ELAC DF63 Floor standing Speaker Review

[Thomas_A]

When it comes to drivers, there might be brands that have quite some varations. Cannot speak for Elac but I’ve heard it from other speaker manufacturers that they have been through some issues of driver quality. Among the cheap Peerless cone tweeters I use for DIY, I had to measure and select the best ones.

 

[tmtomh]

Exactly, yet half this forum thinks that's how it works. One data point and it's a world full of "I'll pass" or "disappointing, glad I didn't pull the trigger" type comments.
This thread is interesting because it's thrown up the classic case of "I don't measure the same result as you do so who is right?" conundrum. The subjectivists have as much chance of being right at this point.

I don't think that's a fair criticism of the forum membership. The rigorous, large-N random sampling you're talking about is the manufacturer's responsibility, not @amirm 's or any other reviewer's. If we can't be reasonably confident that a review sample will be representative of the performance of a unit we purchase, that's not on the reviewer or on the allegedly poor understanding of forum members. That's on the manufacturer - and that by itself is reason to avoid that manufacturer's products.

 

[Mnyb]

I don't think that's a fair criticism of the forum membership. The rigorous, large-N random sampling you're talking about is the manufacturer's responsibility, not @amirm 's or any other reviewer's. If we can't be reasonably confident that a review sample will be representative of the performance of a unit we purchase, that's not on the reviewer or on the allegedly poor understanding of forum members. That's on the manufacturer - and that by itself is reason to avoid that manufacturer's products.

Yes for reviewing purposes one if a sample behaves strange , get one more is sufficient usually one sample is obviously broken .
You will not catch the typical deviation in a whole product line . But one properly working sample is good enough, and the diff vs a broken one is obvious is it not ?

 

[Bernard23]

I don't think that's a fair criticism of the forum membership. The rigorous, large-N random sampling you're talking about is the manufacturer's responsibility, not @amirm 's or any other reviewer's. If we can't be reasonably confident that a review sample will be representative of the performance of a unit we purchase, that's not on the reviewer or on the allegedly poor understanding of forum members. That's on the manufacturer - and that by itself is reason to avoid that manufacturer's products.

I'm not sure I agree, I've seen multiple comments that are a clear assessment of a device based on one review. Of course we can't carry out statistical testing, I made that point earlier, but interpreting the data is what forum members do! I'm gullty of it, I bought a pair of Elac dbr62 mostly off the back of Amir's review.

 

[beagleman]

I'm not sure I agree, I've seen multiple comments that are a clear assessment of a device based on one review. Of course we can't carry out statistical testing, I made that point earlier, but interpreting the data is what forum members do! I'm gullty of it, I bought a pair of Elac dbr62 mostly off the back of Amir's review.

I think simply looking at the forums ratings of this speaker speaks volumes about how they tend to react.

Anyone even mildly aware would realize erin tested it "recently" and had differing results.

My shock came at, instead of pointing out that as a scientific point, many simply bashed this speaker, while being totally ignorant of the fact we have another set of data available, that is also quite credible.

All while have next to zero curiosity about WHY we had 2 sets of data that conflicted.

 

[Bernard23]

I think simply looking at the forums ratings of this speaker speaks volumes about how they tend to react.

Anyone even mildly aware would realize erin tested it "recently" and had differing results.

My shock came at, instead of pointing out that as a scientific point, many simply bashed this speaker, while being totally ignorant of the fact we have another set of data available, that is also quite credible.

All while have next to zero curiosity about WHY we had 2 sets of data that conflicted.

Exactly. So few people challenge the data period, or rather fail to recognise the uncertainty that is associated with it. Amir's testing or anyone's testing comes with some pretty big caveats.

 

[tmtomh]

I'm not sure I agree, I've seen multiple comments that are a clear assessment of a device based on one review. Of course we can't carry out statistical testing, I made that point earlier, but interpreting the data is what forum members do! I'm gullty of it, I bought a pair of Elac dbr62 mostly off the back of Amir's review.

My point is this: the ability to mass-produce hi-fi audio components to a standard of reasonable consistency and uniformity has existed for many decades now. If you go back 50 years, it was in most cases a fairly simple matter for any manufacturer with sufficient capital and basic competence to spew out 100s or 1000s of amps, receivers, turntables, cassette decks, preamps, and yes even speakers whose performance was very similar. They were all made from parts that were themselves mass-produced to fairly detailed specs, and they were all assembled in mass-production environments where major variation was rare and a product of a clear error rather than an uncontrollable aspect of the production process.

So if a speaker reviewed here or anywhere else is not a representative example, that is either a fluke or a result of a fundamentally flawed manufacturing process (or perhaps in rare cases of fraud - although personally I think the "golden sample" phenomenon is a lot rarer than some people seem to think). Either way, that's beyond the ability of a reviewer to consistently discover - and it's also beyond the scope of a review.

To your point, it might indeed be true that your Elac DBR62s produce different sound than the one @amirm tested. But even if it were true (which I doubt it is, BTW), that does not mean that you, or Amir, or anyone else made an error in "interpreting the data."

The only way to "correctly" interpret the data, by the logic of the argument you are putting forth here, would be to say, "this is only one sample and I can have no confidence based on N=1 that this is the performance I will get if I buy this."

But if you make that so-called "correct" interpretation, then your only course of action is to ignore the review's measurements, and in that case you have no basis upon which to decide whether or not to buy the DBR62 or any other speaker or piece of gear. That leads to the IMHO absurd conclusion that you arrived at in one of your prior comments, which was that we here are on no more solid ground than the subjectivists are. And more importantly, it dictates that you should not buy any piece of gear that you cannot audition in-person (and for speakers it means rejecting any speaker you can't audition in-person in your own listening space for an extended period of time).

Now, if your response to that is something along the lines of, "We can do better if we also look at other reviews to see if they get the same results as Amir did," of course that's reasonable. But in that case you are still dealing with a very small N (since a lot of times there is only one other review that has comparable measurements so your N only goes from 1 to 2) - and while such cross-checking is of course a good idea, that procedure would not address the sweeping claims you're making about the limits of our ability to know how the majority of all the DBR62s out there in the world actually perform. So in my opinion you're advocating for something very modest and reasonable in practice - but you're criticizing the entire membership here (including yourself) based on an argument that doesn't hold water.

Finally, I think you're also acting like we have no way of knowing whether or not a modern, mass-produced component has major, wild unit-to-unit swings in performance. While we can never know anything with 100% certainty, there is plentiful evidence out there that gives us a very high degree of confidence that unit-to-unit variation is quite small across the board in the industry. And as others have pointed out, in the rare cases when significant deviation in performance does exist, it is usually pretty easy to detect because something weird jumps out of the measurements, prompting further investigation. And now that I think about it further, in most cases when Amir has found a goofy measurement result, the problem was a design flaw or an internal component/wiring layout flaw, which existed across the board and uniformly. Manufacturers, when they have addressed these issues based on Amir's reviews, have done so by revising the layout or production process, not by tightening up their unit-to-unit QC, because unit-to-unit variation was not the problem in most cases.

 

[Bernard23]

I think you've partly misunderstood my point, my fault for not being so clear. You've made a few assumptions too, which is fine for audio gear but that's not the approach for industrial engineering. Manufacturing is essentially the process of managing variation; variation in the parts you're making and the system you're using to verify them. Maybe Elac did all of the testing and could verify their products with a high degree of confidence, or maybe they didn't. All we usually have as the consumer reading ASR forums is one single measurement data set and my point is that some (and no i didn't say the entire community!) take that that single data set, and assume that this is representative of every product that the company has made. Now, if you're making aircraft engines, that is the case, because that's the process of certification: you have to prove your product is compliant, but that's not the case in consumer audio, there are few regulations or standards that a product can be verified against.
The fact that Erin and Amir have got two very different data points using the same test method suggests either one of the tests was conducted incorrectly (an error in the test method) or that there is a significant part to part variation in Elac's product, but we don't have enough data to conclude anything more than that.
All I'm saying is that you can't be 100% certain of the performance of something that's subject to variation on a sample of 1, or even 2; all we can do is form a general opinion, and you're probably right, it's highly likely that most products confirm to perfectly acceptable limit of variation between products, but as you said yourself we can't be 100% sure, and that was my point.

Anyway, we are veering wildly off topic so let's move on.

 

[amirm]

All we usually have as the consumer reading ASR forums is one single measurement data set and my point is that some (and no i didn't say the entire community!) take that that single data set, and assume that this is representative of every product that the company has made.

No one is assuming that. Measurements are specific to the model being tested. Further, everyone knows that they could get a faulty product.

It is strange to bring up such concern in this context because we have no less than 3 measurements! One from me, another from Erin and a third from European publication. With respect to such things as frequency response, the correlation for something mechanical with different people measuring them is actually quite excellent. Take this overlay of PIR that a member put together:

This indicates very high level of correlation between these independently measured speakers.

From what I recall, the CEA-2034 considers 3 dB variation to be good enough. Whether you use my measurements or Erin's, story is the same here.

Yes, distortion measurements vary. And I hear clear artifacts in a song I played. I consider this faulty driver/manufacturing. The probability of this happening when you buy the product either bothers you, or doesn't.

What it doesn't do is entitle the membership to demand that we buy another $700 speaker and waste opportunity cost of $2,000 to measure another sample. If it is that important to you all, then raise the $2,700, donate it to the forum and I will buy and test another.

Until then, this is not a science project. By now, ELAC has read or should have read my review and they have done nothing to deal with this situation.

Net, net, my testing is highly representative of product performance. There is some degree of unknown which I cannot eliminate. You have to be comfortable with this as without my data, that unknown rises to near infinity seeing how the manufacturer is not provide any measurements.

 

[Bernard23]

It is strange to bring up such concern in this context because we have no less than 3 measurements!

Net, net, my testing is highly representative of product performance. There is some degree of unknown which I cannot eliminate. You have to be comfortable with this as without my data, that unknown rises to near infinity seeing how the manufacturer is not provide any measurements.

Ah I wasn't aware of the third tester!

Agreed, it's highly representative of the sample you have in your room!

And absolutely, caveat emptor and all that. By the way I'm not seeking to diminish the work you and others do as a free service, because without it we have independent verification. And yes, I've come at it from the perspective iof a science project, but ultimately measurement is just that.
Fwiw should I ever find myself lucky enough to be randomly selected to be given a ton of cash then I'd happily invest some of it turning it into a bit of a science project!

 

[Salt]

No one is assuming that. Measurements are specific to the model being tested. Further, everyone knows that they could get a faulty product.

It is strange to bring up such concern in this context because we have no less than 3 measurements! One from me, another from Erin and a third from European publication. With respect to such things as frequency response, the correlation for something mechanical with different people measuring them is actually quite excellent. Take this overlay of PIR that a member put together:

This indicates very high level of correlation between these independently measured speakers.

From what I recall, the CEA-2034 considers 3 dB variation to be good enough. Whether you use my measurements or Erin's, story is the same here.

Yes, distortion measurements vary. And I hear clear artifacts in a song I played. I consider this faulty driver/manufacturing. The probability of this happening when you buy the product either bothers you, or doesn't.

What it doesn't do is entitle the membership to demand that we buy another $700 speaker and waste opportunity cost of $2,000 to measure another sample. If it is that important to you all, then raise the $2,700, donate it to the forum and I will buy and test another.

Until then, this is not a science project. By now, ELAC has read or should have read my review and they have done nothing to deal with this situation.

Net, net, my testing is highly representative of product performance. There is some degree of unknown which I cannot eliminate. You have to be comfortable with this as without my data, that unknown rises to near infinity seeing how the manufacturer is not provide any measurements.

I think, he means, that testing just one sample of hundreds of thousands of the same production output might give an underestimate of the variety of the whole bunch, what indeed might reflect reality.
But reality in testing audio gear is simply evaluating one single item, and this is reality for last decades.
Two pair of shoes.

Besides QC should be done before release in the factory.

 

[Marin]

I have this speaker and it has a nasty cabinet resonance around 190Hz. even knocked over my wooden ashtray that I had placed as a decoration on top of the speakers.
I was playing music around 90Db continues level all evening in dark room and in the morning my ashtray was on the ground???I put my hands on the speaker and it was vibrating like sex toy
All this happened until I couldn't hear any overtones of that resonance even a few centimeters from the sides of the speaker.
I certainly don't have a golden ear at age 47 and I didn't go through harman training, but I would definitely hear something other than music.
I'm surprised no one has explained that since the box vibrated like crazy and could dance on the table however it wanted while the Klippel machine was taking measurements and that is how measurements around 1000Hz got corrupted!
I don't believe Armim was sitting there during the measurement and make sure that doesn't happens?
Even blue tac didn't help spikes had to be mounted and that stabilized the speaker.
This speaker has oversized motor structure and it weighs only 15 kilos.

 

[smclk89]

I found a df63 at a very affordable price and bought it just to try it out. I've been using the MA SİLVER 100 7G for a year. My only comment for ELAC is that although it's a tower speaker, it can never replace a 100 7g. The treble is virtually non-existent. At low volumes, the sound feels like it's coming from logitec satellites. Thank goodness I got it for the price of a DAC, otherwise I would have died of grief. The Ma Silver is a real speaker. A true champion, indeed.

 

[kyuu]

If the 100 7G is anything like the 50 7G, sounds like you're used to a tilted up treble response. Which is fine, and can make sense if you only listen at low volumes. Wouldn't want to listen to that MA speaker at elevated volumes though.

 

[smclk89]

If the 100 7G is anything like the 50 7G, sounds like you're used to a tilted up treble response. Which is fine, and can make sense if you only listen at low volumes. Wouldn't want to listen to that MA speaker at elevated volumes though.

Yes, that's exactly what I was looking for. There's something about these Silvers that bothers me for no reason. There's something about the bass tone that doesn't make me feel good at certain moments. I think each speaker definitely has its advantages and disadvantages. I'll try to get used to these elacs because the bass rhythms are very good.

 

[Marin]

Yes, that's exactly what I was looking for. There's something about these Silvers that bothers me for no reason. There's something about the bass tone that doesn't make me feel good at certain moments. I think each speaker definitely has its advantages and disadvantages. I'll try to get used to these elacs because the bass rhythms are very good.

Yes, I understand what you're talking about, but that function that you obviously like is implemented in AVRs, integrated amplifiers, etc.
It has different names like in the early days "loudness" Button was always on on my dad Marantz stereo amp.
Now days Yamaha has mix polcy about that, like stereo integrated Yamaha use variabile control that you can use as you like or not at all.
I have Yamaha AVR and they implement fix control like Denon, and others manufactures, that is coled "YPAO volume" and it's always on even if I em purist and all other enhancement are off like EQ,and "adaptive dinamic range" that implement some harshness at higher volumes,so it off all the time.
YPAO don't do that and I really like that function because it adds some life in misic at low volume.

 

[Marin]

From inside



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

O I was one of those if you have some reinforcement that you OK but speakers are that is being built get smaller and lighter no wonder he will have problem controlling those 3 woofers.
You can simply call this reinforcement, thank God the standards are stricter for building houses, otherwise we'd all be screwed

 

[Jimster480]

Interesting but not surprising, most speakers seem to have glaring flaws these days... I really wonder what companies are doing when they engineer such speakers and why they don't test them with a comprehensive set of test tracks...

 

[smclk89]

Interesting but not surprising, most speakers seem to have glaring flaws these days... I really wonder what companies are doing when they engineer such speakers and why they don't test them with a comprehensive set of test tracks...

In fact, it's the opposite; each speaker has great capabilities, but it also has shortcomings compared to other speakers with great capabilities.

 

[Marin]

This is a review, listening tests, and detailed measurements of the ELAC DF63 Debut 3.0 Tower speaker. It was kindly drop shipped and donated by a member. It costs US $649 each.
View attachment 474472
The DF63 looks no worse or better than I expected which means, it is fine. Nothing exciting in the back other than the two ports:

View attachment 474473
I appreciated close to the ground terminals as there is less chance of speaker wire resonating against the cabinet. And rear ports means that their high frequency distortion is less likely to be audible.

As usual, if you are not familiar with my speaker measurements, please watch this video first:

ELAC DF63 Speaker Measurement
As usual, we start with our family of anechoic frequency response graphs generated by Klippel Near-field Scanner:
View attachment 474474
If you stand back (or filter the graph), the on-axis response generally looks level. Closer examination shows something untoward happening around 900 to 1000 Hz. And a few resonances at 90 and 180 Hz. Treble response also has some variability. But again, at high level, this is not bad.

Sensitivity is specified at 87.5 dB and it gets there, making it about 1 to 2 dB better than average (can get louder with the same amplification power). This is a common benefit of tower speakers with multiple woofers.

Directivity is reasonably smooth although doesn't follow a monotonic trend. This messes (badly) with early window reflections:
View attachment 474475

And as a result, in predicted in-room frequency response:
View attachment 474476

Showing just one port and woofer, we see the reason for some of the anechoic frequency response errors:
View attachment 474477

While this is a 3-way rather than (messy) 2.5 way speaker, it still suffers from multi-sourced interference including a port/cabinet resonance.

Back to directivity, we see very well controlled one horizontally above 1 kHz:
View attachment 474478
View attachment 474479

Vertical is messy despite the midrange:
View attachment 474480

See the messiness around 1 kHz in both graphs.

Distortion test generated very surprising results:
View attachment 474481
View attachment 474482

I have not ever seen such a massive jump in distortion with 10 dB in this class speaker! We go from better than average to worse than average. Here is a comparison with 5 dB increments:
View attachment 474483

I could hear some audible distortion at 96 dBSPL which got highly exaggerated at 101 dBSPL, rattling the entire structure of the Klippel NFS. In other words, what there is, is amplified by the cabinet.

Impedance is on the low side:
View attachment 474486

Waterfall and step responses are as expected:
View attachment 474484

View attachment 474485

ELAC DF63 Speaker Listening Test
Listening tests were made in our massive (volume wise) living room as you kind of see in the review picture. Immediate impression was that of (pleasant) warmth and full range response I don't expect to hear in this class of speaker. Fidelity was good enough that I started to question what I had measured. So I jumped right into my sub-bass test tracks to chase down the distortion there:

Sub-bass was reproduced better than this speaker has any business delivering! I almost stopped there but glad that I did not. At 30 seconds, there are some high frequency (strings?) that normally don't bother any speaker I have tested. Oh boy. Did it upset the DF-63. Massive, and I mean massive resonance set into tweeter, creating screeching sound that was nearly as loud as the notes themselves! I reduced the volume down to some -10 dB but I could still hear it. My wife was next to me so I asked her if she could hear it at around -5 dB and she could! This is a flat out failure in my book and reminds of another ELAC speaker with the same problem although here, it is far worse.

I tried to detect the problem in other clips and it was much harder. I could hear strangeness in high notes in other tracks but wouldn't bet my salary on it and nowhere as clear as the above track. I can't imaging it not being there in other clips though.

As an aside, above is a great example of using carefully selected audio tracks for speaker testing, than whatever you normally listen to.

On bass, the deep notes were very satisfying. As I cranked up the volume, I could tell they would start to get distorted and muddy but this was at rather elevated levels.

On EQ, I filled in the 1 kHz hole and it opened the sound fair bit, making the stock sound wooly and too warm and closed without it.

Conclusions
The DF63 seems to have a different recipe than many other speakers I have tested. It seems to aim for deeper bass at the expense of higher distortion/lower SPL playback. There is also some interference around 1 kHz which EQ seems to paper over. The main issue as you can imagine, is that tweeter. I don't mind gradual distortion or power limited but not this kind of massive and sudden break up. Yes, I was listening kind of loud. But once I got sensitized to it, even playback at moderate levels was audible.

On the positive front, the deep bass is extremely satisfying, making the speaker nearly full range which is remarkable in this price class. Sensitivity is higher than normal, meaning less demand on the amplifier.

I am sad to give a failing grade to ELAC DF63. Some technical flaws I can't get over and such is the case with tweeter response here.
-----------
As always, questions, comments, recommendations, etc. are welcome.

Any donations are much appreciated using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/

I think I finally understand what is wrong with these speakers around 1000hz.
So the problem is in the crossover, if you look where the woofer has a notch, the midrange also has one.
This is because the crossovers between woofer-midrange and midrange and tweeter connected, Andrew Jones talked about in first series that this is because of reducing costs and no need to change or switch parts or their values much between book shelf model and Towers.
I don't know how to describe it exactly, maybe someone knows better what it is about, but it seems to me that the two crossovers are connected and where there is a deep notch for the woofer also affects the midrange?
This is because two 4 ohm woofers are connected in series and that gives about 6 ohms to drop to 3 and a half where they connect to the midrange.
Andrew Jones use the same thing Although he preferred 8 Ohm speakers(with voice coil around 5.5) so it would be 11 or 12 Ohms around the port frequency and drop to 5 and a half where they connect to the midrange.
The same thing happens to my first series F5.
Unfortunately, I don't have klippel, but here is pink noise and the spectroid app (look red line,) near field measurements