ELAC DF63 Floor standing Speaker Review

[daniboun]

From 7 review found this
We see a serious hump between about 100hz and 1,000 Hertz.

 

[Power Pop 23]

This sample was donated to ASR. As the owner, @amirm could initiate a warranty claim with ELAC (USA).

 

[Salt]

One problem with using only one speaker, you cant tell if it sounds different than the other one.

Thats very true, and that's is why at audioxpress' voice coil magazin's Vance Dikeason always test two specimen to determine the variance of such.

 

[amirm]

One problem with using only one speaker, you cant tell if it sounds different than the other one.

Seeing how i could find this flaw with a single speaker shows the value of listening to a single speaker!

 

[daniboun]

Seeing how i could find this flaw with a single speaker shows the value of listening to a single speaker!

Therefore, all that remains is to test the second one) we will probably be fixed
Let s observe the Ockham's razor : « pluralitas non est ponenda sine necessitate » )))

 

[Salt]

An attempt of analysis:

Certainly something is going wrong that should not do so, but where's the fault?
Elac, being in the market for over a century, should know what they do, not?

First point is the (estimated) single driver's response, that shows well attenuated bass at xover frequency, that then levels up to -25dB at 1-2kHz, where the mid schould play alone.
Same figure at mid, that is attenuated well between 2.5 and 5 kHz, and then levels between 6 and 9 kHz at -25 dB.
Both levels of residuals may be audible at 96 (measured at ~100) dB listening level.
But two symmetrical errors in two different branches of xover? (That would mean a bypass for woofer as well as mid fo rising frequency was implemented for both ways, and why should one do this?).
Could this explane the percepted irregularity?, or is this is probably more a fault of measuring setup? (drivers are not measured singular, not?)

Second point is 86 dB vs 96 dB THD shows lots of distorsions in lower frequencies, but not above 1 kHz.
That would refer to resonances up to 1 kHz, but not at 5 kHz (that's seen in the upper mentioned figure)?

Third point is waterfall does not present any disturbance above 1 kHz.

Fourth point is, of course, the difference to Erin's results.
The presentation of measurement results differ, but if we have a look at the step response, it is very smooth at Erin's measurement, but has some, ehem, 'steps', in Amir's measurements.

 

[amirm]

From 7 review found this
We see a serious hump between about 100hz and 1,000 Hertz.

Good find. Assuming it is some kind of in-room, gated measurement, it more closely matches my estimated in-room measurement:

Than Erin's:

 

[daniboun]

An attempt of analysis:

Certainly something is going wrong that should not do so, but where's the fault?
Elac, being in the market for over a century, should know what they do, not?

First point is the (estimated) single driver's response, that shows well attenuated bass at xover frequency, that then levels up to -25dB at 1-2kHz, where the mid schould play alone.
Same figure at mid, that is attenuated well between 2.5 and 5 kHz, and then levels between 6 and 9 kHz at -25 dB.
Both levels of residuals may be audible at 96 (measured at ~100) dB listening level.
But two symmetrical errors in two different branches of xover? (That would mean a bypass for woofer as well as mid fo rising frequency was implemented for both ways, and why should one do this?).
Could this explane the percepted irregularity?, or is this is probably more a fault of measuring setup? (drivers are not measured singular, not?)

Second point is 86 dB vs 96 dB THD shows lots of distorsions in lower frequencies, but not above 1 kHz.
That would refer to resonances up to 1 kHz, but not at 5 kHz (that's seen in the upper mentioned figure)?

Third point is waterfall does not present any disturbance above 1 kHz.

Fourth point is, of course, the difference to Erin's results.
The presentation of measurement results differ, but if we have a look at the step response, it is very smooth at Erin's measurement, but has some, ehem, 'steps', in Amir's measurements.

This is why I mentioned Ockham's razor ))) principle of rationality which aims to opt for the least expensive solution in ad hoc hypothesis: namely testing the second speaker

 

[amirm]

Third point is waterfall does not present any disturbance above 1 kHz.

I can make that graph show any number of resonances you want just by changing one parameter I set by hand. Here is another version:

All I did was lower the noise floor. This is why I repeatedly caution people from relying on this type of graph.

 

[amirm]

Could this explane the percepted irregularity?, or is this is probably more a fault of measuring setup? (drivers are not measured singular, not?)

A lot of caveats go into those measurement as all of them are playing at once. Also, I only showed one port and one woofer. The two different slightly from each other (due to differing location in the speaker). I thought the graph would be impossible to read if I put those in there as well.

 

[ttimer]

actually rather complimentary from Erin's side and more contrasting from Amir's side.... the question is what explains this @amirm ?

Bad QC and Erin got lucky?

 

[amirm]

This is why I mentioned Ockham's razor ))) principle of rationality which aims to opt for the least expensive solution in ad hoc hypothesis: namely testing the second speaker

There are also other simple explanations. Erin's sample came from the company itself (mine was bought from Amazon I think from ELAC USA direct). It could have been an earlier production sample, built locally instead of China, used different parts, etc.

Bottom line here is that we have a speaker that is in perfect condition without a single visible damage to it. Let's say I test another one and it is different. What then? How do you know which one you get if you purchase it?

 

[Salt]

I can make that graph show any number of resonances you want just by changing one parameter I set by hand. Here is another version:

View attachment 474642

All I did was lower the noise floor. This is why I repeatedly caution people from relying on this type of graph.

That would explain some irritations in lower mids, but not what You reported about distorsions in mid-highs.

Don't misunderstand: it's no criticism of Your high standard measurements, it's a query to understand how this could go (wrong) this way.

 

[Power Pop 23]

Bad QC and Erin got lucky?

Or good quality control and Amir got unlucky

 

[beagleman]

Kind of shocked. Erin regarded this speaker quite well in his review.

But more alarming, Erin's distortion measurements at 96db seem substantially lower meaning?? Especially in the 100-200 or so hz range...odd

 

[amirm]

Broken tweeter?

I should clarify that I am not 100% sure it is the tweeter. I went close (a few inches) to the tweeter/midrange but it was too loud for me to stay there too long. I thought it came from that combo area. It is possible that I am hearing high order harmonics of the midrange.

 

[amirm]

Here is a question. If there is serious damage to the speaker, why is the on-axis response so good?

That doesn't indicate any damaged parts. Most likely their QC is around frequency response and rub/buzz which this speaker would pass.

The issue around 1000 Hz is predictable due to so many sources interfering: two woofers, two ports and a midrange.

 

[Salt]

I should clarify that I am not 100% sure it is the tweeter. I went close (a few inches) to the tweeter/midrange but it was too loud for me to stay there too long. I thought it came from that combo area. It is possible that I am hearing high order harmonics of the midrange.

That would confirm Your measurements, if the mid would go up to 5-8 kHz at -20 dB ... but this would not be intended by engineering.
Something is faulty.

 

[PristineSound]

Love to see more reviews like this!

What a missed opportunity for Elac.

 

[amirm]

Something is faulty.

Maybe. But you really needs this one clip to hear the problem in all of its glory. You could listen to ton of other tracks and not hear it.