OP
- Thread Starter
- #361
Answer is revealed in the Revel F35 review just posted: https://www.audiosciencereview.com/forum/index.php?threads/revel-f35-speaker-review.12053/
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
KEF R3 Speaker Review
- Thread starter amirm
- Start date
Interesting review Amir, myself and my HiFi buddy assessed R3s on few different occasions and always came back with the same wonderful experience. It does pack a punch for a stand mount. We shall have a listen to the Revel M16 in near future based on your review.
- Joined
- Dec 31, 2018
- Messages
- 64
- Likes
- 82
I wasn’t trying to be rude. But, the reality is that just two units tells us nothing “useful” about the manufacturing sampling variation - unless only two units were made of course.
Suppose a 1,000 units were manufactured and you test two and two of them are good. Assume the distribution is bimodal (only “good” = meeting specs or ”bad” = not meeting specs). The variation could still be anywhere between 998 bad & 2 good to 1,000 good and 0 bad. If you want to argue that my example does indeed say “something” about the variation - then I’ll concede that is true. However I don’t think it’s a very “useful” something - certainly not enough to warrant changing people’s mind about making purchases. I’m simply making an argument that, statistically, drawing any meaningful conclusion about manufacturing variability on a sampling of just two units could be completely wrong.There have been some cases where a unit measured poorly (on this site) and the manufacturer has contacted @amirm to provide a second unit for testing. The measurements often prove to be much better on the second unit. In these cases the only thing we really know at the end of this is that (1) the manufacturer cares enough about the measurement results on this web site to try to get a poor result fixed, and (2) the “architecture” of the device is “capable” of achieving the measured performance. We really still cannot conclude anything about unit-to-unit variation or if components were hand picked for the second measurement or not. The initial (poorly measuring) unit might have been a one off bad unit, damaged during shipping, etc. or it might be that many units perform poorly and only a few happened to have the correct set of internal components that enable the architecture to measure well. But we won’t know unless the issue (with the first unit) is ”root-caused” and results published with the before and after measurements. Even then we’d be trusting nobody was lying to us.
@amirm ’s work here is so important (IMHO) because it at least identifies companies which are just lying about (or not even measuring) the performance of their products. If it measures poorly and the manufacturer doesn’t at least supply a unit capable of reaching the specifications - then we know that at least one unit is bad and that they don’t care enough to provide a corrected/fixed unit. (Or they’re not aware of the site and the review?
) In this case there may be units out there that actually do meet specifications but we don’t know if any of them exist or not. However, if the unit measures well and meets specifications then we do indeed know that at least one unit was able to reach specifications! After buying a unit ourselves we’d really need to measure it to know if our specific unit also meets specifications (unless it is so bad that the defect is audible). But at least now we know it’s possible to get a ”good” unit.Again, I don’t mean this to come across as a rant or to be rude. I belabored the point a bit simply because I’ve seen people very disappointed because they’ve drawn the wrong conclusions about a large distribution from too small a (and/or a poorly selected) sample set.
Random aside, but one thing I don't think that has been mentioned in this thread with regards to the bass is that KEF includes a two-step port bung.
Though I did not measure the effects when I had the speaker, I do remember subjectively it had a significant effect. With my own placement nearish to the front wall, I thought using the outer ring of the port bung sounded cleanest without losing too much bass. Here's the vague image from KEF's manual, in which they seem to suggest using the port bungs for placement close to wall.
I did not need this with EQ and a sub, but it could be valuable for cleaner sound for those without a sub who plan to keep the speakers near the wall.
Though I did not measure the effects when I had the speaker, I do remember subjectively it had a significant effect. With my own placement nearish to the front wall, I thought using the outer ring of the port bung sounded cleanest without losing too much bass. Here's the vague image from KEF's manual, in which they seem to suggest using the port bungs for placement close to wall.
I did not need this with EQ and a sub, but it could be valuable for cleaner sound for those without a sub who plan to keep the speakers near the wall.
tuga
Major Contributor
Could you confirm if using both bungs seals the cabinet?
I do not have the R3 any more(Im a reviewer and this was many moons ago). From recollection doubt the foam was thick enough to create a perfect seal, but it did have an audible effect on the bass.
- Joined
- Jan 15, 2020
- Messages
- 6,903
- Likes
- 16,917
The included double bung certainly does it on my LS50, don't have a R3 though.
tecnogadget
Addicted to Fun and Learning
They completely seal the back of the port. If your concern is about the bungs making it behave similar to a sealed box, I can't give you a confirmation without measuring it.
All I can say is I do play a lot with the positioning of the cabinets and put the bungs in and out just for the fun and testing (my current liberty of movement for the speakers is ridiculously small until I make room modifications and change furniture). They are an extremely useful tool for special cases. Once I get final placement, manual PEQ or Dirac and subwoofer, there won't be any need for them...or at least for swapping between no bung, half bung or full bung.
tuga
Major Contributor
I agree. Had a pair of PSB T2s not too long ago which came with rubber plugs for each of the 3 individual ports:
tecnogadget
Addicted to Fun and Learning
tuga
Major Contributor
I'm not absolutely certain but think that I left one of the ports open on the right speaker too.
As you've probably realised the speakers were not positioned symmetrically in the room:
I also have R3s, and had a pair of Revel M20s in to compare.
My measurements (and those of my Anthem ARC room correction) generally correlate with the others here.
Once ARC has does its magic the dip at 1.5K is filled in giving results much like that flat R3 graph earlier.
Subjectively, I preferred the EQed R3 to the EQed M20.
I find it more natural. The M20 was more lively.
My measurements (and those of my Anthem ARC room correction) generally correlate with the others here.
Once ARC has does its magic the dip at 1.5K is filled in giving results much like that flat R3 graph earlier.
Subjectively, I preferred the EQed R3 to the EQed M20.
I find it more natural. The M20 was more lively.
I second that, when we used to live in the rented accommodation my LS50s were placed close to the rear wall (~25cm) and I had to use the bungs all the time. Now they are at least a metre away so no bungs.
I find it interesting that Amirm is finding that this speaker sounds 'uninvolving'. After a very respected audio engineer told me he discovered that the coax driver series that KEF use here measures phenomenally well a couple of years ago when the KEF Blade was released, I went to hear a set local to me in a well acoustically treated room.
Awful.
I wish I knew why. I knew they measured very well.
Tried an A/B test against an ATC SCM11. Ha! No comparison. Subjective comments are just that, but........
The KEF sounded dead, boring and lacked any kind of drive some how. Even with the bass EQ'd up, it just didn't sing. Detail seams to be there, but not in the upper mid.
Amirm: I have a sneaky suspicion that if you ran a multi tone test- they will show up that coax is no good. Remember, you subjectively liked the JBL 305......and that has less bass. I know you know there is more to this speaker game
What you heard was the Kef midrange vibrating and modulating the tweeter. A problem with all loudspeaker drivers with the tweeter in center of the midrange. The midrange membrane is acting like a waveguide to the tweeter. As the membrane is vibrating , the waveguide for the tweeter constantly changes, changing the frequency response of the tweeter when playing music. The measurements may look fine, though.
If you think that modulation is not visible in frequency response nor in THD measurement why do you think it is audible at all?
Although that might be a problem with the LS50/Q150, in the R3/300 the midrange crossover is 400hz so the cone is moving 2-3 mm, hardly enough to present a problem.
The measurements are made with static signals. Music doesnt work that way. An acoustic piano plays all the frequencys at the same time, meaning it will exite for example 30-20000 Hz at the same time.
The problem with coaxial drivers can be seen with measurements playing 300 Hz tone and 8000Hz at the same time, at high levels.
Are you trying to say coaxial drivers would fail IMD multitone test?
Im sorry to say there is still a minor problem. Just consider the wavelengt of 10000 Hz . How much does a tweeter have to vibrate to play that frequency ?
But you are right that the problem is much lesser when crossing over at 400 Hz, than fullrange.
It would be interesting if Amirm could do a test like that.
Similar threads
