Page 1 of 2 12 LastLast
Results 1 to 10 of 19

Thread: NFS Luna / Duo Follow up

  1. #1
    Join Date
    Aug 2003
    Location
    USA
    Posts
    4,606

    Default NFS Luna / Duo Follow up

    Now that the dust has settled there is a bit more to discuss so a follow-up with pertinent data is important.

    I have had our Klippel NFS running, on average, between 1-2 tests per day since we assembled it. I would estimate I have logged about 600 hours of usage. At this point, I would say I am as qualified and experienced with it as anyone else who doesn’t work directly for Klippel.

    First off, the Klippel NFS is a tool and like all tools, it is only as good as the operator. It is anything but a “hit the button and walk away” testing device. Far from it, there is a lot of room for error in the setup of each speaker. When used properly and with enough knowledge and experience, it is without question, the most accurate device for full range anechoic directivity measurements currently available.

    There are basically 2 ways of using the NFS. For directivity, once various parameters and measurement distances are set, the robot goes to work moving the microphone around the speaker taking potentially thousands of measurements which are then compiled, and various calculations are processed. This is where field separation and gating algorithms are applied to remove the room and generate an anechoic response. You see these measurements in ASR’s CEA-2034 graphs, early reflections, estimated in-room and horizontal and vertical directivity. These directivity measurements are mathematically computed.

    ASR’s other graphs, near field driver components, distortion, CSD (waterfall), impedance/phase are different and do not utilize the advanced field separation functionality of the device as these are single sweep measurements.

    To take these measurements, the operator uses a remote control to position the microphone to where he/she wants it, then they manually perform the transfer function sweep depending on which measurement they are taking. It is not much different than any other quality loudspeaker testing device, only difference is that the NFS does induce some minor reflections in the results (these mostly get averaged out in the directivity results) and precise microphone placement is complicated due to the way the NFS stepping motors work and the limitations in measuring distance (difficult to measure at a distance further than ~ 30”, but it can be done)

    These manual measurement sweeps also allow for gating to remove the effects of the room.

    One of the first things I did with our NFS was to check it against multiple references to determine its accuracy. With our good friends and top tier engineers at SEAS, they took directivity measurements of a driver in their anechoic chamber. I then compared their results with our results using the same exact driver (they sent it to me) and I was very happy to see how closely the measurements aligned.

    I also purchased a Revel M105 to compare results with Amir’s review of the M105.

    Here is our measurement of the M105:

    m1052034.jpg

    Here is Amir's:

    https://www.audiosciencereview.com/f...nts-png.73659/

    If you compare with Amir’s measurement, they are similar but not quite the same. You can see in our measurement many more jagged ripples. We have setup our NFS to use 20 points per octave, which is called for in the CEA-2034 spec. Just by looking at Amir’s measurement of the m105 one can see that there is quite a bit of smoothing. The more smoothing, the better the measurement looks. Our measurement of the M105 does not look nearly as good.

    I do know Amir took these measurements over a year ago, perhaps he has changed his setup parameters since then to use 20 points per octave. His more recent measurements appear less smoothed. His more recent measurement of the Revel F228Be do not look nearly as smoothed as the m105 and do appear to be at 20 points per octave.

    Here are our NFS directivity measurement of the exact same Luna he tested:

    OGLunaCEA2034p.jpg

    Here is Amir's:

    https://www.audiosciencereview.com/f...nse-png.90567/

    There are some differences here as well. In our measurement, we are getting a slightly different low end response (more midbass and doesn’t roll off as soon in our measurement). I believe Amir addressed this with different measurements later on, by using a microphone starting point further away from the speaker. Still, they are relatively close in this regard.

    I have more concern about the large peak in the tweeter response at 6-7 kHz in Amir's measurement. I immediately recognized this as being vertical mic positioning, something I was able to confirm with the engineers at Klippel after examining Amir’s data. This isn’t really the fault of Amir, but with the ribbon tweeter being quite directional vertically, this gets amplified with nearfield measurements such that even a cm off from dead center of the ribbon will not be quite accurate. This really isn’t a concern with domes, but one must really have experience with true ribbon tweeters in order to be able to accurately and fairly measure them.

    More to this same point…

    Here is Amir's Luna nearfield driver components measurement:

    https://www.audiosciencereview.com/f...ses-png.90568/

    In Amir’s nearfield driver components measurement, his measurement of the tweeter is far off. This relates back to the manual measurement sweeps I mentioned above and having both the knowledge and experience necessary to trust a measurement prior to publicly publishing it.

    The NFS took this measurement accurately, but – this was taken with a mic distance of approx 1cm. A true ribbon tweeter like a RAAL can not be accurately measured in the extreme nearfield. Different sections of the ribbon produce different frequencies, with the middle of the ribbon (where there is the most flex) producing the lower highs. With the mic at dead center and at 1cm away, this is what the NFS mic picked up, where you see the highs roll off sharply starting at 10kHz.

    To accurately measure the ribbon, the mic must be positioned at least as far as the total vertical height of the ribbon (or as far as the longest length, be that vertical height or horizontal width). I recommend measuring at a distance of 3”. This is with any measurement device, NFS, MLSSA, LMS etc.

    Here you can see the differences using the NFS at different microphone distances:

    Luna tweeter at 1cm.jpg

    Luna tweeter at 3 inches.jpg

    Notice the rather extreme difference in the response above 10kHz. Ignore everything from about 3kHz and lower as the NFS is picking up the response from the woofer.

    The same goes for a slot port, if the mic is positioned to closely, it will pick up all sorts of artifacts from air-turbulence and reflections.

    OG Luna port at 1cm.jpg

    OG Luna port at 3.jpg

    Note, the above measurements are taken with no gating and no smoothing and are not mathematically derived. These are highly accurate and perfectly match measurements we have taken with our other test equipment.

    What is interesting here is notice how the nasty port resonance at 600Hz, taken with a mic distance of 3" is dramatically reduced compared to the 1cm measurement.

    Now, here is where things getting very interesting.

    Here is the single sweep transfer function measurement of the original Luna taken with a mic distance of 22.5". This measurement does not use any gating nor does it use field separation, there is no mathematical processing or smoothing. As such, there is some influence from the room.

    OG Luna at 22.5.jpg

    Notice in the above measurement, the port resonance at 600Hz is still somewhat evident, but dramatically reduced and not nearly as offensive as the directivity measurements showed.

    Here is the same measurement, but gated to remove room effects:

    OG Luna at 22.5 gated.jpg

    So, where am I going with this?

    Here is the measurement of the same Luna taken with the NFS at the industry standard of 1 meter. Again, this is a single sweep, non-processed measurement that is not mathematically derived from nearfield measurements.

    OG Luna at 1 meter.jpg

    The peaks and dips are now larger because there is more room influence due to having a further mic distance.

    Let's gate this measurement and remove the room:

    OG Luna at 1 meter gated.jpg

    Where is that horrible port resonance?

    Let's add a bit of smoothing:

    OG Luna at 1 meter gated smoothed2.jpg

    Now, if you examine the above measurement and compare to our published measurement of our reference Luna, they do track very closely.



    Overlaying the 2 measurements on top of each other, and throwing out everything below 200Hz.

    compare.jpg

    These 2 gated measurements taken with a mic distance of 1 meter, one of our original reference Luna speaker taken using our MLSSA system, and the other of the ASR measured Luna taken with our NFS do indeed closely match each other.

    The following is important:

    Below is our 1 meter NFS measurement of the OG Luna, that is derived mathematically from over 800 nearfield measurements using field separation + gating to remove the room compared to the NFS single sweep gated measurement. Derived is in Red, measured in Blue:

    OG Luna derived vs measured.jpg

    These 2 measurements track each other closely, except at the 600Hz port resonance.

    I have a few important conclusions to make.

    First and foremost, nasty comments were thrown our way because our published measurement of both the Luna and Duo did not match Amir's published directivity measurement. I have proven a few things here:

    Using NFS gated measurements, our published measurements do closely match the Luna Amir tested. It is only in the derived directivity measurements where there appears to be large discrepancies, and this is mostly centered around port resonance frequencies.

    The NFS directivity measurements are highly accurate and are capable of the highest resolution possible today. There is simply no other test equipment available that comes close. However, there is still a nagging question regarding how NFS field separation calculations handle the response from a front slot port generated resonance. I suppose at this point, the only way to confirm is to measure this same Luna in an anechoic chamber, which I intend to do sometime in the future.

    As accurate as the NFS is, there is still plenty of room for mistakes to be made.

    Regardless of whether or not the port resonance picked up in the directivity measurements is audible, Duo V2 and Luna V2 show no port resonance in either NFS directivity measurements or NFS single sweep gated measurements
    .
    .
    .
    Good Sound To You!

    David Fabrikant
    www.ascendacoustics.com

  2. #2
    Join Date
    Mar 2004
    Location
    SouthWest of Cleveland
    Posts
    1,781

    Default Re: NFS Luna / Duo Follow up

    Dave, thanks for your knowledgeable post. Although I have no plans on owning either the Lunas or Duos I always enjoy your technical expertise!
    Ed

  3. #3
    Join Date
    May 2007
    Location
    Stouffville,Ont..
    Posts
    338

    Default Re: NFS Luna / Duo Follow up

    Thorough as always Dave, keeping it simple for us common folk. We appreciate your hard work and also understand it's your passion.

    Regards, Bill
    Speakers: Towers & STC(w RAAL), MA CP-WT260 In-Wall
    Sub: Funk Audio 18 in Santos

    Source: Martanz NR1710, Oppo BDP 103D + UBK-90 4K & LG B9 65"

  4. #4
    Join Date
    May 2015
    Location
    Houston, Texas
    Posts
    775

    Default Re: NFS Luna / Duo Follow up

    +1 on Bill's and Ed's comments (although I currently have the Luna Duo center until my Horizon ships :-) )

  5. #5
    Join Date
    Nov 2011
    Posts
    351

    Default Re: NFS Luna / Duo Follow up

    Two things stand out to me in this further investigation. As mentioned previously, measuring speakers is not a simple thing to do accurately, and no matter the technology involved, a professional with decades of hands on experience will have a better understanding of how to interpret the results from that technology.

    Well done Dave, I think you will help Klippel in understanding the anomalies between measured and derived results, it would appear there is some work left to be done.

    Jay
    Last edited by petmotel; 10-13-2021 at 01:26 PM. Reason: grammar correction

  6. #6
    Join Date
    Aug 2003
    Location
    USA
    Posts
    4,606

    Default Re: NFS Luna / Duo Follow up

    Thanks all. This is really an interesting study, well - at least to me

    I have spent way too much time and money on this and I don't foresee us recouping the cost of our Klippel NFS purchase (anytime soon). However, from the standpoint of what I do, of what I have been doing since around 1987 or so, it was well worth it.
    .
    .
    .
    Good Sound To You!

    David Fabrikant
    www.ascendacoustics.com

  7. #7
    Join Date
    Nov 2011
    Posts
    351

    Default Re: NFS Luna / Duo Follow up

    Quote Originally Posted by davef View Post
    Thanks all. This is really an interesting study, well - at least to me

    I have spent way too much time and money on this and I don't foresee us recouping the cost of our Klippel NFS purchase (anytime soon). However, from the standpoint of what I do, of what I have been doing since around 1987 or so, it was well worth it.
    Not quite sure if I'm hearing a bit of buyer's remorse mixed with rationalization, or if you're actually quite pleased that you now have an absolutely state of the art speaker lab. Maybe a mixture of both. In uncertain times, it probably isn't one's first instinct to make a huge investment in R&D.

    In any case, as always, it was a memorable journey, huge props that you are considerate enough to give your customers a glimpse behind the curtain. I can't think of many (any actually) manufacturers that operate at this level of transparency. Thanks for ride along, great stuff!

    Jay

  8. #8
    Join Date
    May 2007
    Location
    Stouffville,Ont..
    Posts
    338

    Default Re: NFS Luna / Duo Follow up

    Quote Originally Posted by davef View Post
    Thanks all. This is really an interesting study, well - at least to me

    I have spent way too much time and money on this and I don't foresee us recouping the cost of our Klippel NFS purchase (anytime soon). However, from the standpoint of what I do, of what I have been doing since around 1987 or so, it was well worth it.
    Dave just think of this (klipple) investment as having that Porsche 918 Spyder in the garage... it's there when you want to take it out for a spin...
    Speakers: Towers & STC(w RAAL), MA CP-WT260 In-Wall
    Sub: Funk Audio 18 in Santos

    Source: Martanz NR1710, Oppo BDP 103D + UBK-90 4K & LG B9 65"

  9. #9
    Join Date
    Aug 2003
    Location
    USA
    Posts
    4,606

    Default Re: NFS Luna / Duo Follow up

    Quote Originally Posted by petmotel View Post
    Not quite sure if I'm hearing a bit of buyer's remorse mixed with rationalization, or if you're actually quite pleased that you now have an absolutely state of the art speaker lab. Maybe a mixture of both. In uncertain times, it probably isn't one's first instinct to make a huge investment in R&D.

    In any case, as always, it was a memorable journey, huge props that you are considerate enough to give your customers a glimpse behind the curtain. I can't think of many (any actually) manufacturers that operate at this level of transparency. Thanks for ride along, great stuff!

    Jay
    Sorry, I didn't mean to portray that conclusion - I have no buyers remorse whatsoever.

    I am just a little frustrated as to how much time I have spent on this and have yet to come to a final conclusion as to which measurements accurately represent the port response in the far field. I wish I could just let this go but I can't, lol. I need to know for my own curiosity and continued research.

    I am thinking a might call in a favor with my friends at SEAS. Send this speaker to Norway and have them measure it in their chamber. They do not have an NFS (few companies do) but they do have a true anechoic chamber and use Klippel gear for their measurements, so this would be an appropriate comparison. The results of those measurements would give me my answer.

    It's a real pity that here in the states, gaining access to a chamber - even for scientific purposes - is nearly impossible.
    .
    .
    .
    Good Sound To You!

    David Fabrikant
    www.ascendacoustics.com

  10. #10
    Join Date
    Sep 2014
    Posts
    4

    Default Re: NFS Luna / Duo Follow up

    +1 for me on Duo v2 availability list! (long term Sierra v2 owner, recently picked up 4 HTM-200SEs for surrounds for new basement HT setup)

    This has been quite a ride. I've been saddened by the cynicism about the measurements, awed and inspired by Dave's reactions in word and deed.

    It amazes me in these kind of niche "hobbyist" markets (my other personal one being hobby boardgames) when people's first thoughts jump to cynical assumptions when something isn't quite right. The people working in this kind of industry (especially at Dave's scale) are so clearly driven by the love of the thing and could be making more money for the same or less effort doing something else.

    Thanks for all the dedication and, as others have said, the transparency - it's fascinating to have a window into all of this.

    -vas

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •