you will see how the tweak was formulated. I took 30 measurements of 22 drivers, simulated the enclosures the drivers (with measured t/s) were measured in, in dimensionally correct enclosures and adjusted the Bl to curve fit the sim to the measurement. Then I plotted normalized inductance vs the Bl multiplier required to make the sim match the measurement and derived the formula from the plotted data.

So yes it's a generic formula. As you can see from the scatter chart in the paper some drivers match the trend line much better than others, so the generic formula will be more accurate for some drivers than others.

There's also a way to get a more accurate tweak - do your own curve fit. Measure your driver in a small sealed box. Sim the driver (with measured t/s) in that small sealed box. Adjust Bl until the sim matches the measurement. Then use that value of Bl for all successive sims.

When adjusting Bl the sim will come out a couple db below the measured value but the frequency response curve and all other graphs the simulator provides should be pretty accurate. Nothing was done to account for the couple db less sensitivity in a Bl adjusted graph, but you can just easily add a couple db to your tweaked sims.

The adjustment is not perfectly accurate, but as the 30 examples in the paper show, it's a lot more accurate than not doing the adjustment. The 30 examples in the paper are shown using the generic tweak - the formula in the paper as used by Hornresp. The curve fit method is even more accurate, that's how the generic version was formulated, I simulated each of the 30 measured enclosures, plotted the normalized Le vs Bl multiplier and derived the adjustment formula from that trend line.

Last edited by diy speaker guy; 06-30-2016 at 04:23 PM.
diy speaker guy is offline  
Old 06-30-2016, 04:16 PM - Thread Starter
Advanced Member
 
diy speaker guy's Avatar
 
Join Date: Jan 2011
Posts: 854
Mentioned: 25 Post(s)
Tagged: 0 Thread(s)
Quoted: 265 Post(s)
Liked: 234
Quote:
Originally Posted by chrapladm View Post
SO I am curious to know more about what exactly causes this inductance to see if its possible to have a new VC made for an existing motor for much less inductance.

Thanks again
The cause of the issue is unclear. I mean, it's clearly related to large voice coils and inductance has a scaled effect (higher normalized inductance = less simulation accuracy) but I have no suggestions on how to eliminate the issues while keeping the benefits of large coils. I spent zero time investigating WHY these effects occur but I did take a bit of time to point out what type of drivers are usually affected. If the excursion and/or power handling and/or normalized Le for a certain driver are much higher than average for the driver size - and especially if all 3 factors are present - (for example a 12 inch driver with 40 mm xmax and 4000 wrms power handling and normalized inductance = 4) then you can bet it's going to produce an extremely inaccurate sim unless you adjust for the large coil effect.

If you want to investigate this yourself, you can check the validity of the assertions in post 25, as suggested by a guy at diyaudio.

ALL drivers of this type that I checked were affected to some degree, even the ones with low normalized Le and low(ish) power and excursion, like the Dayton HO and UM and the MTX 9515, three drivers you wouldn't suspect. Even the TC LMS that everyone assumes sims accurately was much more accurate when the large coil was accounted for using my method.

Pro drivers do not seem to be affected much (if at all). Although you should note that I didn't specifically study any pro drivers. I did however note that none of the pro drivers measured at data-bass had the telltale "inductance hump" that all of these 22 large coil drivers that I did study had. The "inductance hump" is just a high q bump in response indicating the enclosure is too small to give a flat frequency response.

So if you want a driver that is not affected by this issue you might want a voice coil and motor more similar to a pro driver. But then you lose the massive excursion capability that I assume is the whole reason you are using Sundown in the first place.

This issue is not a big deal for sealed boxes (including IB), a bit of eq will fix everything right up. It's also not a huge problem with front loaded horns, as long as you account for the issue in the sim. The low knee gets a bit lower in frequency than expected but it's not a huge deal. But in ported boxes, transmission lines, and all other resonant enclosures that don't have a sealed chamber the results go really wonky if you don't account for the issue in the sim. You get a huge "inductance hump" above tuning and rolled off response on both ends. This is just an indication that the box is way too small. And since this issue presents as a weaker than expected motor, if the driver qts is already relatively high, it might not work at all in a resonant type enclosure, as the issue is synonymous with higher than expected qts, the effect of which can be pretty accurately fudged by lowering Bl.
diy speaker guy is offline  
Old 06-30-2016, 04:36 PM
AVS Special Member
 
chrapladm's Avatar
 
Join Date: Mar 2010
Location: Adelaide, South Australia
Posts: 3,908
Mentioned: 9 Post(s)
Tagged: 0 Thread(s)
Quoted: 645 Post(s)
Liked: 237
Ya I read the paper and adjusted my simulations accordingly. I then saw the large coil adjustment in HR after I had used your method and read your paper. You had mentioned a long time ago that the Zv3 had high inductance when I was simulating a dual 18 horn. I just never knew why they have such a high inductance.

So now I am going to see what coils work for minimizing inductance and design using those drivers. I am not worried about pro audio drivers and their small winding depth/heights. More focused on 30mm + drivers although I may just need to study the 5100's coil better.

Last edited by chrapladm; 06-30-2016 at 04:40 PM.
chrapladm is online now  
Sponsored Links
Advertisement
 
 
Thread Tools


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off