The only thing that's clear from those measurements is that the design is atrocious. The driver is unsuitable for ported box loading, the qts is way too high. The box is way too small, probably too small even for a moderate qts driver. The MF implementation is botched to the point that it's not doing anything useful at all.
Look at the ported box sim I posted. There's no 18 db roll off (the tuning is low enough that it won't reach 18 db until single digit frequencies if at all), in fact the roll off is quite similar to the sealed box even though you can't see much below tuning.
The port and driver are a system, they work together, they do not antagonize each other until below tuning where they cancel to some degree due to phase. Just because the cone motion is not linear with frequency does not mean the ported box is flawed. Cone motion is not completely linear in sealed boxes either, it ramps up with decreasing frequency and then it levels out. And if it's a high qtc alignment it bumps (in excursion, phase and group delay) before leveling out.
There's so much valuable information to be gained from acoustic theory, simulations and even basic measurements. You can't take a single example measurement of a fatally flawed design and make sweeping statements about an entire genre of enclosures.
Last edited:
Thomas Aquinas the medieval theologian would be impressed with the subtle distinction you are making*.
Whatever the largely theoretical shortcomings of VC-based methods as you cogently outline them, some respected manufacturers think it is the route to take. They must think you get a better result using imperfect VC feedback - even using a non-temperature compensated resistor - than just letting the starched fabric spider, round rubber surround, and tubby box rule the cone motion.
OT: models are great, even with fudge-factors like "lossy inductance" in. I just think it wrong for a DIYer to plug numbers into paint-by-number sims they don't fully understand or any other tool.
Ben
*Is this the first occasion when a medieval theologian has been cited at DIYaudio?
Last edited:
Really? Again with the sims? The one sim he showed laid right on top of the measurement almost perfectly. I've shown you dozens of sims with similar accuracy compared to measurements. There are literally hundreds of examples of sims accurately matching measurements on this forum alone, thousand of examples on the internet.
I would think this indicates a large number of people intimately understand how simulators work, and the proven accuracy of simulations are a testament to their inherent value for education, design, etc.
Your constant remarks and consistently negative attitude towards simulators and the people that use them to good effect has never gained any traction here (this is a scientifically minded forum after all) and it never will except with other like minded people who don't care to learn how to use them.
At this point, after seeing so much evidence of the validity of sims, to deny their usefulness or the inability of others that have SHOWN you accurate sims to understand how they work is like denying the existence of gravity.
Simulators are based on theory that was mathematically proven decades ago, it's not guesswork. People that are proficient with simulators are well aware of what they do, how they work, and their limitations. Stop implying that everyone here is dumb as a bag of hammers or take some responsibility and show us what we don't understand and how it affects the accuracy of our sims.
The point is what you are doing with the VC current signal, e.g. which transfer function is used to inject it into the amp's feedback loop. Assuming a real-valued transfer, a plain scalar factor, three regions of operation might be distinguished, looking at the resulting output impedance Zo of the amp vs VC static impedance (Rdc, dominantly):
a) 0 > Zo > -Rdc : <insert fancy name here>. Assuming a Zo only slightly less (absolute value) than Rdc, the Rdc is almost fully cancelled and the voltage the amp "sees internally" is almost pure back-EMF, with little content of resisitive voltage drop acrross the effective "Rdc" (= Rdc - Zo). Back-EMF is another word for a voltage being proportional to cone velocity ("microphonic voltage").
The result is a true motional feedback (high level of driver-internal feedback), the amp controls the cone velocity directly by applying whatever force (via current) is needed for the Back-EMF to equal the input voltage, velocity ~ input.
The transfer function of sealed box is then one with heavy pole splitting, having two first order high-passes wide apart, and between those two points we have a -6dB/oct slope, therefore EQ is needed to get back to whatever target.
Problem : VC static impedance is neither constant nor purely resistive, and Back-EMF isn't linear. Any LF error signal (like induced dynamic DC-offset etc) of the cone beyond what the loop can control will settle without ringing but taking a lot of time (overdamped behavior). Chaotic and unstable with high excursion.
b) Zo = 0 : Standard Voltage drive. Someting in between both extremes and mostly a good compromise. Driver are developped to fit in this condition, after all. Moderate level of internal feedback.
c) Zo >> 0 : Current Drive. Back-EMF is ignored, and the amp controls the force applied to the cone, force ~ input. A force controlled system with no damping other than mechanical. No internal feedback of any sort.
Need again EQ to reach a proper SPL transfer function.
Not a good idea for woofers. Any error signal of the cone, be it external or internal (coming from distortion) will ring with the system resonance.
Can be chaotic and unstable with high excursion, but for a different reason: jump resonance.
Last edited:
Bolserst,
The port response is very weird indeed for a standard "normal size" duct.
I have measured (literally) tons of ported PA and home stereo cabinets, none of them exhibiting the "twin peaks" of the Sony SA-W2500 “subwoofer” port. Every "standard" BR all have a single Fb peak which varies from as little as 1/3 octave to as much as an octave or more, with a little narrow upper pipe resonance peak/dip as seen in the rather huge (and obviously un-equalizeable) 700 Hz peak.
At any rate, the output only would be considered "subwoofer" territory for top cabinets that roll off around 200 Hz, perhaps a good choice for the OP ;^).
Art
Last edited:
You misunderstood my post about Klipsch Heresy - I would lay my dirty old hands at that beauty anytime! My room is big enough an I do prefer professional/PA loudspeakers over those wimpy hi-fi speakers.
Here in Europe apartments in a old-fashion masonry style are preferred over gypsum drywalls, so no wall cavities.
Last edited:
I measured many examples of "twin peaks" port resonances and they were typical for a high Qts drivers.
The problem you have is not that DIYer uses sims that they don't understand, but it's that DIYer can have wrong goals regarding their expectation at the moment they decide to build.
DIYer wants to learn, and simming programs gives them a way to test a lot of thing by trial-error process without spending any money in real project. But of course, step by step, in this learning process, DIYer wants to try real-life stuff sometime too early.
Using sims tools as hornresp, or akabak it allow to understand seeing graph, that in some conditions, speaker can exceed Xmax (for basic thing). Or we can understand the need to make things phase coherent to sum right (seeing graph again with xsim for example). Or we can see high order slope ringing without understanding anything about what linkwitz explain on his website (seeing akabak step response graph - i talk for me). So it can be used to learn, before spending too much money in a project.
Using audacity, we can discover how ear-sensitive we are to different crossovers. Using some vst, we can learn how much sensitive we are to distortion, and at what freq (i generally like low range distortion making the false impression of a lot of low end, but nowdays some modern mastering already include those type of effect).
While coherent enough, the measured BR chart differs in a few ways from the textbook version. But for BR boxes, nobody is obliged to hew to the textbook. You can screw around with your choices to please yourself. This has been obvious from the early well-known example, the Altec A7 "Voice of the Theater", hundreds of years ago, that departed from a textbook ("sim") design to sound better in its natural theater (and home) settings.
Ben
Ben
Last edited:
Well, since I'm mentioned here I probably should respond.
You will get no argument from me that the iNuke and the 18's will outperform the Rythmic in sheer power and SPL, no doubt about it.
Matter of fact, I still may change my mind on all of this, I still have the time to do so...
However, the whole reason I looked into the Rythmic is the feedback loop being able to manipulate/control the driver. It's also the reason I would pay a premium if it performs as it claims. Now it's been thrown around that this may or may not be of any benefit, and the jury in me is still out on this aspect (unfortunately).
Incidentally, the Rythmik kit you sited is not the one I was speaking of... I was looking at getting (2) kits with the 1501 (500w) drivers and the HX400 (Hypex 400w rms) amps, which do not have the PEQ, as I own a 5-band per side, stereo parametric with L&HPF capability..
Those kits are $499 ea., but there is a discount when you order two, so the price when you order those is $948 shipped for both.
Also point taken on your size argument, if one has the available space behind the drivers for the infinite, then sure, why not? But this system is in my living room and I cannot cut a hole in either the floor, walls, nor ceiling, so for me this is not possible. (maybe someday in the basement HT)
Not that it matters much, but one final thing is that the Rythmik can be put into any enclosure type, they ask that you let them know and they will make changes to the amp boards accordingly. From the website:
*edit* Oh, and when I was speaking of the Rythmik being priced similar to non-servo kits, I was speaking of an 'all on sub' type of kit (self contained), and I still have not seen anything that would convince me otherwise. Just to clarify.
Cheers
-Steve
Last edited:
My statement concerning source vs room contributions to quality had nothing to do with MFB. I should have made that clear, sorry. It probably would have saved you a lot of typing. All I was saying was that you can hear a poorly performing subwoofer even in an untreated room. It is worth pursuing improvement in your subwoofer design (by whatever means) even if you may not be able to treat your room due to domestic or other reasons. If it helps you understand my posts better, MFB is not my go-to solution for every design...maybe only 1 in 10. I also try my best to avoid subtle implications.
Last edited:
All respected manufacturers using VC-based feedback use a second sensing coil, not a bridge arrangement.
The measurements I provided show the practical reasons why.
Thanks for the links, a few papers I hadn’t seen yet. I’ve learned a lot for Toole as well. I particularly like how he lets the data do the talking rather than trying to make the data fit a particular theory or notion.
Most if not all of the subwoofer-room modal response is non-minimum phase, so EQ cannot fix it. The REW manual actually has a fairly good write-up in it on how to use group delay to determine what parts of the room response are minimum phase and which aren’t to help identify when EQ might be of help.
Sealed boxes are minimum phase systems(at low frequencies) so overall transfer function can be manipulated with EQ. That was the topic of my experiments. But, you bring up an interesting question…do vented enclosures behave as minimum phase systems? A simplified lumped model would certainly says yes, and most well designed enclosures will likely be. But depending on the exact enclosure proportions and port location I can imagine it is not necessarily always the case. I will put that on my list of things to test before I die.
With a minimum phase system, the same flat response that produces the least distortion to pulse waveforms will produce the least distortion with music as well. Speaking of minimum phase, weltersys has recommending you look into using your DSP to go linear phase in the bottom octaves to fix up your impulse and square wave temporal response. I have yet to give that a try, but it’s on my short list. Linkwitz had a link to a decent summary/study on the topic, as well as some information on causes and audibility of group delay. Skip to the last 10 pages of the PDF to see real life examples of what linear phase correction can do.
Frontiers
http://www.linkwitzlab.com/Attributes_Of_Linear_Phase_Loudspeakers.pdf
>>>
Quote from the summary:
"It is clear, that designing loudspeakers using frequency-domain characteristics as the main (or only) criteria leads to stagnated, oversimplified, and ultimately inaccurate system…So, here I am. Struggling to come out of the “frequency-domain box” and into the new world of time/frequency/space-domain characteristics of contemporary loudspeakers."
Last edited:
Me too!
It wasn't until reading a post by Rythmik designer that the connection between the two patented circuits finally "clicked" for me.
http://www.diyaudio.com/forums/subw...arameters-electrical-means-3.html#post1589127
A great edition of the REW manual I have was written (or re-written) by a brilliant guy named "Bolser". Yes, a great tutorial. Thank you. (You are too modest.)
The Toole EQ argument about rooms looks at the total timed tonal compass sound package reaching human ears as the criterion (so dispersion index by frequency becomes a key player in shaping and mis-shaping the sound). Thus simple room response FR misses a lot and EQ to fix the FR misses it too.
B.
Right, those neat flat curves from hornresp may just not be the whole story, at least for Linkowitz.
A friend boasts that his Quad ESLs produces square waves. Not something I've ever been able to coax my ESL cells to do, yet. But even though you, Linkowitz, and weltersys can tweak speakers to produce accurate waves, I can't see how anything but MF can take the errors out of cone motion (such as resonances, non-linearities, etc.).
The rumble on the IPAL tone bursts are not a frequency problem. Granted, in some highly abstract sense, driver and box parameters which are not practically attainable (without feedback) could in theory do it. But I don't think it is helpful to continue to say "If a Martian could build a driver like that, it would sure enough sound like MF".
For example, I suppose an MF driver has a Qms that is unobtainably tiny since it has a much shallower resonance.
Ben
Last edited:
How many times do you need to be told that simulators don't control the design process? They calculate the results of the user inputs, and the design is completely up to the user, not the simulator.
The only simulator I know of that isn't 100 percent user driven from beginning to end is WinISD, that program has a few mathematical (textbook) alignment options to get you started, but from that point it is possible (and highly advisable) to tweak the sim according to the design goals.
You think the Altec A7 appeared hundreds of years ago? Really? A speaker named "Voice of the Theater" was produced hundreds of years before moving pictures were invented, not to mention the tech and parts to build the speaker?
This off axis dispersion issue is one (of many) categories your 1 meter square ESL panels fail at. Miserably. I can't imagine Toole would do anything but throw them in the garbage, they don't have any of the qualities he describes in well designed speakers.
Are you only looking at frequency response? I know that's usually the only thing that people post, but Hornresp has a lot of different graphs. I can assure you, Linkwitz isn't investigating any obscure factors that good simulators can't predict. Hornresp can show you about about phase, group delay, timing (ringing), dispersion.
If you would actually LOOK at a simulator you would find that they are not a dumb interface that chooses your design for you and shows you only frequency response.
Resonances are not errors, non-linearities are not an issue in properly designed and sized systems.
Last night I presented two properly sized system examples that achieve the same goals that MF pursues. These systems do not use Martian theoretical technology, it's currently available stuff and this stuff is actually really affordable too.
When you look at this from a performance goal perspective it should be clear that the non MF system design philosophy I've been talking about achieves lower distortion than your cheap 15 inch driver with an accelerometer taped to it could ever achieve. Just because a system doesn't have active cone correction doesn't mean it can't have vanishingly low distortion levels.
Hi bentoronto,
Just a quick addition to my response to the links provided in Post #191, and Posts #202/203:
Post #202: "Pity you didn't have a chance to read the article. He demonstrates the opposite point of view, namely that basic frequency ("amplitude") correction is not the right (or not the whole) way to achieve your perceptual house curve. He is hostile to relying on room EQ as the core approach to fixing rooms."
The quote I posted in #199 was from Dr. Toole, and clearly indicates his position on the usefullness of parametric equalization for low frequencies. Additionally, in the AES paper (1st link in #191) he addresses when and how equalization is correctly applied, and his general dislike of "generic “room curves,” or more specific “house curves.”"
After reading through the linked materials I can see what you are referring to when you say: "He is hostile to relying on room EQ as the core approach to fixing rooms." I misinterpreted this as asigning a general "hostility" towards equalization to Dr. Toole, which you didn't do, and which is obviously not the case.
Thanks for the links.
Regards,
Just a quick addition to my response to the links provided in Post #191, and Posts #202/203:
Post #202: "Pity you didn't have a chance to read the article. He demonstrates the opposite point of view, namely that basic frequency ("amplitude") correction is not the right (or not the whole) way to achieve your perceptual house curve. He is hostile to relying on room EQ as the core approach to fixing rooms."
The quote I posted in #199 was from Dr. Toole, and clearly indicates his position on the usefullness of parametric equalization for low frequencies. Additionally, in the AES paper (1st link in #191) he addresses when and how equalization is correctly applied, and his general dislike of "generic “room curves,” or more specific “house curves.”"
After reading through the linked materials I can see what you are referring to when you say: "He is hostile to relying on room EQ as the core approach to fixing rooms." I misinterpreted this as asigning a general "hostility" towards equalization to Dr. Toole, which you didn't do, and which is obviously not the case.
Thanks for the links.
Regards,
[
You might want to quantify “vanishingly low” though.
This is likely the heart of your intellectual struggle. Like the quote from Bohdan Raczynski where he says he was locked in a frequency-domain mind set, ignoring the possibility of time-domain improvements by using DSP to adjust phase independent from response, I think perhaps you are locked in a MFB mind set based on your experience with SOA drivers you experimented with 50 years ago. You are ignoring the improvement in woofer design( electrical , magnetic, & mechanical) and modeling tools to optimize the woofer system. If you had available to you 50 years ago a SOA woofer from today, you would not have been nearly as impressed with MFB improvements as you were. You may have even come to the conclusion that there were better ways to achieve equal improvement in performance. All that being said, I still agree that MFB is a very intellectually satisfying design approach.
+1
You might want to quantify “vanishingly low” though.
Last edited:
I don't have a mic right now so I can't provide distortion measurements but I did try to quantify this by stating that my normal listening level with my Klipsch Heresy bedroom system is at power levels well below 0.01 watts. If you can guess the likely distortion levels in that situation you will have an idea of what I'm talking about. I don't want to guess as the margin for error in a guess like that is pretty high.