It is not correct to compare single driver disto meas with full system measurements especially when meas K3 !!! Too many hidden variables ...
Yes, that can be an issue. FWIW, here is a before & after of HD2 & HD3 for a system with a Seas aluminium / magnesium dome tweeter; one 'stock', the other with a parallel notch on its ~26KHz concentrated breakup peak. There were no other changes made to the filter, passband frequency response or anywhere else. Behaviour is as expected (& as indicated in the Purifi pdf & other instances).
Attachments
Last edited:
Guys,
You have to expand this! 26kHz peak? You have one measurement of raw driver, one with filtered response in a finished speaker, which is a magnitude better with the same level of 90db. We're talking about the distortion in bass response, measured, not simulated!
You have to expand this! 26kHz peak? You have one measurement of raw driver, one with filtered response in a finished speaker, which is a magnitude better with the same level of 90db. We're talking about the distortion in bass response, measured, not simulated!
My SEAS magnesium cone drivers have an LC network directly across the speaker terminals as per their recommendation.
thats the ‘wrong’ type of notch if we want to reduce distortion generated at sub multiples of the peak frequency.
Sure. The system HD (as in the HD of a complete speaker) is affected to some extent by the entire filter & particularly where you're crossing. For instance, if we have a driver with a 6KHz breakup peak (it's just an example), you'll see HD3 amplification at 2KHz. But if you cross below 2KHz, you'll be partly avoiding that 2KHz peak, so your system HD will look different -how much so depending on the other drivers you're crossing to! It's one way of getting around the issue: the difficulty sometimes is being able to do so, because you can get into a situation where you end up robbing Peter to pay Paul. So in broad terms you have to be a bit careful when comparing a 'raw' driver's HD response to the system HD of a full speaker as there can be a number of variables at work.
The measurements I posted above were the before / after HD2 and HD3 plots of a complete speaker (Tafal). The only difference was that in the second set of plots, the tweeter's 26KHz breakup mode was notched out with a parallel LC circuit inserted in series with the tweeter. No other changes were made, and the frequency, impedance etc. up to ~20KHz show very little difference. It makes quite a difference to the HD performance though, & is a good example both of the benefits of stamping on a major stopband (loose use of term for brevity) breakup peak in this way, and also that it isn't just an issue for bass, midbass & midrange drivers: there can be benefits to treating tweeters in this way as well, even though the breakup peak itself is well above the HF limit of human hearing.
Last edited:
I still do not understand why a filtered system would be wrong to compare with a raw driver distortion. If this shows a lower distortion, it's all good. This is after all what we listen to. Btw, the files I attached is a 12db electrical, acoustical cross at 2kHz, should be the worst possible according to the theories in this thread...
Last edited:
It's a broad principle rather than a specific: if you compare a raw driver to a complete system, you're changing multiple variables rather than looking at the effects of a single feature -in this case that of adding a parallel LC in series with the driver tuned to its main breakup peak or, stretching a point, that and the rest of the low pass. The system HD, depending on the particular design, may be affected by the HD performance of other units through the region of interest. It's not a given, but it is possible in some cases, so at least for bass / midbass / midrange units, it's strictly speaking preferable / best practice when looking at the effects of one particular approach to ensure it's the only unit operating. A bit pedantic, but it can sometimes help avoid problems, since that way we always know for certain exactly what it is we're looking at.
This isn't to deny the really important part of course, which is that the system HD is indeed lower.
This isn't to deny the really important part of course, which is that the system HD is indeed lower.
Thanks for the elaborate answer! I totally agree when it comes to find the reasons/solutions for problems, looking at it in an isolated situation is the best. In this case, the distortion is low enough IMO! (Except for the distortion fairies on this forum;o) More components are not necessary according kiss philosophy. What makes a successful engineer is how to fix a problem in a system with as little effort as possible! What makes a true genius, is to find several ways to solve the issue and be open for more solutions...
One query, if I may? The images you attach appear to show the Hobby Hi-Fi VisaMon 170 project. I don't have access to Hobby HiFi, so I don't know the exact details of the filter, but the image you posted above doesn't show the 12dB electrical topology you mention? The circuit diagram shows the midbass has a 3rd order electrical low-pass, with a shunt LCR and RC (plus large series cap used to get some more LF out of an acoustically undersized box). The tweeter has a 4th order electrical with fixed L-pad. Have I missed something? FWIW though, doing a little digging, I found the kit listed on Lautsprechershop, which also noted the crossover frequency of roughly 2.1KHz. Assuming the low-pass creates an LR4 acoustical slope or thereabouts (I suspect this is a fairly safe assumption given the drivers, box size & filter topology) that inherently helps with the HD3 amplification at roughly that frequency (and to a lesser extent the HD5) because the driver is already down -6dB. Not 'ideal', but does a reasonable job.
Visamon 170 it is! And you are right about electrical third order on bass driver! I mixed it up, with what I believe was made by Klang+Ton, which has an electrical second order filter. However, I'm not sure...
Last edited:
Nice to see some variations on the theme TBH! 🙂
Without access to the full design details we're speculating somewhat; however, I located the kit for this version on Lautsprecher, where it notes a crossover frequency of 1.8KHz, so it's actually crossing below the 2KHz peak in HD3. Quite a bold choice with that tweeter, but I digress... 😉 Anyway, same thing applies: assuming 2nd order acoustic slopes (which seem unlikely given that the speaker appears to have a flush front baffle and no obvious compensation for the acoustic-centre offsets) then it will be at least -6dB at 2KHz, so like the KE25SC version, I'd expect to see some reduction in that HD3 spike, which appears to be the case.
Without access to the full design details we're speculating somewhat; however, I located the kit for this version on Lautsprecher, where it notes a crossover frequency of 1.8KHz, so it's actually crossing below the 2KHz peak in HD3. Quite a bold choice with that tweeter, but I digress... 😉 Anyway, same thing applies: assuming 2nd order acoustic slopes (which seem unlikely given that the speaker appears to have a flush front baffle and no obvious compensation for the acoustic-centre offsets) then it will be at least -6dB at 2KHz, so like the KE25SC version, I'd expect to see some reduction in that HD3 spike, which appears to be the case.
Last edited: