Should we be careful when measuring old speakers? A not-too-faulty (but still faulty) speaker measured

[wwenze]

After learning passive speakers can spoil too, I measured my pair of Diamond 8.2 that have sat in the living room for years

Close mic of tweeter. The bad speaker was measured twice (orange and brown below)

Hoo boy.

My concern is that if we didn't have a second unit to compare against, we wouldn't really think the broken speaker is broken. We would suspect it is a bad speaker and blame the designer.

 

[Juhazi]

It is not broken, just other driver wired in wrong polarity! Please check impulse and step response in REW to see the difference.

More info here, it looks like 2nd order http://techtalk.parts-express.com/forum/tech-talk-forum/12868-2nd-order-reverse-polarity-question

 

[wwenze]

Bad unit step response

Good unit step response

Bad unit impulse response

Good unit impulse response

 

[sergeauckland]

I've had a similar experience when I bought my 801s. I couldn't hear them at the seller's house, as he'd sold his HiFi as he was emigrating, and the 801 were all that was left, so very trustingly, he let me take them away, and if they were OK he trusted me to pay for them. Anyway, I took them home and tried them, and all three drivers were working, but I didn't pay much attention to what they sounded like as my intention anyway was to make them active, so as long as the drivers were all working, any crossover faults would go away. Also, I was keen not to abuse the seller's trust and make the payment as soon as I was satisfied they were working.

When I had more time, I measured the 'speakers as they were, and found that both tweeters were working, but well down on output, one as much as 10dB, the other something like 4 or 5. According to B&W, it could be the ferrofluid drying out, (I also remember KEF telling me once about the same thing, Ferrofluid goes gummy after a long time, so although the tweeter still works, output is well down.) Sadly, my tweeters didn't have any ferrofuid as it wasn't used in my version, which was built just at the transition to ferrofluid, and B&W could supply me with new tweeters for a modest price, so I decided to replace them.

Crossover faults can also upset the measurements, especially very old unpolarised electrolytics which tend to go high in value and therefore change the crossover points. Again, all the drivers will provide output, but the sound (and the measurements) will be off. One problem with loudspeaker faults is that many take place over a long time, so one gets used to the slowly changing sound, and it's only on measurement or hearing something else that the faults become more obvious.

I think that judging vintage loudspeakers, especially if measuring only one, can be misleading. At least measuring both of a pair would show up any significant differences, indicating that one (or both) may be at fault.

S.

 

[Juhazi]

Yes, the "broken" unit has both drivers connected in same polarity. Open it (bass driver is often easier to take off) and switch wires. Then check that both have impulse peak in same direction. If they are not, you can change amplifier wire polarity or reopen the "problem speaker" and change tweeter wire polarity and reverse the change in woofer. And measure again...

Most likely the broken speaker has had a driver failure and the repairman has made a mistake when assembling. Sometimes this happens at the factory too with new workers! I have seen a speaker where wires coming from crossover had "wrong colors" for the tweeter, obviously to make life of assembly workers easier!

 

[Juhazi]

Here my measurements of old speakers that were a mess, both of them

! Other of double woofres were broken, and sometihing weird with double mids as well. I ended with going dsp and single mid+tweeter!

 

[wwenze]

The step response shows the polarity configuration in both speakers are the same... no?

 

[andreasmaaan]

The step response shows the polarity configuration in both speakers are the same... no?

Yes actually, you're right!

Perhaps you could try swapping the tweeters. If the same tweeter measures differently connected to the other crossover, then you'll know the issue is in the crossover. If the tweeters measure the same regardless which crossover they're connected to, the problem is with the tweeter.

 

[Juhazi]

Yes actually, you're right!

Perhaps you could try swapping the tweeters. If the same tweeter measures differently connected to the other crossover, then you'll know the issue is in the crossover. If the tweeters measure the same regardless which crossover they're connected to, the problem is with the tweeter.

I noticed this too, but impulse responses are different and steps are not exactly same. Perhaps measurement axis/height was not identical? Look at how differently speakers sum woofer and tweeter response around 10kHz, the bad speakers has higher spl...

 

[wwenze]

The plastic grille has become stiff with age, and feels like it will break apart with force like the xbox 360 enclosure.
But the drivers are screwed in from the front, and the wire is soldered onto the crossover PCB...

Crossover of both measure identical in-circuit. Replacement tweeter stocks are... not optimistic.

Probably will end up in the storeroom while I get the excuse to take the S3000Pro out. Is ignorance bliss or not in this case?

 

[ctrl]

Close mic of tweeter. The bad speaker was measured twice (orange and brown below)

How is it that the two measurements orange and brown are so different? Was the measuring microphone not mounted on a microphone stand?

When measuring, you should always indicate the measuring distance, so that others can estimate how big other influences are.

For such a near field measurement the measuring distance to the tweeter should be <1cm (0.5'') to minimize the influence of the other drivers.

If the measuring distance was already below 1cm I would guess that the blue curve shows a crossover-defect.

 

[Kvalsvoll]

Bad unit step response

Good unit step response

Bad unit impulse response

Good unit impulse response

This clearly shows that the hf unit on the bad speaker is connected with wrong polarity. Swap the wires on the tweeter, and this speaker may work as intended. The situation was described in post #5, with a likely scenario for how this can be possible.

What amazes me here, is that readers on a technical-oriented audio forum do not see this clearly from those measurements. Like - you claim you can judge a loudspeakers performance by looking at measurements, but fail to see the simple and obvious.

 

[ctrl]

This clearly shows that the hf unit on the bad speaker is connected with wrong polarity.

It may be true, but without more detailed information about the measurement conditions this need not apply.

Assuming the crossover frequency is 2500Hz, the wavelength is 0.14m. If the phase alignment during the crossover is optimal, we get total extinction when the wavelength shifts by 0.07m.

This means that if the sound travel time difference between tweeter and woofer changes by 0.07m in the near field measurement, we still get perfect cancellation in the near field measurement with a correctly connected crossover.

 

[Kvalsvoll]

It may be true, but without more detailed information about the measurement conditions this need not apply.

Assuming the crossover frequency is 2500Hz, the wavelength is 0.14m. If the phase alignment during the crossover is optimal, we get total extinction when the wavelength shifts by 0.07m.

This means that if the sound travel time difference between tweeter and woofer changes by 0.07m in the near field measurement, we still get perfect cancellation in the near field measurement with a correctly connected crossover.

No assumptions or wavelengths necessary, it is obvious that the two speakers are different and that this difference is caused by the tweeter having inverted polarity. This you can see in the IR graph. Then you can use the frequency response as additional information to decide which one is faulty, in this case it is likely the one with the reduced level has tweeter with wrong polarity.

 

[ctrl]

No assumptions or wavelengths necessary, it is obvious that the two speakers are different and that this difference is caused by the tweeter having inverted polarity. This you can see in the IR graph. Then you can use the frequency response as additional information to decide which one is faulty, in this case it is likely the one with the reduced level has tweeter with wrong polarity.

Since details of the crossover are missing, it can only be speculated. Nevertheless IR and SR of the "faulty" speaker do not look like the polarity of the tweeter is inverted.

This would look something like this (measured at the reference point of the speaker):
In phase:

inverted phase of the tweeter:

Doesn't it look more like a crossover or tweeter/chassis defect?

 

[andreasmaaan]

No assumptions or wavelengths necessary, it is obvious that the two speakers are different and that this difference is caused by the tweeter having inverted polarity. This you can see in the IR graph. Then you can use the frequency response as additional information to decide which one is faulty, in this case it is likely the one with the reduced level has tweeter with wrong polarity.

The reason I'm less certain than you is that the IR and the step response of the "bad" speaker don't seem to reconcile. The step suggests that the tweeter is connected in positive polarity, while the IR suggests the opposite.

I'm also a little baffled by the "near-field" tweeter measurement. If a polarity reversal is responsible for the suck-out (which I also think is the most likely explanation BTW), I wouldn't expect to see that so clearly in the near field of the tweeter, where the tweeter's output should dominate.

 

[Kvalsvoll]

Since details of the crossover are missing, it can only be speculated. Nevertheless IR and SR of the "faulty" speaker do not look like the polarity of the tweeter is inverted.

This would look something like this (measured at the reference point of the speaker):
In phase:
View attachment 93736


inverted phase of the tweeter:
View attachment 93737

Doesn't it look more like a crossover or tweeter/chassis defect?

Looking at those curves once more, you may be correct. A fault in the tweeter og crossover that causes changes in frequency reponse could also result in the IR changing shape, so it looks like it caould be inverted. And the lading edge is in-phase for both.

 

[Kvalsvoll]

The reason I'm less certain than you is that the IR and the step response of the "bad" speaker don't seem to reconcile. The step suggests that the tweeter is connected in positive polarity, while the IR suggests the opposite.

I'm also a little baffled by the "near-field" tweeter measurement. If a polarity reversal is responsible for the suck-out (which I also think is the most likely explanation BTW), I wouldn't expect to see that so clearly in the near field of the tweeter, where the tweeter's output should dominate.

There is significant leakage from the bass/mid driver, tweeter does not play down to 100hz alone, but response is also different av very high frequencies. So there may be a defect in the tweeter or crossover. Swapping out tweeters or just try to change polarity gives the answer to that.

 

[andreasmaaan]

There is significant leakage from the bass/mid driver, tweeter does not play down to 100hz alone, but response is also different av very high frequencies. So there may be a defect in the tweeter or crossover. Swapping out tweeters or just try to change polarity gives the answer to that.

Which is precisely what I suggested the OP should try to do.

 

[Kvalsvoll]

Which is precisely what I suggested the OP should try to do.

I should have added "like @andreasmaaan suggested".