Hi Charles,
Yep, I'm still looking for a DSP that meets my needs. Most of the units have plenty of MIPS these days (200-400 is typical) so that's no problem, but the bandwidth is. There are units with 96K sampling, enough to avoid the brick wall filter issues, and there are units with flexible CAD-style signal routing, but I haven't found one with both features.
In theory, I suppose you could do the time delay in the analog domain but I'd think it could get ugly stringing that many allpass filters together. Depending on the slope of the lowpass, you're looking at a delay on the order of 1/2 to 1 cycle at the nominal crossover frequency. Any thoughts on that?
Regards,
Dennis
Yep, I'm still looking for a DSP that meets my needs. Most of the units have plenty of MIPS these days (200-400 is typical) so that's no problem, but the bandwidth is. There are units with 96K sampling, enough to avoid the brick wall filter issues, and there are units with flexible CAD-style signal routing, but I haven't found one with both features.
In theory, I suppose you could do the time delay in the analog domain but I'd think it could get ugly stringing that many allpass filters together. Depending on the slope of the lowpass, you're looking at a delay on the order of 1/2 to 1 cycle at the nominal crossover frequency. Any thoughts on that?
Regards,
Dennis
Hi Eric,
the reason why I experimented with an indirect tweeter is because the Manger seems to have a quite directional radiation (due to the 70mm diameter voice coil this is not surprising) and while comparing it to a conventinal speaker (with 25mm dome tweeter) I felt the upper range somewhat underrepresented.
Therefore I thought it might be a good idea to increase the overall radiated energy in the high frequency range by trying an indirect radiating tweeter. I choosed a 10mm Polycarbonat-tweeter. I think it was a modified (removed the Ferrofluid from the air gap) Audax TW74A - (I believe it is the same used as direct radiating tweeter in the "Medea") I had at hand. I run it via passive first order highpass and tried values of the xover between 9kHz - 13 kHz and also had a variable L-pad attenuator to align the volume level.
Actually IMO this gave better results. When trying this You have to take care that the sound from the indirect tweeter does not interfere with waves directly coming from the Manger which means the faceplate of the tweeter has to be big enough that lower frequencies (note that with only a first order highpass the tweeter radiate quite a bit lower frequencies as well) do not bent around. Also You need a considerable delay between the indirect
and direct waves so keep the tweeter away from nearby reflecting walls otherwise the tweeter could harm more than it helps. Also try different crossovers points and volume levels. Depending of the reflecting properties of Your room and walls the results can be different.
For experimental purpose with such an indirect tweeter once I made a very handy small xover board with variable xover-point L-pad attenuator and a switch for comparing the tweeter switched on and off. Below a picture of the board.
P.S.: if You want to try the Audax (without Ferrofluid) You may want to use equalisation (LCR) for the tweeters lower resonance. If I remember right the TW-74A had quite a bump at around 3 kHz due to an underdamped resonance. With only a first order highpass (even crossed at so high frequencies) this could be an issue.
Attachments
Hi Christoph,
I am still confused as to where you placed the tweeter. In the same baffle as the MSW? Somewhere completely different?
I would be afraid that it screws up the excellent temporal characteristics of the MSW.
I have been wondering whether it would actually be possible to place it behind the MSW membrane. But then it is difficult to say how much the membrane will transmit and how much diffraction and reflection will occur on the quite large metal structure of the MSW.
Cheers,
Eric
I am still confused as to where you placed the tweeter. In the same baffle as the MSW? Somewhere completely different?
I would be afraid that it screws up the excellent temporal characteristics of the MSW.
I have been wondering whether it would actually be possible to place it behind the MSW membrane. But then it is difficult to say how much the membrane will transmit and how much diffraction and reflection will occur on the quite large metal structure of the MSW.
Cheers,
Eric
Eric:
I placed it on the rear of the Manger enclosure (I haven`t tried the Manger with a open baffle but with closed box!). IMO placement does not matter so much as far it is ensured that the tweeter does not interfere with the direct radiating waves.
I´m not sure what You mean by "temporal characteristics" but I assume You mean transient response or time-response respectively .
Well, as the tweeter is not intended to radiate from the front baffle but just to raise the high frequency power spectrum of the indirect sound field (reverberation field) I do not consider this as a problem.
The transient response is meaningful only for the first arrival wavefront which are the soundwaves that arrive at the listener position without being reflected from walls or other things.
Only this sound that strike the ears first determine the perception of direction.
Only in case that the tweeter contributes to the direct sound due to effects already mentioned (bending around baffle/enclosure or too short delay between direct and indirect soundwaves due too less space between the tweeter and reflecting things/wall) I could see a problem that it spoils the Manger response.
I do not see a problem doing so. Actually diffraction and reflection of high frequencies is what we would want here. Why don`t You just give it a try? The 10mm Audax are dirt cheap and for experimenting they are so lightweight that You can first mount them with adhesive tape or similar.
Here a link to ebay for the 10mm Audax (Euro 5 a pair!).
Audax 10mm on ebay
Sorry typo
In my last but one post I wrote .........removed the Ferrofluid from the air gap) Audax TW74A - (I believe it is the same used as direct radiating tweeter in the "Medea")......
It is not "Medea" (that is from Audio Physics and do not has a tweeter at all) but the Newtronics "Mega" (always mix that up
)!
And BTW Eric:
As I quit the Klang & Ton subscription quite a while ago because I disliked their new style (I changed to Hobby-Hifi then - the title sounds somewhat silly but the articles and measurements are much better IMO) I would like to ask You if it is possible that You could email me a copy of the Manger article from the Klang & Ton issue 1/03 You mentioned (do You have a scanner)?
It would be very nice to have it.
In my last but one post I wrote .........removed the Ferrofluid from the air gap) Audax TW74A - (I believe it is the same used as direct radiating tweeter in the "Medea")......
It is not "Medea" (that is from Audio Physics and do not has a tweeter at all) but the Newtronics "Mega" (always mix that up
)!And BTW Eric:
As I quit the Klang & Ton subscription quite a while ago because I disliked their new style (I changed to Hobby-Hifi then - the title sounds somewhat silly but the articles and measurements are much better IMO) I would like to ask You if it is possible that You could email me a copy of the Manger article from the Klang & Ton issue 1/03 You mentioned (do You have a scanner)?
It would be very nice to have it.
Step response
Attached you can find the in-room step response of one of my Manger speakers, driven by a cheap two-channel IC amplifier.
The x-over is a 2nd/1st order active subtractive crossover, followed by some notch filter (to at least partially tame the MSW's behaviour around it's fs).
I am sorry for the bad picture quality caused by the file conversion (the green line starts at the y-axis zero, the time axis markers are spaced at 2 ms intervals just to give you some ideas).
The woofer is a Dynaudio 20W75. I also added an LT circuit to achieve an f3 around 30 Hz but I used it only for some casual listening (i.e. it is not included in this measurement). The outcome is that I can get astonishingly deep bass reproduction but the small amp is running out of steam much too early. I will use LF equalizing again as soon as I 1.) have a larger amp at hand and 2.) the experimental box for the more robust Audiotechnology woofer is ready.
The green line in the graph shows how the desired step response of a phase_accurate
speaker would look like (approximately !). The most interesting part of the response are the first few milliseconds where it comes quite close to the wanted response (keep in mind that with an ordinary crossover, the zero line would have been crossed at least once before the 1st millisecond is over !).
Since I aim for a better response and a crossover that is easier to trim, I will try out another crossover topology later on (which is also phase-accurate). The current experimental one I will most probably use with a cheap Monacor broadband (that seems to be better behaved around its fs) and some cheap woofers in woofer-broadband-woofer (didn't dare to call it MTM) configuration.
Regards
Charles
Attached you can find the in-room step response of one of my Manger speakers, driven by a cheap two-channel IC amplifier.
The x-over is a 2nd/1st order active subtractive crossover, followed by some notch filter (to at least partially tame the MSW's behaviour around it's fs).
I am sorry for the bad picture quality caused by the file conversion (the green line starts at the y-axis zero, the time axis markers are spaced at 2 ms intervals just to give you some ideas).
The woofer is a Dynaudio 20W75. I also added an LT circuit to achieve an f3 around 30 Hz but I used it only for some casual listening (i.e. it is not included in this measurement). The outcome is that I can get astonishingly deep bass reproduction but the small amp is running out of steam much too early. I will use LF equalizing again as soon as I 1.) have a larger amp at hand and 2.) the experimental box for the more robust Audiotechnology woofer is ready.
The green line in the graph shows how the desired step response of a phase_accurate
speaker would look like (approximately !). The most interesting part of the response are the first few milliseconds where it comes quite close to the wanted response (keep in mind that with an ordinary crossover, the zero line would have been crossed at least once before the 1st millisecond is over !).Since I aim for a better response and a crossover that is easier to trim, I will try out another crossover topology later on (which is also phase-accurate). The current experimental one I will most probably use with a cheap Monacor broadband (that seems to be better behaved around its fs) and some cheap woofers in woofer-broadband-woofer (didn't dare to call it MTM) configuration.
Regards
Charles
Hi Charles,
looks pretty good! According to the Manger data sheets, the combination of 1st order HP at 370 Hz and 3rd order low pass at 100 also shows near-ideal step response behavior. Do you have a measurement for this also?
Back to the subtractive XO:
Was the 2nd order on the woofer or the MSW? What was the Q of the 2nd order filter?
You are driving the MSW from 0 impedance, so you wouldn't need a notch (or rather, a Zobel) to keep the impedance flat. I gather you are talking about the sensitivity increase towards low frequency that Manger usually cures by choosing to high a HP frequency?
By the way, I have finished my Manger enclosures (20 l with damping beginning 10 cm away from the driver and tiles on the inner sides). I am currently running them full range as I am still working on the woofer. I meant to switch back to the T+As after a quick test, but I am hooked. Imaging is perfect, highs are good in the sweet spot, and bass, although lacking, is sufficient. Bass attacks, which are currently not filtered, do not seem to stress the MSWs anywhere near as much as other FR speakers.
Regards,
Eric
looks pretty good! According to the Manger data sheets, the combination of 1st order HP at 370 Hz and 3rd order low pass at 100 also shows near-ideal step response behavior. Do you have a measurement for this also?
Back to the subtractive XO:
Was the 2nd order on the woofer or the MSW? What was the Q of the 2nd order filter?
You are driving the MSW from 0 impedance, so you wouldn't need a notch (or rather, a Zobel) to keep the impedance flat. I gather you are talking about the sensitivity increase towards low frequency that Manger usually cures by choosing to high a HP frequency?
By the way, I have finished my Manger enclosures (20 l with damping beginning 10 cm away from the driver and tiles on the inner sides). I am currently running them full range as I am still working on the woofer. I meant to switch back to the T+As after a quick test, but I am hooked. Imaging is perfect, highs are good in the sweet spot, and bass, although lacking, is sufficient. Bass attacks, which are currently not filtered, do not seem to stress the MSWs anywhere near as much as other FR speakers.
Regards,
Eric
Hi Eric
2nd order was on the woofer since I didn't want to spoil the midrange and because the MSW can take a fair bit of LF content as you have already experienced by yourself.
The Qes (if applicable at all) of the MSW is quite high, resulting in a Qtc of >1 in the box. That's what the notch is aimed at. The notch filter is consisting of a highpass and a lowpass in parallel so there is not much supression at all. I have to undig the details of the circuit then I will supply them as well.
With your 20 liters things will of course be better than with my 10 liters but for me (read my SO) size is an issue .....
I already ordered the components for a different kind of crossover (including some LM3886 for the power amps) where I intended to build the notch with a simulted LC filter. But I think I will leave this for the sake of simplicity and sound quality. The x-over will also be of the subtractive/analog computing type. I will give info on this one as well.
Regards
Charles
2nd order was on the woofer since I didn't want to spoil the midrange and because the MSW can take a fair bit of LF content as you have already experienced by yourself.
The Qes (if applicable at all) of the MSW is quite high, resulting in a Qtc of >1 in the box. That's what the notch is aimed at. The notch filter is consisting of a highpass and a lowpass in parallel so there is not much supression at all. I have to undig the details of the circuit then I will supply them as well.
With your 20 liters things will of course be better than with my 10 liters but for me (read my SO) size is an issue .....
I already ordered the components for a different kind of crossover (including some LM3886 for the power amps) where I intended to build the notch with a simulted LC filter. But I think I will leave this for the sake of simplicity and sound quality. The x-over will also be of the subtractive/analog computing type. I will give info on this one as well.
Regards
Charles
Hi Charles,
I am currently throwing together a low-component count, yet high quality discrete power amp. I should have it tested and will publish the schematic in about two weeks. Having just built a 3886-based amp and listened to it on my T+As before giving it away, I guess a 3886 will not bring out the full potential of the Mangers.
Did you measure pulse response with the regular Manger filter?
Regards,
Eric
I am currently throwing together a low-component count, yet high quality discrete power amp. I should have it tested and will publish the schematic in about two weeks. Having just built a 3886-based amp and listened to it on my T+As before giving it away, I guess a 3886 will not bring out the full potential of the Mangers.
Did you measure pulse response with the regular Manger filter?
Regards,
Eric
I do not intend to use it with the 3886 amp eternally. This is just meant for measurements and trimming response etc (and some casual listening).
But you should always be aware that almost any amps and drivers perform better when used with a line-level crossover !
I never made any measurements with the original Manger filter but the step response looks exactly how it should look like with that type of crossover and an inverted woofer (there are several sites where you will find it on the web).
Regards
Charles
But you should always be aware that almost any amps and drivers perform better when used with a line-level crossover !
I never made any measurements with the original Manger filter but the step response looks exactly how it should look like with that type of crossover and an inverted woofer (there are several sites where you will find it on the web).
Regards
Charles
The original Manger crossover performs a 180 phase change during the transition, what a subtractive crossover won't do.
Therefore the first zero-crossing of the step response occurs after 700us approx.
To make it short: An original Manger system isn't actually 100%phase_accurate but it is still much better in this repect than most speaker systems (at least through the whole midrange area !).
A big advantage of the original crossover is that both drivers are down by 6 dB at the effective crossover frequency so that it is definitely better regarding efficiency and power handlicng compared with the subtractive crossover.
Furthermore the lowpass is much steeper than the subtractive one, avoiding that the LF drivers spoil the midrange.
OTOH I am still tempted to achieve a system that is transient-perfect.
Regards
Charles
Therefore the first zero-crossing of the step response occurs after 700us approx.
To make it short: An original Manger system isn't actually 100%phase_accurate but it is still much better in this repect than most speaker systems (at least through the whole midrange area !).
A big advantage of the original crossover is that both drivers are down by 6 dB at the effective crossover frequency so that it is definitely better regarding efficiency and power handlicng compared with the subtractive crossover.
Furthermore the lowpass is much steeper than the subtractive one, avoiding that the LF drivers spoil the midrange.
OTOH I am still tempted to achieve a system that is transient-perfect.
Regards
Charles
Yes, I was wondering how the Manger would take the hump you get in the 6 dB branch of a subtractive XO.
Regarding the HF cutoff of the woofer, what is keeping you from making the woofer branch 3rd or even 4th order?
What is causing the phase change in the original Manger XO? I don't remember seeing it in the phase response plots of the Zerobox 109. Will have to do some searching and thinking...
Regards,
Eric
Regarding the HF cutoff of the woofer, what is keeping you from making the woofer branch 3rd or even 4th order?
What is causing the phase change in the original Manger XO? I don't remember seeing it in the phase response plots of the Zerobox 109. Will have to do some searching and thinking...
Regards,
Eric