SL3000 Failures and Repairs

cds

In the past few months I’ve had 4 SL3000’s die. The failures were characterized either by no output or some serious distortion. After repairing them and checking the ones that were “OK”, I started wondering if there is a materials aging issue with the drivers or some other common cause for failure. What follows are my observations and thoughts on the reasons for the failures. Hopefully this may help others who have had the drivers mysteriously quit.

Before I get started, let me state that I do play the SDA’s loud at times using a Carver A-760X. Hard not to do. I have done this for years and have never noticed the tweeter protection device activating so I did not have a concern that they were approaching electrical meltdown. My only “problem” is that during large bass events I have to watch the volume to prevent the 6.5” drivers from bottoming…. not a good thing.

Initially I suspected some of the tweeter materials were showing signs of age or crossover caps had become leaky allowing too much LF energy to reach the tweeters. Now after evaluating the failures I also believe sound pressure from inside the cabinet is another likely scenario.

During repairs I found either the voice coil open or the voice coil detached from the dome. Initially one would think the open coil would be an electrical failure and the other to be mechanical. However, I now believe both are mechanical in origin.

The detachment of the coil from the dome appears to be an adhesive failure. The open coil failures in all 3 cases looked like mechanical work hardening of the wire rather than electrical burnout. My conclusion is based on examination of the break using 60x magnification.

After repairing them, I decided to perform a basic test to evaluate the repairs. Applying an input from 1 – 20KHz in 1KHz steps, I measured the SPL about 6” from the dome. Besides measuring the SPL at each point, I also fed the SPL meter into an oscilloscope to monitor for distortion. Crude yes, but significantly better than by ear alone.

I compared the performance of the repaired units and found them to be within 2 dB of each other over the full frequency range. When I checked the ones I thought were OK, I was surprised to see the output 6-8 dB lower than the repaired units with noticeable distortion at various frequencies. Taking these apart I found no problems like the others, so I did a little maintenance (cleanup and adjustment) and voila, they now performed on par with the repaired units.

If my suspicion is correct both types of failures are due to excessive dome movement. In one case the wire fractured and in the other, the glue separated from the dome. My thought is that excessive low frequency energy is reaching the domes. Either 1) there are leaky caps in the crossover allowing excessive low frequencies to reach the tweeters but not enough current to trip the protection device, or 2) air pressure inside the cabinet finding its way through the rear of the tweeter and moving the dome.

There is a gasket in the tweeter under the dome assembly that should prevent this. It looks like time has started to take its toll on these gaskets and they have taken compression set. While still usable, the mounting plate screws have to be torqued appropriately to make the air seal. If I keep having this problem I may eventually make a new gasket since I doubt this part is available. It looks to be made out of the same material used for the gasket to mount the drivers to the cabinet. In performing the repairs, I noticed that some of the screws were not quite snug. Perhaps they were tight at time zero, but 15 years later? Who knows?

Note: If you elect to check the screws on your drivers, alternate the tightening sequence to keep the voice coil centered and don’t tighten them excessively.

I’m planning on replacing the crossover caps when I get a chance just to be on the safe side. I have also been thinking of building a sealed enclosure behind the tweeters to isolate them from the internal pressure of the cabinet, thereby making the gasket less critical.

Any thoughts?

One last thing. I’m assuming my repairs have restored the 3000’s but I have nothing to reference them to. I have ordered one of the new 3000 equivalents from Polk just to see how it compares. I’ll post an update after I’ve had a chance to test is against the repaired versions.

I am also interested in comparing original 3000’s to the repaired versions. If anyone has spares sitting around and would like to loan them, let me know.

Cheers

PolkWannabie

Spares ? ... Sitting around ???

disneyjoe7

Have you tested the Carver amp out with the scope, using the test signal.

Thinking that 99% of tweeter death are clipping related.

If not the amp yes a bad cap in the crossover? Can you check this with a spare speaker (full range) if any LF is present you would hear it.

cds

Yes I’ve checked the amp. With a 4 ohm load it was clean to over 600 watts so I don’t think it’s the problem.
I haven’t checked the crossover as you suggested but I agree it would indicate LF at the tweeters. I was planning to run a frequency sweep and plot the crossover response before and after changing the caps.

One thing that would be interesting to know is how much LF the 3000s can accept. ie. What should they be crossed over at and what slope is desirable?

Anyone ever see specs?

hoosier21

Some great info, thanks for taking the time to inspect/repair/test and give a crap about those tweeters. Could same folks some money.

Sealed tweeter section would be an easy "upgrade" to do for most anyone.

disneyjoe7

Given the input you have stated in your post, I did assume you have checked the amp. But I just had to ask not to overlook the obvious.

Originally posted by cds
Yes I’ve checked the amp. With a 4 ohm load it was clean to over 600 watts so I don’t think it’s the problem.
I haven’t checked the crossover as you suggested but I agree it would indicate LF at the tweeters. I was planning to run a frequency sweep and plot the crossover response before and after changing the caps.

One thing that would be interesting to know is how much LF the 3000s can accept. ie. What should they be crossed over at and what slope is desirable?

Anyone ever see specs?

Tour2ma

A leaky cap ouught to be apparent from a simple VOM check... no?

cds,
Can you post a little more on the actual repair and "adjustment" of the tweets? Maybe some pics?

F1nut

Yeah, what Tour said.

disneyjoe7

Tour,

Just to say this yes a meter could check for a leaky cap, my method was more say back woodzee and would give me a warm fuzzy

madmax

I found your post very informative. I have added a few more ideas to add to my check list for someday when I tweak my SDA's. Thanks!
madmax

cds

A leaky cap ouught to be apparent from a simple VOM check... no?

Can you post a little more on the actual repair and "adjustment" of the tweets? Maybe some pics?

Since all caps have some leakage, just how much is too much? And how does a little leakage effect the low frequency cutoff of the filter?
I don’t believe the answer is trivial, but I probably could come up with a test procedure that would quantify leakage vs. low frequency response….basically as the cap equivalent impedance changes, how does the overall circuit response change?

I have an LCR meter that can measure very low leakage, but there is the other variable, namely how does dielectric absorption change after a voltage is applied? A real leaky cap would probably be easy to characterize, but good caps are often leaky after not being used for a while and after the dielectric reforms with applied voltage, they return to like-new levels.

Somehow, I think replacing the caps with new ones would be a hell of a lot easier than analyzing and characterizing a couple hundred caps. What would be the bottom line anyway? Replace the caps!
Also, disneyjoe7’s method of using a LF speaker would probably be as good as all this mumbo jumbo.

In my case before I had taken the drivers apart my first thoughts were that the crossover caps might have been leaky due to several years of inactivity. I’ve never seen a study relating shelf life to leakage, but it’s common practice when repairing old electronic stuff to apply voltage gradually because old capacitors, notably electrolytic, have a nasty habit of shorting when rated voltage is applied suddenly after a long period of non use. When testing caps on the LCR meter even good caps show high initial leakage and gradually drop within specified limits if the voltage is maintained.

I blamed my own exuberance of wanting to hear my SDAs again after so long that I “Turned the key and floored it”….. no one to blame but myself for spinning the bearing.


OK, enough of this. Let’s get back to the second question that may be more useful for others in the group. Specifically, how did I “tweak” the working drivers. The answer is that I’m not sure if what I did had anything to do with the improvement. Statistically, the sample size was small so it could have been simply the fact that I opened them, but you be the judge.

I’ll outline what I did with the functional units and add that based on this experience, if I ever get others for spares or replacements, I’ll likely perform the same steps. That’s just me however, your mileage may vary.


After testing the “good” units using the method described earlier, I was surprised that their output was kind of crappy. Makes me wonder what I had been listening to. After disassembling them I found the following “issues” which I also noted when I had the dead units apart..

1) Ferrofluid Distribution
The ferrofluid seemed to be located in one section of the magnet gap. Does this mean the stuff responds to gravity and migrates to the location of the gap at the 6 O’clock position? I don’t know, but the magnetic field is so strong in the gap that I have a hard time believing gravity would be a factor. Unfortunately I had all the drivers out of the cabinet so I could not correlate the driver mounting orientation to the puddled fluid. All of my drivers exhibited this puddling, and to me, it just didn’t “look right”.
If anyone has a definitive theory on this, please share it.

2) Ferrofluid Chunks
Because I was working under a microscope, I could easily examine the fluid. Most of it had a light mineral oil consistency but there were also lumps of stuff which looked like the charcoal crust on a good steak, floating in the fluid, along the edges of the gap, and on the voice coil. Normal? I have no clue, but again, it didn’t “look right”.

Using a Q-tip and alcohol, I cleaned the inside and outside of the voice coil. I next took a small sliver of paper about the thickness of a computer punch card (remember them?) and moved it horizontally and vertically through the fluid. When I took it out of the fluid, if there were lumps attached, I cleaned them off and repeated the procedure. When the lumps in the fluid were few, I moved the paper sliver around the “dry” portions of the gap. Again, as it collected chunks of stuff, I cleaned them off and repeated the process until there was not much left.

I then used the sliver of punch card as a spatula. Moving it 360 deg. around the gap I attempted to more evenly distribute the fluid. It sorta of worked, but was not easy. The fluid did not like leaving the portion of the gap where most of it had puddled. Makes me more suspicious of how gravity would do this. I also placed the coil in the gap and rotated it a few times to further “distribute” the fluid around the gap.
Currently I’m out of ferrofluid, and my source is out of stock, but when I get more, I may experiment by chemically flushing the gap and evenly adding new fluid.

That’s about all the internal stuff. During assembly of the mounting plate I apply a signal to the coil while positioning it in the gap and tightening the screws. By monitoring the output waveform, I can adjust the coil and screw torque for minimum distortion. I especially pay attention to the frequencies below 2KHz. I have seen the output waveform appear fine at 8KHz but look like crap below 2KHz. If I adjust for a clean 1KHz waveform, the upper frequency range looks great.

That’s it. Simple enough?

BTW, What type of pictures would be of interest?
If I can figure out how to get something useful through the microscope the next time I have things apart I’ll attach a couple.

Tour2ma

The parallel with bringing old amps up on a Variac never occurred to me. Nice bit of insight.

Agree that ID'ing bad caps may not be as simple as I initially made it sound, but out of the circuit, a significantly failed one would show itself. But that's after they're already out...

As for pics, I'd say certainly any critical points in the dis-/ re-assembly and other points of interest along the way.

BTW there's another thread here on the 3000's rebuild and in particular the choice of ferrofluid replacement.
http://clubpolk.polkaudio.com/forum/showthread.php?s=&threadid=13849&highlight=ferro+fluid
Ignore some of the trivial stuff. There's some info about halfway down the page...

dorokusai

Neither post has displayed anything but words about this situation.