Hi Paul,
I don't care for hijacking, its fine for me. 🙂
Classic music is often well recorded and highly dynamic with few compression on the recording, so you need a speaker that delivers. You've some very nice equipment on stock, where I see different options for implementation to achieve that. Maybe someone other has additional good ideas or experience that I don't see or have:
1) If a longer journey is planned, and you do not hesitate to build prototypes for experiments, why not try the Purifi with the T25Bs just as it is? If you care a bit for baffle geometry and driver placing, and be smart with the filters, there are nice examples to get very good power/in room response and preference rating even with classic configurations w/o waveguide (but I admit that the T25B radiates the sound very widely in comparison to other tweeters...):
https://www.erinsaudiocorner.com/loudspeakers/ascend_sierra_1_v2/
https://karl-heinz-fink.de/the-measurements-checking-out-the-epos-14n
Perhaps you will be fully satisfied with the result once you listen to them. When you look especially at the Sierra 1 V2 measurements you may have to sacrify a little bit on-axis linearity, but that gets more and more insignificant with increasing listening distance > Critical Distance where the indirect sound/reflections dominate.
2) Change the tweeter to one with a Waveguide, or one where you do not have reservations to add one; e.g. a SB26ADC in a Somasonus WG. Or Seas DXT, there is already a well documented open source project available with very good measurements:
https://www.audiosciencereview.com/...r-directiva-open-source-speaker-review.27094/
Also here: If a longer journey is planned, and you do not hesitate to build prototypes for experiments, why not buy a quite affordable pair of Seas DXTs and compare them with the T25B? When you do not like it you can easily sell the DXT here with managable loss.
3) the already discussed - and preferred by you at the moment, as it seems - option with a 4" mid:
As argued, larger mids are not recommended because the reason/effect for/by adding it is continuing and matching the very wide dispersion of the tweeter. The one that gives the highest dynamic headroom I could find is the Kartesian Mid120_vHE that I use. At least in rough simulation, it will give no significant headroom increase to the overall system in comparison to using the Purifi as a midwoofer. Maybe the flat membrane will give you a slightly plus in openess and neutrality by avoiding funnel refections in the mids on the deep woofer cone and the low loss fabric surround slightly more resolution / detail.
With the Flex8 you can handle the breakup resonances starting from 4kHz very well.
Also here, again: If a longer journey is planned, and you do not hesitate to build prototypes for experiments, you can buy a cheaper mid for comparison to the 2-Way solution and you can even switch between the configurations when you arrange the drivers W-T-M.
My personal recommendation for a value mid is SB12MNRX2-25-4 which is on 2nd place in my "dynamic headroom" shortlist (I simulated even more 4" drivers in the past than on the diagram a few post above) and has a flawless frequency response (and even a shorting ring which is not mentioned in the feature list as usual). When you like it, you can think about "upgrading" to the Kartesian (it fits even roughly the same rebate with few rework) and sell the SB with managable loss, or even keep it when the headroom and resolution is sufficient for you.
I cannot recommend the Purify 4" after further looking at the data and as predicted by @b_force , because efficiency is low and it needs double the power for same output than the SB for a fraction the costs.
Found a nice pic of a prototype speaker from Telegrapher for some inspiration:
With the 12" Subwoofers avaiIable, I would strongly recommend to put the Purify just in a closed box. An additional vent is not needed for headroom and will complicate the xover to the subs, and increase the group delay which gets already borderline with very low xover. When you want to have the option for using the satellites standalone, implement a port for headroom in that case, but make it closable.
Using the Purifi as a kickbass in a very limited range >> 80 Hz to << 400 Hz is a bit "throwing pearls to swine": The Purifies excel with large bandwith, you pay for the capability to move air in the lows and at the same time reproducing undistorted mids. But in that 4-Way configuration they don't have to do either one or the other. Well they will perform perfectly, but are quite overqualified for the job...
Best regards
Peter
I don't care for hijacking, its fine for me. 🙂
Classic music is often well recorded and highly dynamic with few compression on the recording, so you need a speaker that delivers. You've some very nice equipment on stock, where I see different options for implementation to achieve that. Maybe someone other has additional good ideas or experience that I don't see or have:
1) If a longer journey is planned, and you do not hesitate to build prototypes for experiments, why not try the Purifi with the T25Bs just as it is? If you care a bit for baffle geometry and driver placing, and be smart with the filters, there are nice examples to get very good power/in room response and preference rating even with classic configurations w/o waveguide (but I admit that the T25B radiates the sound very widely in comparison to other tweeters...):
https://www.erinsaudiocorner.com/loudspeakers/ascend_sierra_1_v2/
https://karl-heinz-fink.de/the-measurements-checking-out-the-epos-14n
Perhaps you will be fully satisfied with the result once you listen to them. When you look especially at the Sierra 1 V2 measurements you may have to sacrify a little bit on-axis linearity, but that gets more and more insignificant with increasing listening distance > Critical Distance where the indirect sound/reflections dominate.
2) Change the tweeter to one with a Waveguide, or one where you do not have reservations to add one; e.g. a SB26ADC in a Somasonus WG. Or Seas DXT, there is already a well documented open source project available with very good measurements:
https://www.audiosciencereview.com/...r-directiva-open-source-speaker-review.27094/
Also here: If a longer journey is planned, and you do not hesitate to build prototypes for experiments, why not buy a quite affordable pair of Seas DXTs and compare them with the T25B? When you do not like it you can easily sell the DXT here with managable loss.
3) the already discussed - and preferred by you at the moment, as it seems - option with a 4" mid:
As argued, larger mids are not recommended because the reason/effect for/by adding it is continuing and matching the very wide dispersion of the tweeter. The one that gives the highest dynamic headroom I could find is the Kartesian Mid120_vHE that I use. At least in rough simulation, it will give no significant headroom increase to the overall system in comparison to using the Purifi as a midwoofer. Maybe the flat membrane will give you a slightly plus in openess and neutrality by avoiding funnel refections in the mids on the deep woofer cone and the low loss fabric surround slightly more resolution / detail.
With the Flex8 you can handle the breakup resonances starting from 4kHz very well.
Also here, again: If a longer journey is planned, and you do not hesitate to build prototypes for experiments, you can buy a cheaper mid for comparison to the 2-Way solution and you can even switch between the configurations when you arrange the drivers W-T-M.
My personal recommendation for a value mid is SB12MNRX2-25-4 which is on 2nd place in my "dynamic headroom" shortlist (I simulated even more 4" drivers in the past than on the diagram a few post above) and has a flawless frequency response (and even a shorting ring which is not mentioned in the feature list as usual). When you like it, you can think about "upgrading" to the Kartesian (it fits even roughly the same rebate with few rework) and sell the SB with managable loss, or even keep it when the headroom and resolution is sufficient for you.
I cannot recommend the Purify 4" after further looking at the data and as predicted by @b_force , because efficiency is low and it needs double the power for same output than the SB for a fraction the costs.
Found a nice pic of a prototype speaker from Telegrapher for some inspiration:
With the 12" Subwoofers avaiIable, I would strongly recommend to put the Purify just in a closed box. An additional vent is not needed for headroom and will complicate the xover to the subs, and increase the group delay which gets already borderline with very low xover. When you want to have the option for using the satellites standalone, implement a port for headroom in that case, but make it closable.
Using the Purifi as a kickbass in a very limited range >> 80 Hz to << 400 Hz is a bit "throwing pearls to swine": The Purifies excel with large bandwith, you pay for the capability to move air in the lows and at the same time reproducing undistorted mids. But in that 4-Way configuration they don't have to do either one or the other. Well they will perform perfectly, but are quite overqualified for the job...
Best regards
Peter
Last edited:
When there are subs, especially multiple, I don't see why someone would go lower than 100Hz anyway.
Or you can even go higher when the subs are symmetrically placed next to the tops, but that all becomes a bit room and situation specific.
I also feel the same about the over qualification. Especially because our ears are also less sensitive in that range already and IMD is low because there isn't much cone excursion.
An 8 or 10 inch already has very low distortion in that range.
Personally, I would go for 2-way with a good waveguide.
But I do see and understand why a 3 way can be appealing for the wider directivity.
Last edited:
Yes, with the Purify you can build a SOTA 2-way + sub system with constant directivity characteristics that takes advantage of the woofer capabilities. I can fully understand in hesitating to slaughter the T25B for applying a waveguide, but then the consequent alternative is to use another tweeter. Beside the mentioned Somasonus solutions and the DXT, here are some nice measurements with different tweeters in commercially available WGs that also work well: https://heissmann-acoustics.de/en/treibertests/
(BTW: Another option with just the small Bliesma may be a concept like favored by ME Geithain: https://www.me-geithain.de/en/rl-904.html
They achieve some directivity control by diffraction on the tweeter little baffle edges, but this will need some loops to find the correct dimensions and may be a bit too experimental and the result too uncertain - but the idea just came in my mind. Don't take it too serious....)
When concepting the "Schnuckelchen" I also thought about using a pair of 10"-12" subwoofers right from the start in addition (for multiple reasons not related to performance I discarded the subs), and likely I would have decided to take this little puncher instead of the WO24P as kickbass supporting the T25A/Mid120 unit in the fundamentals: https://www.eighteensound.it/en/products/lf-driver/8-0/8/8NW650
Selling the Purifis will release right the money budget free to buy the Mid120_vHEs and also a pair of 8NW650 or similar; this will still result in a SOTA setup with more balanced performance budgets for my taste...
(BTW: If you find the paper basket rims of PA drivers not so pretty - well as I respect the unique performance of the Purifis, I would never ever use one because the surround shape is the ugliest thing I have ever seen on a driver...)
Best regards
Peter
(BTW: Another option with just the small Bliesma may be a concept like favored by ME Geithain: https://www.me-geithain.de/en/rl-904.html
They achieve some directivity control by diffraction on the tweeter little baffle edges, but this will need some loops to find the correct dimensions and may be a bit too experimental and the result too uncertain - but the idea just came in my mind. Don't take it too serious....)
When concepting the "Schnuckelchen" I also thought about using a pair of 10"-12" subwoofers right from the start in addition (for multiple reasons not related to performance I discarded the subs), and likely I would have decided to take this little puncher instead of the WO24P as kickbass supporting the T25A/Mid120 unit in the fundamentals: https://www.eighteensound.it/en/products/lf-driver/8-0/8/8NW650
Selling the Purifis will release right the money budget free to buy the Mid120_vHEs and also a pair of 8NW650 or similar; this will still result in a SOTA setup with more balanced performance budgets for my taste...
(BTW: If you find the paper basket rims of PA drivers not so pretty - well as I respect the unique performance of the Purifis, I would never ever use one because the surround shape is the ugliest thing I have ever seen on a driver...)
Best regards
Peter
🤣😆😀
You're not the only one 😀
Btw, for rims one can 3D print something nice an smooth 🙂
Or just use rear mount.
Gentlemen. Let’s keep this on topic. The mods have done cleanup on this thread.
Power is power, and in theory it doesn't matter if the axis of rotation differs a little like this.
The ideal of using the acoustic centres (not that there is a definable axis, hence the need to specify the baffle front), seems to some degree to suit the sensibilities of some that propose it.. except that it's not easy to argue against without an explanation like this, since it is the most logical place for it. It might be said that if you did use the source centre, it would be easier to visualise and understand the output, but it's clearly not that simple.
The individual response measurements are not affected by the axis location, except that they may be at slightly different angles and distances.
@Mods: Thanks for sweeping out the troll posts!
@b_force: Yep, the Purifi surround reminds me of HR Giger designs, dropping off alien slime in my nightmares.... 😆😛🤣
Lazyness! I build that turntable ~17 years ago with few knowledge and a big roller bearing a friend gave to me. Results are not perfect and not 100% comparable, but it is there and till now I never felt the need to improve because other things are more challeging, also the setup is very compact and can be easily stored. And surprisingly sturdy enough that i can put a ~40kg speaker on it.
With increasing adjustment angle, you get more angle error and distance error = level error. The wider the distance to the microphone gets, the less error you have.
As mentioned by @AllenB you always will have dicussions about the correct acoustic center you turn around during measurements. With the single driver in the shown sketch it seems easy, also still with a small 2-Way, but when you want to make vertical measurements of a big 3-way speaker it gets obvious: When you turn around the baffle midpoint between tweeter and mid, and the mic is 1-2m away, the woofer comes significant nearer and gets louder in the one direction, and in the other direction it gets significant farer away and quieter. But when summing up the power response, most of the effect is then averaged out.
It starts with a baffle like the "Schnuckelchen" have, even on a "ideal" turntable: do I turn around the middle axis of the eclosure, or around the middle axis of the tweeter & mid for the horizontal measurements? In both cases I will get a distance error = level error, in one case of the woofer + the asymmetric edge diffractions, in the other case of the mid&tweeter...
For the example with the larger 3-way, with on-axis mic position between tweeter and mid, and a mic distance of 2m, you get some cm wider distance to the woofer that results in phase and level errors that I calculate and compensate then in Vituixcad before simulation. In specific cases I do simulations in Vituixcad diffraction tool, changing between the mic positions of my measurement setup in comparison to a infinite mic position to see the differences and their relevance. So there are a lot of things affecting the measurements which have to be respected in general, so that my non-ideal turntable causes only a fraction of that.
Aware of that general issues and the additional restrictions of my turntable, often I only take measurements +/- 60° hor and +/- 40° ver as input for filter development in simulation, which is then ca. +/- 70° and ca. +/- 50° in reality, and with a very slightly higher weighting of the on-axis response. I can put the turntable to ~2m height in my garden, and usually measure compact 2-Ways in 1m distance, larger 3-ways in 2m distance and e.g. measurements for "Schnuckelchen" were taken in 1,5 m distance.
With a little experience and all that general and specific deviations in mind, this is totally sufficient for me as input for filter development. The last percents are always hearing, voicing and personal taste to the sound.
Best regards
Peter
@b_force: Yep, the Purifi surround reminds me of HR Giger designs, dropping off alien slime in my nightmares.... 😆😛🤣
Lazyness! I build that turntable ~17 years ago with few knowledge and a big roller bearing a friend gave to me. Results are not perfect and not 100% comparable, but it is there and till now I never felt the need to improve because other things are more challeging, also the setup is very compact and can be easily stored. And surprisingly sturdy enough that i can put a ~40kg speaker on it.
With increasing adjustment angle, you get more angle error and distance error = level error. The wider the distance to the microphone gets, the less error you have.
As mentioned by @AllenB you always will have dicussions about the correct acoustic center you turn around during measurements. With the single driver in the shown sketch it seems easy, also still with a small 2-Way, but when you want to make vertical measurements of a big 3-way speaker it gets obvious: When you turn around the baffle midpoint between tweeter and mid, and the mic is 1-2m away, the woofer comes significant nearer and gets louder in the one direction, and in the other direction it gets significant farer away and quieter. But when summing up the power response, most of the effect is then averaged out.
It starts with a baffle like the "Schnuckelchen" have, even on a "ideal" turntable: do I turn around the middle axis of the eclosure, or around the middle axis of the tweeter & mid for the horizontal measurements? In both cases I will get a distance error = level error, in one case of the woofer + the asymmetric edge diffractions, in the other case of the mid&tweeter...
For the example with the larger 3-way, with on-axis mic position between tweeter and mid, and a mic distance of 2m, you get some cm wider distance to the woofer that results in phase and level errors that I calculate and compensate then in Vituixcad before simulation. In specific cases I do simulations in Vituixcad diffraction tool, changing between the mic positions of my measurement setup in comparison to a infinite mic position to see the differences and their relevance. So there are a lot of things affecting the measurements which have to be respected in general, so that my non-ideal turntable causes only a fraction of that.
Aware of that general issues and the additional restrictions of my turntable, often I only take measurements +/- 60° hor and +/- 40° ver as input for filter development in simulation, which is then ca. +/- 70° and ca. +/- 50° in reality, and with a very slightly higher weighting of the on-axis response. I can put the turntable to ~2m height in my garden, and usually measure compact 2-Ways in 1m distance, larger 3-ways in 2m distance and e.g. measurements for "Schnuckelchen" were taken in 1,5 m distance.
With a little experience and all that general and specific deviations in mind, this is totally sufficient for me as input for filter development. The last percents are always hearing, voicing and personal taste to the sound.
Best regards
Peter
Last edited:
The requirements were
- Screen and speakers left/right next to it should be arranged as the current system.
- The arrangement must be capable to be twisted 90° and aligned to the sofa within a few seconds
Well, within around 1,5 years now, we had twisted the sideboard 90° maybe only 2-3 times. Usually we only each grab an armchair and sit at the position in which the photo below was taken, so the second requirement got a bit obsolete.
Placing the speakers on the sideboard is also not optimal, as it introduces a slighly boomy signature to the fundamentals. Not really disturbing resonances, but... (I put a thick rubber granulate mat between the wooden top plate and cabinet body). And the screen nearby the speakers with their very wide directivity introduces slighly coloration to the mid-highs due to reflections.
It is not really bad or even annoying, but when you put the speakers seperately, free-standing e.g. on the stands borrowed from another small 2-way I own, you obviously hear that some potential is left.
So I was rethinking that setup during the last two month by rebuilding the sideboard to half width, and getting back to floorstanders again (also improving the dynamic range and maybe precision in the low frequencies a bit, by the use of a larger cabinet and two woofers):
So I have something new in my mind where I will reuse the mid + tweeter, passive network and FA503 but replace the WO24P with two fast, low-mass Scan Speak 22W/8534G00 with extra magnets in a new cabin; Project "FastTrack":
Last weekend I discussed that proposal with my wife - but she got upset and had then "forbidden" me to "slaughter the cute Schnuckelchens" because she likes the sound and loves the look, and wants to keep the speakers as they are. But she agreed to reduce the sideboard's width and place them seperately on stands - so it came out exactly the other way around where I expected reservations from her... 😆 (and the project shown above got obsolete)
One option I'm thinking about now, is to use black anodized aluminum construction profiles to build the stands from:
Another option could be to ask a local metalworker company to weld some similar cube from steel tubes, which then can be painted (my wife preferres white color) and be filled with sand.
Or to ask a carpentry, if they can saw me massive wooden blocks from birch or maple of the needed size.
Still not decided and unsure, maybe someone has another good idea?
_____________________________________________________________________________________________________________________________________________
When I planned the speakers, I sized the cabinet volume to 24l according to the TSPs from the SB Acoustics datasheet, giving a nice Qtc of ~0,71-0,72 in CB.
But after measurement when the drivers arrived, I got significantly higher Qts than given by the manufacturer. Well, the CNC wood cut was already ordered, and I also had hesitated to increase the cabinet size much in general, so I had to live with a Qtc >0,8 within the final design, knowing that I can equalize with the DSP to my needs easily thanks by Mr. Linkwitz...
Afterwards I did some research then on the TSP deviations, asked around in a german diy forum and searched the internet, and came out with following results I want to share for your interest:
It is noteworthy, that the commonly used measurement of cone diameter is taken from the mid of the surround, but SB takes only the pure cone diameter which leads to lower values for Sd. Some just took the value over without own measurement.
And you can see that 6 datasets match pretty good to each other, but there are two "outliers", one of them published by the manufacturer - interestingly, it promises the "better" parameters than the ones you might get...
After it was clear last weekend that the Schnuckelchen will stay, I decided to order extra magnets to push BL and decrease Qes, hoping for a Qtc improvement:
It worked out nicely as hoped for:
So yesterday I ended up with the alignment Qtc = ~0,71 that I initially targeted for. 🙂
To keep the original frequency response I did two TSP based simulations, substracted both frequency response results and implemented a correction filter to the DSP:
I did no 1:1 comparison, but I think the sound got a little more controlled and precise in the fundamentals - but I don't want to exclude the expactancy effect that it just feels better when some potential was not left, and the alignment is working now in the optimized operating point.
Best regards,
Peter
Last edited:
SB Acoustics has a tendency to measure TS parameters at slightly higher voltages.
I have no seen and measured quite some different drivers and quite a few seem to have the same issue that the Qts as well as the Fs are both higher. (I mean the measured parameters compared to datasheets)
I have also done quite some experiments with bucking magnets.
With a well optimized and modern motor design, 10-15% range in change of Qt is basically what you can expect.
With older and more simple ones, it's around 20% max.
Curious, but where did you order those magnets? 🙂
It was also the case with SB15NBAC35-4 some time ago, both values higher. Tested also increased and lower exication levels with them, but was not able to match manufacturer data.
Magnet was ordered here (german online shop):
https://magnetkontor.de/de/ferrit-y35-ringmagnet-o115-50-x-20-mm.html
With the same magnet I achieved +9% BL for a 26W/8534G00
+5% BL for the SB15NBAC35-4 with this magnet applied:
https://www.magnetladen.de/ringmagnet-90x36x20-mm-y35/
and +7% BL for the SB15NBAC35-4 with this expensive N42 one:
https://www.ebay.de/itm/29127909593...VRneh+kDxjVAsjyXDpQ9JFv9JK|tkp:Bk9SR7KfsfyOYw
Magnet was ordered here (german online shop):
https://magnetkontor.de/de/ferrit-y35-ringmagnet-o115-50-x-20-mm.html
With the same magnet I achieved +9% BL for a 26W/8534G00
+5% BL for the SB15NBAC35-4 with this magnet applied:
https://www.magnetladen.de/ringmagnet-90x36x20-mm-y35/
and +7% BL for the SB15NBAC35-4 with this expensive N42 one:
https://www.ebay.de/itm/29127909593...VRneh+kDxjVAsjyXDpQ9JFv9JK|tkp:Bk9SR7KfsfyOYw
The lower these values is a reasonable indicator how well optimized the motor is.
Although it can depend a bit on its size.
Sometimes it seems that either a bigger or smaller one will do better.
I still want to play a bit with neodymium cilinders, but they are more difficult to mount.
Keep in mind that with a well designed motor, with a very symmetrical BL curve, you could push it asymmetrical sometimes.
Although it can depend a bit on its size.
Sometimes it seems that either a bigger or smaller one will do better.
I still want to play a bit with neodymium cilinders, but they are more difficult to mount.
Keep in mind that with a well designed motor, with a very symmetrical BL curve, you could push it asymmetrical sometimes.
Yes, e.g. there is some potential for improving the SB15NBAC35-4, but you pay a lot for the last 2% with the N42 magnet. Value is much better with the Y35 for 6.- €, the driver seems to be maxed out with it (at least the backplate steel saturation).
I assume this is a good example for a nicely matching geometry of the bucking magnet, adding it's magnetic field to the existing speaker main magnet at roughly the same position:
(Source: https://karl-heinz-fink.de/and-now-the-final-one )
When you would put e.g. a smaller diameter cylindric magnet in the middle of the backplate, that may also increase BL the same or even more for small signals, but there is high risk to disturbe the shape of the original magnetic flow and you might increase asymmetry of Bl(x). Taking the original geometry into account and just continuing it lowers that risk, and hopefully only increases BL and maybe slightly improves Le(i).
I think there is a reason why KHF did not just take a disk magnet instead of a ring if there would have been room for it... 😉
Neodymium rings with large, matching diameter are hardly available and expensive; with my experiences so far I think you get the most out of it with quality ferrite magnets for low invest....
Best regards
Peter
EDIT: Just some addition to your observation "Sometimes it seems that either a bigger or smaller one will do better":
When you look at the example driver's crosscut above, there might be designs with thicker backplates and smaller diameter pole pieces - so in the middle at the pole piece there is good saturation originally, and some small diameter bucking magnet near that position will not do much. With increasing diameter (with even increasing crosscut) the saturation decreases, and a large diameter bucking magnet will do it's work fine there.
The other way around: A thin backplate that is already well saturated will not be that sensitive to a large diameter bucking magnet. When the driver provides then a large diameter pole piece, the BL will increase the "smaller" the magnet gets, meaning it pushes the saturation directly within the pole piece. But this configuration may lead to the mentioned additional asymmetries of BL(x)....
I assume this is a good example for a nicely matching geometry of the bucking magnet, adding it's magnetic field to the existing speaker main magnet at roughly the same position:
(Source: https://karl-heinz-fink.de/and-now-the-final-one )
When you would put e.g. a smaller diameter cylindric magnet in the middle of the backplate, that may also increase BL the same or even more for small signals, but there is high risk to disturbe the shape of the original magnetic flow and you might increase asymmetry of Bl(x). Taking the original geometry into account and just continuing it lowers that risk, and hopefully only increases BL and maybe slightly improves Le(i).
I think there is a reason why KHF did not just take a disk magnet instead of a ring if there would have been room for it... 😉
Neodymium rings with large, matching diameter are hardly available and expensive; with my experiences so far I think you get the most out of it with quality ferrite magnets for low invest....
Best regards
Peter
EDIT: Just some addition to your observation "Sometimes it seems that either a bigger or smaller one will do better":
When you look at the example driver's crosscut above, there might be designs with thicker backplates and smaller diameter pole pieces - so in the middle at the pole piece there is good saturation originally, and some small diameter bucking magnet near that position will not do much. With increasing diameter (with even increasing crosscut) the saturation decreases, and a large diameter bucking magnet will do it's work fine there.
The other way around: A thin backplate that is already well saturated will not be that sensitive to a large diameter bucking magnet. When the driver provides then a large diameter pole piece, the BL will increase the "smaller" the magnet gets, meaning it pushes the saturation directly within the pole piece. But this configuration may lead to the mentioned additional asymmetries of BL(x)....
Last edited:
and added a passive network to the midrange and tweeter:
Series resistors for local current feedback (there is much headroom in voltage swing of the FA503 available) and serial notches for the main diaphragm resonances. This should help to decrease the already low current induced distortions and thermal modulation even more.
It is fantastic builds you show with lots of good info!
Would you explain deeper how you design the passive network? I have pretty good understanding of the DSP side, not so much on passive. I also read the Article Purify have posted on their site. But it primarily show the effects.
I would like to test it on my Fusion but don’t have an idea how to progress.
Hi, thanks for your reply!
Value of the resistor was selected under the assumption, that for much current feedback it has to be as high as possible.
As I have limited rail voltage (53V RMS = 350W @ 8 Ohm with the FA503) the higher the resistor value, the lower the max. electrical power the Hypex is able to push through the driver. I decided to keep 100W electrical power budget for the driver. The Hypex can deliver ~400W to a 6 Ohm driver, so if ca. Rdc = Rv the available mid's max power is quartered and 100W stay for the driver. Here some Basta! simulation and measurement:
Did few analysis regarding power compression / thermal modulation for a 18s 6ND430 in the past, here some sim result with effect of voice coil temp and Rv = Rdc:
As I hopefully remember correctly my past calculations had shown that there should be 1/3 less of power compression / thermal modulation in amplitude with Rv=Rdv (and expectable same amount less of current distortion) so I decided that this might be a good rule of thumb for a meanigful Rv value to evaluate and select power budgets. Tweeter channel has less rail voltage ~28V RMS, but tweeter needs / takes also less power.
Both tweeter and mid are kind of that break thermally before they start even slightly to distort, so there is not much noisefloor to gain. (Did no reliable THD measurements till now, but ordered a ISEMCON mike and have a new Motu M2 here for in future hopefully better measurements, soldered new measurement wiring harness just yesterday evening!)
Specifically with these noisefree drivers I would not recommend to invest in a bigger amp module for burning power in a resistor, the effect is not big - I only did this because the FA503 and the power was there before the speaker. In general, for active tweeter / mid channels current amplifiers would make sense, that drops all thermal modulation and current distortion to zero. (NOT the woofer channel, on the resonance frequency I wanna have defined electrical control by voltage amp when the drive is stroking, and not leave the cone to poor Qms highly nonlinear damping properties! Little pre-resistor to allow for fatter woofer magnet selection and then readjusting Qes with Rv to a tunable value for closed or vented box is allowed; "more damping factor = more good" is a myth, but properly sized impedance matching at the Amp-Speaker Interface might have advantages)
As there is not that much contribution to improve distortion that I could hear, I feel that with elevated levels mid and tweeter have a bit more "bang" and dynamics with the resistors due to the thermal support effect of Rv. When the VC temperature and so the restistance increases, Rv has the effect that the drivers clamp voltages increases also to compensate and push more current. But the coil gets even hotter then, so this is a positive feedback loop that tightens the threshold the coil might burn with the next peak...
The Impedance of the parallel notch in series of the driver is tuned (with Virtuixcad, measured FR and IMP responses) to the area of cone resonances to surpress them in the current distortion FR within the driver's band of use. Leave the resistor in the notch out to drop FR as deep as possible, tune Q with L and C to something meaningful. A notch makes much sense with metal/hard cone/dome mids/tweeters, for paper mid cones with more wideband breakup behaviour I would just take a coil in series as lowpass, or leave it out when the breakup is low Q. Again: Specifically with these noisefree drivers there is not much to gain - but these passive parts are not so expensive, and it's a hobby to fiddle around with the speakers... Have TAC/GB + SB15NBAC30 here on FA123 with passive notches on the massive but exact breakups, with these speakers the hearable effect of the notches is obviously positive.
Value of the resistor was selected under the assumption, that for much current feedback it has to be as high as possible.
As I have limited rail voltage (53V RMS = 350W @ 8 Ohm with the FA503) the higher the resistor value, the lower the max. electrical power the Hypex is able to push through the driver. I decided to keep 100W electrical power budget for the driver. The Hypex can deliver ~400W to a 6 Ohm driver, so if ca. Rdc = Rv the available mid's max power is quartered and 100W stay for the driver. Here some Basta! simulation and measurement:
Did few analysis regarding power compression / thermal modulation for a 18s 6ND430 in the past, here some sim result with effect of voice coil temp and Rv = Rdc:
As I hopefully remember correctly my past calculations had shown that there should be 1/3 less of power compression / thermal modulation in amplitude with Rv=Rdv (and expectable same amount less of current distortion) so I decided that this might be a good rule of thumb for a meanigful Rv value to evaluate and select power budgets. Tweeter channel has less rail voltage ~28V RMS, but tweeter needs / takes also less power.
Both tweeter and mid are kind of that break thermally before they start even slightly to distort, so there is not much noisefloor to gain. (Did no reliable THD measurements till now, but ordered a ISEMCON mike and have a new Motu M2 here for in future hopefully better measurements, soldered new measurement wiring harness just yesterday evening!)
Specifically with these noisefree drivers I would not recommend to invest in a bigger amp module for burning power in a resistor, the effect is not big - I only did this because the FA503 and the power was there before the speaker. In general, for active tweeter / mid channels current amplifiers would make sense, that drops all thermal modulation and current distortion to zero. (NOT the woofer channel, on the resonance frequency I wanna have defined electrical control by voltage amp when the drive is stroking, and not leave the cone to poor Qms highly nonlinear damping properties! Little pre-resistor to allow for fatter woofer magnet selection and then readjusting Qes with Rv to a tunable value for closed or vented box is allowed; "more damping factor = more good" is a myth, but properly sized impedance matching at the Amp-Speaker Interface might have advantages)
As there is not that much contribution to improve distortion that I could hear, I feel that with elevated levels mid and tweeter have a bit more "bang" and dynamics with the resistors due to the thermal support effect of Rv. When the VC temperature and so the restistance increases, Rv has the effect that the drivers clamp voltages increases also to compensate and push more current. But the coil gets even hotter then, so this is a positive feedback loop that tightens the threshold the coil might burn with the next peak...
The Impedance of the parallel notch in series of the driver is tuned (with Virtuixcad, measured FR and IMP responses) to the area of cone resonances to surpress them in the current distortion FR within the driver's band of use. Leave the resistor in the notch out to drop FR as deep as possible, tune Q with L and C to something meaningful. A notch makes much sense with metal/hard cone/dome mids/tweeters, for paper mid cones with more wideband breakup behaviour I would just take a coil in series as lowpass, or leave it out when the breakup is low Q. Again: Specifically with these noisefree drivers there is not much to gain - but these passive parts are not so expensive, and it's a hobby to fiddle around with the speakers... Have TAC/GB + SB15NBAC30 here on FA123 with passive notches on the massive but exact breakups, with these speakers the hearable effect of the notches is obviously positive.
Last edited:
P.S.: This is indeed a long unrecognized advantage of typical passive speakers vs. active. E.g. Bernd Timmermanns of Hobby HiFi Magazine has long ago stated and often shown measurements that mid distortion that looks like a echo of the cone resonances e.g. with magnesium cone drivers can be "filtered out" after applying the passive lowpass (he never wrote an explanation why this is so). But with a active speaker you loose this effect and so the advantage. I love my active speakers, and with just a coil and/or a notch I can get that advantage back 🙂
Why does it needs to be either/or?
You can just simply go for an hybrid approach?
Or I guess that's why you mean with that last bit. 🙂
You can just simply go for an hybrid approach?
Or I guess that's why you mean with that last bit. 🙂
Aren't we talking about the same back-EMF induced distortion that Purifi/Lars Risbo addressed? I could imagine a cone in resonance applies forces on the voice coil.
@Kwesi: Thanks a lot for the very good explanation. That Muto 2 looks nice, and you may have conviced me to buy Dayton Dats V3.
Your info is mindblowing. In a good way of course 😀
So what I get from your explanation above.
Bass
We should skip passive filter and let amp take full control of the driver.
Resistor for midrange and tweeter
Without resistor in series with midrange and in series with tweeter, with digital amp we may actually have a response that is non-flat depending on Voltage/SPL and coil tempature. Effectively we have bit of highpass/more energy in lower spectrum than highs. In worst scenario we have relative added energy in the sensitive area of the hearing.
By adding series resistor we add more in the signal path, that intuitively (with no deeper understanding) should be worse it is rather the complete opposite. I guess by adding tweeter we may even improve gainstructure a bit? With series resistor we offload all the bad remaining stuff from the amp into the resistor and let the good stuff pass to the tweeter.
I have Hypex Fusion123 and looks like same tweeter amp as in FA503.
Would you then say that good starting point for resistor in tweeter and midrange is 5.6 Ohm? Like:
https://www.soundimports.eu/en/mundorf-mres20-560.html
Is 20 Watt resistor overkill and should we expect it to get warm? (If it gets warm a bad idea to just throw it in cabinet with damping material to close)
Your info is mindblowing. In a good way of course 😀
So what I get from your explanation above.
Bass
We should skip passive filter and let amp take full control of the driver.
Resistor for midrange and tweeter
Without resistor in series with midrange and in series with tweeter, with digital amp we may actually have a response that is non-flat depending on Voltage/SPL and coil tempature. Effectively we have bit of highpass/more energy in lower spectrum than highs. In worst scenario we have relative added energy in the sensitive area of the hearing.
By adding series resistor we add more in the signal path, that intuitively (with no deeper understanding) should be worse it is rather the complete opposite. I guess by adding tweeter we may even improve gainstructure a bit? With series resistor we offload all the bad remaining stuff from the amp into the resistor and let the good stuff pass to the tweeter.
I have Hypex Fusion123 and looks like same tweeter amp as in FA503.
Would you then say that good starting point for resistor in tweeter and midrange is 5.6 Ohm? Like:
https://www.soundimports.eu/en/mundorf-mres20-560.html
Is 20 Watt resistor overkill and should we expect it to get warm? (If it gets warm a bad idea to just throw it in cabinet with damping material to close)