Rebuilding the Enigma...

So, I loaned out my Enigma subwoofer to my brother a few weeks ago for use in the background music system for his bar, and staff managed to destroy the driver (not too surprising).

I cut out the moving parts for the driver, and it looks like the fiberglass former fractured at one point. I suspect that's from the INF10 driver bottoming out - a particular flaw with that driver, so that's not too surprising. The bottom 2/3rds of the coil also seems to be blackened, and I'm a bit curious as to why the top 1/3rds is not. The driver has a vented pole piece, and the pole piece is t-shaped at the gap, but the driver is ordinary otherwise, e.g. no under-spider venting. It's rated at 100W, and was connected to an amp that also does 100W and, while the coil is a bit blackened, I think the failure was more likely due to mechanical damage from bottoming out too hard.

Rag Bose all you want - they seem to at least get speaker protection right. So I'm thinking about how I can add Bose-like speaker protection to this subwoofer, to make sure that power delivered to it does not exceed say 70W for a continuous period. PE doesn't seem to have circuit-board mountable lamps, and there's not enough information available on their polyswitches (e.g. insertion resistance) to really make an informed choice about them.

Any suggestions? I plan to raise the order for the new driver later today, once I've identified one that sims well in the current box (another big challenge).
 

Attachments

  • IMG-20190423-WA0003.jpg
    IMG-20190423-WA0003.jpg
    123.8 KB · Views: 906
Too bad the woofer has been damaged :(



You could add a limiter to the amplifier. If you think the limiter will be defeated, mount it inside of the amplifier chassis.
Audio Amp Power Limiter
Mechanical damage can be prevented by using a (at least 2nd order) high pass filter. PA woofers also are more robust.

Is the amplifier still working? The fact that only half of the coil is blackened could mean that it was subject to dc, which could have been caused by a blown amplifier.
 
Last edited:
Too bad the woofer has been damaged :(

Too bad in truth. Apart from that particular flaw that allowed the former to slam into the back plate even though the it had an extended pole piece, that driver had some interesting parameters, and the bandpass speaker was designed with those parameters in mind. The only replacements I've found so far are some Dayton drivers with Xmax around 5~6mm (the original driver had an 8mm Xmax).


You could add a limiter to the amplifier.

Not possible in this case. The amplifier in question is an old Technics SA-EX700 receiver, which has a built-in powered subwoofer channel. If any limiting is to be done, it has to be done at speaker level.


Is the amplifier still working? The fact that only half of the coil is blackened could mean that it was subject to dc, which could have been caused by a blown amplifier.


Yup, it's still working, though to be honest I haven't checked it for DC offset. Being a receiver, it probably does a self-check for this before engaging the internal circuit breakers for the speaker outputs anyway.
 
I'm surprised the receiver is still working. Hi-fi equipment usually does not have sufficient cooling for PA applications.

This thing is a bit of a beast. 100Wpc into 5 channels continuous (including a subwoofer channel) and has a built-in fan, so cooling isn't an issue. Class H+ output circuitry, and it still runs pretty warm.

It was my first HT receiver, so it's likely going on almost 15~20 years old.
 
Hmm, this looks a bit interesting. Converting the Enigma 4th order BP to a 6th order series-tuned BP, using the Eminence Kappa Pro-10LF. It looks too good to be true, in fact. Not only is the resulting bandpass response wider, it's also at a higher SPL level for the same input voltage.

The Kappa Pro-10LF is a little more expensive than what was prepared to pay, and I had to derive the semi-inductance parameters using a trace of its impedance curve as the ZMA data was not available, but the predicted response of 38 Hz to over 160 Hz looks very promising.

At least that's what Hornresp is predicting. Can anyone cross-check this for me using another software program that can sim series-tuned 6th order BP systems?

Driver parameters are as follows:
Eminence Kappa Pro-10LF
Sd=376.90
Bl=14.60
Cms=3.61E-04
Rms=1.26
Mmd=41.92
Le=0.97
Re=5.47
Pmax=600
Xmax=7.2
Re'=5.17
Leb=0.70
Le=5.67
Ke=0.13
Rss=50.08

6th order (series-tuned) parameters as follows:

V1= 21.9 litres
Vent Area = 91.2 cm^2 (two 3" vents)
Vent Length = 10 cm
V2 = 21.8 litres
Vent Area = 40.6 cm^2 (two 2" vents)
Vent Length = 16 cm long

The image is a comparison of the sim'd response of "Enigma2" against the sim'd response of the original Enigma subwoofer. The second is an export of the Hornresp file, just in case anyone's interested in having a look at the sim.

Of course there are some compromises to the design - particle velocity through the vents, for example. And GD of 22ms just under 40 Hz. But still, it does look very promising...
 

Attachments

  • 20190424-enigma2.png
    20190424-enigma2.png
    57.4 KB · Views: 782
  • enigma2.txt
    1 KB · Views: 85
Is it possible to pick larger ports? At 100 watts, the internal port reaches 40 m/s at 40 Hz. At these velocities the ports stop working.

Unfortunately larger vents just wont fit in the box. The 2" vents will flare to around 3.5" at the ends (if I do this, I'll be using Precision Ports), and of course they'll be terminated in the box, so some of the resulting distortion will be filtered by the bandpass filtering of the second chamber and vents, but still, yes, particle velocity is one of the concerns I have about this idea.

The bigger concern I have is if the sim'd FR is correct, because it does look a bit optimistic.
 
This driver has been damaged by high output signals with frequencies below the subwoofer's LF capability. So the driver was pushed too hard without any acoustic load on it and the VC hit the back plate.
You may want to put a 2nd or 3rd oder HP filter somewhere in the subwoofer's signal path that rolls off at 30 Hz.


Best regards!
 
I was hoping that you could tell me :).

Oh dear, LOL. I might just have to build one to confirm. I'm not sure yet whether I'm willing to use this particular build for confirmation.

FWIW, I did a "sanity check" by comparing the same build modeled two different ways - one using Hornresp's OD feature, and the other using Hornresp's BP6S (with Ap2 set to zero) feature. The results are close, but not quite the same (see figure 1 below). Of the two, the BP6S model came closest to predicting the measured performance of the Enigma (figure 2), so that's the one I'll be using going forward. It would be nice though if the BP6S model could include the impact of stuffing :).

Neither of the models get the impact of the offset perfect though. The measured offset is 21.5 cm, but I had to adjust it to 18.5 in the models to get a better match for the response around 400 Hz. I'm guessing this likely because the Hornresp model assumes that the vents are going to be mounted outside of the box's internal volume, when in fact they are not.



I'll have a closer look at that thread, thanks.
 

Attachments

  • 20190425-enigma1.png
    20190425-enigma1.png
    84.6 KB · Views: 363
  • 20190425-enigma2.png
    20190425-enigma2.png
    88.3 KB · Views: 368
This driver has been damaged by high output signals with frequencies below the subwoofer's LF capability. So the driver was pushed too hard without any acoustic load on it and the VC hit the back plate.
You may want to put a 2nd or 3rd oder HP filter somewhere in the subwoofer's signal path that rolls off at 30 Hz.

The former hit the back plate yes, but I think that more to do with poor design rather than overloading. The Enigma is currently a 4th order BP design, and the sealed rear chamber offers some protection against over-excursion at low frequencies.

I'm guessing that they used a linear spider on this driver, and it did a poor job of decelerating the former after the coil starts exiting the gap.

The distance from the top of the gap to the bottom plate is 33 mm, the coil length is 23 mm and the rated Xmax is 8 mm, which puts the gap at 7mm (23-2*8). As the coil is wound all the way to be bottom of the former, that puts about 18 mm (33-(7+8)) between the bottom of the former and the bottom plate. The distance from the spider to the top of the gap is the same 18 mm or slightly more. This means that while there's a lot of space between the former. the former's going to bottom out before the triple joint hits the top plate.

A good progressive spider and/or narrower surround would have started getting stiffer before the coil started leaving the gap, and certainly would have not let the former travel another 10 mm (18-8) past Xmax so easily. I could push the cone of the driver to Xmech with one hand - something I can't usually do with other bass drivers.
 
FWIW, I did a "sanity check" by comparing the same build modeled two different ways - one using Hornresp's OD feature, and the other using Hornresp's BP6S (with Ap2 set to zero) feature. The results are close, but not quite the same.

Hi Brian,

When the BP6S model is used, Hornresp knows that Ap1 and Lp1 specify a port tube, so an internal unflanged end correction is automatically added. When the OD model is used, Hornresp has no way of knowing that a horn segment is acting as a port tube, and therefore no throat-end end correction is added.

Attachment 1 shows the Enigma2 design you originally posted, but with Ap2 set to zero. Attachment 2 shows the equivalent OD design, but with an internal end correction length of 3.30 cm added to the 10 cm physical port length, so that L34 becomes 13.30 cm.

Internal unflanged end correction = 0.1952 * Pi * Sqrt(Ap1 / Pi) = 3.30 cm

Attachment 3 shows the BP6S response (actually changed to a BP4 alignment by setting Ap2 = 0).
Attachment 4 shows the equivalent OD response, with the end correction added.

The two responses are effectively identical up to 300 Hz.

Kind regards,

David
 

Attachments

  • Attach_4.png
    Attach_4.png
    57.6 KB · Views: 88
  • Attach_3.png
    Attach_3.png
    56.7 KB · Views: 234
  • Attach_2.png
    Attach_2.png
    54.3 KB · Views: 229
  • Attach_1.png
    Attach_1.png
    53.8 KB · Views: 250

ICG

Disabled Account
Joined 2007
The former hit the back plate yes, but I think that more to do with poor design rather than overloading. The Enigma is currently a 4th order BP design, and the sealed rear chamber offers some protection against over-excursion at low frequencies.

I'm guessing that they used a linear spider on this driver, and it did a poor job of decelerating the former after the coil starts exiting the gap.

[...]

A good progressive spider and/or narrower surround would have started getting stiffer before the coil started leaving the gap, and certainly would have not let the former travel another 10 mm (18-8) past Xmax so easily. I could push the cone of the driver to Xmech with one hand - something I can't usually do with other bass drivers.

Even with a progressive suspension that can happen, the sealed back chamber is not enough to absolutely prevent this from happening. You just need a 'enough low' frequency in the signal chain. If something that low or a on/off impulse, dc in the signal and/or a quick yank at the EQ is 'enough' to damage a driver. Kay Pirinhas first part of the diagnosis was correct and even a rugged PA driver would have been damaged. Maybe not at the first time but that damages the suspension too, so it would have happened some other time. In such situations a low cut is mandatory. Period.

Looking at the VC, you can clearly see the black burn at the coil end, that driver has been heavily abused. If not the excursion, the power would have done it later. If that wasn't enough, put not one sub with the same (or other) driver up, use several. Without a lowcut and control over the eq the same can still happen though.
 
It was a 100W driver, driven by an amp also rated to do 100W. I pulled the driver out within 24 hours of the bar reporting a problem with it. There was no characteristic smell of burnt voice coil so, if it saw too much power at some point, it could have been some time ago (the driver is about 19 years old). OTOH, being in a bandpass design with an Fb around 58 Hz and the motor located in the sealed section might have led to the coil getting browned over due to heat.

I first used that driver a vented design, where I first noticed its tendency to bottom out with just moderate signal levels). It was the main reason that I shifted to using it in a 4th order BP design instead that had a small sealed section, to exercise a bit more control over its excursion. Looks that it still didn't exercise enough control.

Finally, I've attached a sim of the bandpass design with the INFO10 driven at 100W, the rated output of the amp. Excursion exceeds rated Xmax below 42 Hz, but does not get even close to the calculated Xmech of 18mm. So how was it able to reach Xmech hard enough to break the former? That's what leads me to believe that there was some issue with the driver's design that lead to its failure, and I suspect that's "oil-canning" due to a linear spider.

The driver I'm currently considering replacing it with has a lower rated Xmax (6mm), but also has an aluminum former and cone, which should help to wick heat out of the sealed section, and hopefully it has a better suspension (some feedback I've received suggests that it does).
 

Attachments

  • 2019-04-27 (2).png
    2019-04-27 (2).png
    17.4 KB · Views: 102
I'm trying to come up with a simple power protection circuit for the new driver that will engage if it's overdriven for two long. Interestingly enough there doesn't seem to be much information available about this type of requirement for woofers (there is a bit for tweeters, but somewhat devoid of actual details).

Here's what I've come up with so far. R1 and R3 are polyswitch devices, chosen to keep the power dissipated by the driver to 100W or less. If R1 triggers, this will result in a 3 ohm 25W resistor being put in series with the speaker, which should reduce the power dissipation without significantly affecting the speaker's output (red: original, blue: with 3ohms in series). The third graph shows the power dissipation by the driver when the the protection is triggered. The purpose of R3 here is to reduce the output of the speaker significantly if someone tries to turn up the volume even more.

The driver that I'm considering for use is the Dayton DA270, which is rated for 80W RMS, 160W peak, so I think designing around 100W should work out Ok.

Now I just have to right-size those polyswitch devices. I might just end up buying a bunch of different values and testing them out... :)
 

Attachments

  • 2019-04-27 (6).png
    2019-04-27 (6).png
    15.4 KB · Views: 114
  • 2019-04-27 (7).png
    2019-04-27 (7).png
    23.1 KB · Views: 111
  • 2019-04-27 (8).png
    2019-04-27 (8).png
    20.5 KB · Views: 120
Note the temperature rise of the voice coil. Its dc resistance can double at full power: Speaker Failure Analysis

Good point - I had not taken that into consideration. Perhaps that would help to explain why the "100W speaker circuit breaker" I was looking at would trigger if it saw 3A for more than 5 seconds. 3^2*8 is only 72W. However, 3^2*16 (assuming coil resistance doubles) is 144W.


Thermal damage should not be a problem with a PA woofer and a 100 W amplifier.

I wouldn't call this a PA woofer :).