As for the original question: yes, I could see paying something like $500 for such a service. The problem though is that designs get refined in an iterative way. So, while the benefit is there, it is somewhat limited to a diyer looking to improve a given design. Of course, one would still learn a lot from the results.
You rarely need accurate polar data below 300 Hz, in fact with most designs you don't need it below 1 kHz.
There is no "hate" for Klippel - but there is dislike like for the notion effectively presented that you actually need a Klippel to produce good results. THAT is an utterly stupid premise rooted in extremism.
"Large" is relative (particularly with what Erin was doing: batch measurements).
So lets go through his post then:
1. Problems with baffle-step integration. This is well documented with nearfield and farfield measurement splicing (including port contribution), and yes - it might well require some to test outside on a fairly high stand to achieve a confidence level with what they are doing inside to achieve a good result. Still though, what is "off a fair bit" - and is it particularly meaningful (+/- how much db) in any given room the way the room alters results in the modal range? It might well be meaningful to Erin in his two rooms in the positions he places his loudspeakers, but is that going to be true for most? Ideally you'd want that flat anechoic response, but chances are you aren't going to achieve that with a room so you shoot for a reasonable guesstimate realizing that for most they are never going to actually achieve that result regardless of the loudspeaker's anechoic behavior. Note: in almost all instances the NFS is better, but let's not cr@p on what can produce a decent result with serious attention paid to achieving that result.
2. Strong port resonances are apparent with good nearfield measurements.
3. Strong enclosure resonances are NOT particularly apparent with the NFS. While Erin may be correct with what he has pointed out with problems with some speakers he's measured - there is no definitive proof that what he's hearing as a problem and what he thinks is actually a problem with a small magnitude deviation in linear response is correct. When he started doing this on his videos his confidence level was exceedingly cautious and overtime - not so much. Maybe he is correct? I think he likely is, but I can't say that he actually is with any confidence on my part. As it is - sure if you look close you can spot a deviation in linear response with the NFS that you likely wouldn't have measured in the far-field in a standard room (or could substantively correlate with from the near-field). As it currently "stands" I don't think the NFS is particularly compelling *here.
4. Most people don't use ground-plane measurements for accuracy checks regardless of their understanding. (..it's kind of a shame really, its a great tool - but as noted you need to understand how best to use that tool if you are going to make use of it.) Anecdotally: I rather like Zvu's use of renting his local grade-school's INTERIOR basketball court for ground-plane measurements, in fact I rather like the idea for stand-mounted measurements.
5. I"ll flat-out say the NFS isn't necessary (for DIY), because I don't have extremest views on exactly what is necessary and because the tools more commonly available to the DIY'er IF USED FULLY AND CORRECTLY can do a good job. It may not be a fully "accurate" result - so what? I personally want something that's closer to accurate but that's me, some will just put a fullrange driver in any box and be relatively happy (for a while).
*an exception to this is where the "port" (as with an un-damped transmission line) actually modulates the result (combing) and even modulates the driver itself. In this circumstance I'd want the NFS or realistically: would take the time to do a proper ground-plane measurement (..and splicing isn't really the point of doing this: instead it's about looking for serious "sawtooth" problems with linearity). This sort of design (with a driver on a T-Line operating at higher freq.s) is pretty rare though (at least for multi-ways). Troels has a few DIY designs like this and my guess is his measurements with those loudspeakers are actually FAR worse than what he's managed to capture. Ironically though - they can still sound pretty good despite measuring like cr@p.
Last edited:
But the whole point is what can you do INSIDE
I don’t want to go outside. Where I’m currently living it’s 80% humidity nine months a year. That’s monsoonal weather, tropical cyclones mosquitoes, viruses, soil borne germs of the likes you never even heard of, even if you are a physician.
Another place I’ve lived it’s raining nine months a year, as soon as the sun is out, people are dying for a stroll for some sunshine- ain’t nobody gonna want to go outside to measure anything.
The whole point is what can we do inside indoors using traditional methods, and can it be as accurate as an anechoic chamber or NFS- that remains to be seen.
What do you need, may be different to what I need.
some people don’t need speakers and listen to music from their laptop or earbuds.
Those fellows over on the full range forum don’t need no crossover. Nor good sound 😝
Other want to advance to their understanding and the state of the art. They stopped buying when they realised that nothing as good enough for their needs.
I drive my crapbox to get groceries and pick up the kids, but I don’t tell the weekend race car driver and his fellow enthusiasts they don’t need a dyno, underground pit, or racetrack to test, modify and build better cars do I?
Where do you think outfits like ProDrive start from?
If I do it’s because I don’t care about things getting better.
You guys should just all drive Toyota hybrid crapboxes like me. Living room on wheels.
I don’t want to go outside. Where I’m currently living it’s 80% humidity nine months a year. That’s monsoonal weather, tropical cyclones mosquitoes, viruses, soil borne germs of the likes you never even heard of, even if you are a physician.
Another place I’ve lived it’s raining nine months a year, as soon as the sun is out, people are dying for a stroll for some sunshine- ain’t nobody gonna want to go outside to measure anything.
The whole point is what can we do inside indoors using traditional methods, and can it be as accurate as an anechoic chamber or NFS- that remains to be seen.
What do you need, may be different to what I need.
some people don’t need speakers and listen to music from their laptop or earbuds.
Those fellows over on the full range forum don’t need no crossover. Nor good sound 😝
Other want to advance to their understanding and the state of the art. They stopped buying when they realised that nothing as good enough for their needs.
I drive my crapbox to get groceries and pick up the kids, but I don’t tell the weekend race car driver and his fellow enthusiasts they don’t need a dyno, underground pit, or racetrack to test, modify and build better cars do I?
Where do you think outfits like ProDrive start from?
If I do it’s because I don’t care about things getting better.
You guys should just all drive Toyota hybrid crapboxes like me. Living room on wheels.
Last edited:
..the whole point is what you can REGULARLY do INSIDE (to achieve reasonable accuracy).
Really its about understanding what you are getting with your results inside. It might require that you do some testing on a similar type of design/project either outside OR in a large space inside (..again, like what Zvu is doing).
In this respect then, as you have mentioned - paying for the NFS to achieve that level of confidence could well be worth it IF you have no other good means of obtaining that more accurate comparison (as needed). With more common designs I doubt it's going to be needed provided you have reasonably sufficient space inside. IF you don't have that clear space inside: THEN it gets more difficult for those common designs.
Basically testing outside (or a large space inside, rented or not) shouldn't need to be at all frequent unless the designs you are working on are a-typical in certain respects (and notably a-typical from each other project-to-project).
-you know, a "base-line" for the room and conditions you are testing in. 😉
(..though you run into the same problem if you keep changing your overall testing setup: notably the room etc. and *mic. position.)
*and this can take a while to get used to (because reasonably assuming the room isn't used solely for this, you typically move the mic..
Really its about understanding what you are getting with your results inside. It might require that you do some testing on a similar type of design/project either outside OR in a large space inside (..again, like what Zvu is doing).
In this respect then, as you have mentioned - paying for the NFS to achieve that level of confidence could well be worth it IF you have no other good means of obtaining that more accurate comparison (as needed). With more common designs I doubt it's going to be needed provided you have reasonably sufficient space inside. IF you don't have that clear space inside: THEN it gets more difficult for those common designs.
Basically testing outside (or a large space inside, rented or not) shouldn't need to be at all frequent unless the designs you are working on are a-typical in certain respects (and notably a-typical from each other project-to-project).
-you know, a "base-line" for the room and conditions you are testing in. 😉
(..though you run into the same problem if you keep changing your overall testing setup: notably the room etc. and *mic. position.)
*and this can take a while to get used to (because reasonably assuming the room isn't used solely for this, you typically move the mic..
Always worth a look there, some real gems like NTK's work, just avoid the out of control train wreck threads (every forum sees to have them). Devout objectivists with only half the answers make for fireworks 🙂
I thought so but you referenced the NFS directly (perhaps unintentionally) which is why I wanted to point it out as there is much confusion about how it works.
Did you measure the woofer at the same time? The frequency range that covers with a higher signal to noise ratio would be helpful to see.
Don't automatically assume that because something doesn't measure well that the reason is because the manufacturer couldn't measure it properly. In cheap speakers of which Erin has measured quite a few the faults could be and often are due to cost cutting, in expensive ones it could be the designer wanting to stay close to the "house sound".
If they really were aiming for a different result and were in the dark as to what they made, then yes it could help them. Having an accurate measurement only helps you design a better speaker if you know what to do with it. The measurement shows you where it went wrong not how to fix it.
I'd gladly par a couple hundred for this service but I'm not really willing to pay for or risk the shipping. That is the big drawback as far as I'm concerned.
Klippel range od products goes as high as over 40k€ per specific unit. I don't think DIYers should have any interest in paying any kind of service in that matter. The whole point of Klippel is to tax companies earning big money(?) to earn a little more by presenting fancy looking charts.
If I dyno my car (certified data) it costs me 130€ which equals about 130US$. The procedure is done on a 140.000€ installation in a large garage. The measurement has very high energy (electricity) costs. It takes 30-45 minutes from start to end, as the car has to be tied down. This is only for verification, the real tuning is done on the road with extensive data logging, with equipment much less expensive, typical about 2000€.
So if I wanted one of my speakers measured, 50€ for a Klippel should be OK to learn what I messed up... Would it help me? Not much, if I can not try different configurations and x-over while measuring. So with speakers I use my cheap 500$ equipment, something that would have been industry standard before 2000 or Klippel. Even as Klippel today is the way to go professionaly, there have been some good speakers build without it in the past.
I think if the chassis manufacturer uses Klippel in it's design, most of the advantages have been realised, Maybe that is a reason I can get away without it, at least in my designs. In listening trials my speakers still beat any 5-10 times as expensive main speaker brands modell without question, after a few second of A-B test.
Any serious loudspeaker manufacturer uses a Klippel, as it is the only way to optimize good chassis. Have a look at their reference list.
By the way, Linn does not have a Klippel... which shows me they do not want to know what they did wrong on their products... If you only care for marketing, you don't need Klippel, that's a fact.
I wonder how many Chinese "devlopers" are trying to clone a Klippel, right now as I write this... most units sold seem to be in China. Good luck to our Chinese brothers! Is there something to interpret from the fact Russia does not have a single Klippel system?
So if I wanted one of my speakers measured, 50€ for a Klippel should be OK to learn what I messed up... Would it help me? Not much, if I can not try different configurations and x-over while measuring. So with speakers I use my cheap 500$ equipment, something that would have been industry standard before 2000 or Klippel. Even as Klippel today is the way to go professionaly, there have been some good speakers build without it in the past.
I think if the chassis manufacturer uses Klippel in it's design, most of the advantages have been realised, Maybe that is a reason I can get away without it, at least in my designs. In listening trials my speakers still beat any 5-10 times as expensive main speaker brands modell without question, after a few second of A-B test.
Any serious loudspeaker manufacturer uses a Klippel, as it is the only way to optimize good chassis. Have a look at their reference list.
By the way, Linn does not have a Klippel... which shows me they do not want to know what they did wrong on their products... If you only care for marketing, you don't need Klippel, that's a fact.
I wonder how many Chinese "devlopers" are trying to clone a Klippel, right now as I write this... most units sold seem to be in China. Good luck to our Chinese brothers! Is there something to interpret from the fact Russia does not have a single Klippel system?
Linn does not have a Klippel because Gilad Tiefenbrun would not pay for one, not because r&d did not ask for it.
Weeeeellllll...
As I said previously, there are many designs that get this wrong the most. Unfortunately, GP measurements aren't a one size fits all solution and you have to understand the effect the ground has on the "apparent" baffle. Case in point: measure a tower speaker laying on its side and then measure it upright. You'll get two different results through the BS region (right in the critical midrange area). Measuring it upright is the correct way but many measure it on its side because they assume it won't make a difference.
What I'm providing in the photo above is the on-axis response of (4) measurements:
1) Outdoor GP @ 2m with the speaker laying on its side (blue)
2) Outdoor GP @ 2m with the speaker standing upright, not tilted toward the mic (green)
3) Outdoor GP @ 2m with the speaker standing upright, tilted toward the mic (purple)
4) NFS @ 2m (black)
*Side note: The mid frequency looks different because of baffle step. I expect you understand this but for those who are watching from the sidelines, it's important to not this is a fault of ground plane measurements as well.
All 3 GP measurements match the NFS below 100Hz. The speaker on its side (blue) looks much different in the midrange because it's baffle width is effectively doubled so the baffle step profile has changed. The purple and green (speaking standing upright tilted/not tilted) track the NFS until about 800Hz or so where the exact aiming of the tweeter relative to the microphone matters, and in these cases, it wasn't perfect. That wasn't the purpose of the test, though, so I didn't really care. I just wanted to verify the bass extension and midrange region. As you can see, the NFS lines up in the bass nearly perfectly.
Lesson here? Measure as it would be used. In a tower speaker, measure it upright. You'd think it's common sense but I see many people get this wrong. But also take time to compare a clean* gated measurement so you know that you can trust your GP measurement.
*I'd recommend at least 5ms reflection free to make sure you have reliable data down to 200Hz and your multiples of that are sufficient to have good data at 1kHz. More ideally, you'd have 10ms or more.
Last edited:
Well, that is embarrassing. I thought the measurements were of the woofer but they are not. I’ll try to find if I have something more but it was more than 10 years ago.
Wow @Turbowatch2 that is cheap for a dyno run. In the California bay area it runs several times that. Ya lucky!
I can’t see why it would make any difference, there’s no reference plane in free space.
We test our big speakers upside down in a rented anechoic chamber equipped w Klippel (yes they have modules for that application as well).
We test our big speakers upside down in a rented anechoic chamber equipped w Klippel (yes they have modules for that application as well).
About the bafflestep, what about just a simulation + ground plane measurements (some get good results lifting the speaker actually from the floor) + regular measurements.
I have this so many times, comparing it to either anechoic measurements or Klippel measurements of similar speakers.
It always ends up what was expected.
Unless you create very crazy baffles, bafflestep simulations are pretty accurate and predictable.
I have this so many times, comparing it to either anechoic measurements or Klippel measurements of similar speakers.
It always ends up what was expected.
Unless you create very crazy baffles, bafflestep simulations are pretty accurate and predictable.
That's the default state on any forum.
In fact, that is the starting point before you even say one word.
Two different NFS users, Erin and amir, measuring the same product. I think it proves my case that even using such a refined equipment, the differences are about the same as with an insightful home DIYer using a mic and sound card could obtain... ref point ended up att 100Hz...
//
//
