Interesting that the crossover provided by SEAS for the Thor is not the same as the original. The original has a 45 uF cap and .025 mH providing a notch on the woofers the SEAS filter
doesn't have anything close to a 45 uF cap.
If you enter the 45 uF and .025 mH values into this calculator it shows a 4.75 KHz resonance: L-C Resonance Calculator - Pad2Pad
The SEAS Thor crossover uses a .15 mH and 8.2 uF with the calculator showing 4.5 KHz resonance.
The notch frequencies are close but the Q of the notch will be very different.
doesn't have anything close to a 45 uF cap.
If you enter the 45 uF and .025 mH values into this calculator it shows a 4.75 KHz resonance: L-C Resonance Calculator - Pad2Pad
The SEAS Thor crossover uses a .15 mH and 8.2 uF with the calculator showing 4.5 KHz resonance.
The notch frequencies are close but the Q of the notch will be very different.
Last edited:
Quite so. Just to confirm, this is the original Thor crossover as designed by Joe D'Appolito, and depicted in his AudioXpress magazine article. Crossover is LR4 at 2.5KHz. The stated goals for the filter were to:
It achieves all those things, as can be seen from the measurements in D'Appolito's writeup. Unfortunately it also received significant criticism over the years because many felt it crossed the midbass units higher than ideal for those drivers with LR4 slopes, and provided insufficient suppression of the W18E001's bell mode, resulting in an audible spike in HD3 around 1.5KHz. I can't honestly comment on that point; I've heard the Thor a few times, but not for long enough in familiar systems to come to any conclusions. Speaking purely in the abstract, since the W18's bell mode is at a relatively low frequency I'd probably prefer to cross lower (assuming LR4), but that's just me, and I'm hardly in a position to criticise somebody like Joseph D'Appolito on the subject of crossover design, especially since the macro aspects of the Thor's axial response and (horizontal) polars are excellent.
It achieves all those things, as can be seen from the measurements in D'Appolito's writeup. Unfortunately it also received significant criticism over the years because many felt it crossed the midbass units higher than ideal for those drivers with LR4 slopes, and provided insufficient suppression of the W18E001's bell mode, resulting in an audible spike in HD3 around 1.5KHz. I can't honestly comment on that point; I've heard the Thor a few times, but not for long enough in familiar systems to come to any conclusions. Speaking purely in the abstract, since the W18's bell mode is at a relatively low frequency I'd probably prefer to cross lower (assuming LR4), but that's just me, and I'm hardly in a position to criticise somebody like Joseph D'Appolito on the subject of crossover design, especially since the macro aspects of the Thor's axial response and (horizontal) polars are excellent.
Attachments
Last edited:
Above for easy comparison is the jimangie1973 version.
https://www.diyaudio.com/forums/multi-way/64799-clarity-seas-thor-kit-100.html#post2240904.
E0018-08S W18E001
E0011-06 T25CF002
Terrific rabbithole to go down:
THOR: A D’Appolito Transmission Line | audioXpress.
Many aspects debatable by posterity. Joe d'Appolito doesn't want to revisit it. Feels the main interest was the transmission line. And he's done with that.
Last edited:
Is he? I don't recall where he said that -do you have a link, Steve?
Either way, I personally lean toward filters with lower filter frequencies for drivers like these, but (but) with the aforementioned caveat that I take nothing for granted either since I don't have extensive experience with the original crossover, and anybody who writes off a speaker designed by Joseph D'Appolito without that would be foolish indeed. I'd also note that I'd revise at least some of my original views about the enclosure in light of 15 years further knowledge. The Thor may not be perfect (what is) but it's still an interesting design that stimulates debate, and is enjoyed by many. After all, it's still with us two decades down the line, and I hope it continues to be for many years to come.
Either way, I personally lean toward filters with lower filter frequencies for drivers like these, but (but) with the aforementioned caveat that I take nothing for granted either since I don't have extensive experience with the original crossover, and anybody who writes off a speaker designed by Joseph D'Appolito without that would be foolish indeed. I'd also note that I'd revise at least some of my original views about the enclosure in light of 15 years further knowledge. The Thor may not be perfect (what is) but it's still an interesting design that stimulates debate, and is enjoyed by many. After all, it's still with us two decades down the line, and I hope it continues to be for many years to come.
Yes, Scottmoose, I don't make this stuff up!
Page 1: THOR: A D’Appolito Transmission Line | audioXpress
These days I think MTM with a ribbon is a good idea:
This by Mr. Rick Craig of Selah Audio. I quite like BW3 for flat power response. But there must be some issue with the dispersion and falloff with distance of a standard MTM that MTTM styles fix. Nobody agrees with me except adison and Mr. Rick. But I can do the math.
Joe d'Appolito said:
Page 1: THOR: A D’Appolito Transmission Line | audioXpress
These days I think MTM with a ribbon is a good idea:
This by Mr. Rick Craig of Selah Audio. I quite like BW3 for flat power response. But there must be some issue with the dispersion and falloff with distance of a standard MTM that MTTM styles fix. Nobody agrees with me except adison and Mr. Rick. But I can do the math.
I didn't say you did, I asked if you had a link as I hadn't seen said remarks, only the original article. Perfectly reasonable that he shouldn't be interested in updating the Thor as it was a commissioned project from Seas, and he's not likely to go back and develop a new version unless he's paid to do so. I don't see in that link where he says he's done with TLs though -is there another one?
Be that as it may, since I'm not over-familiar with the Selah Audio range and I don't have time to look into their history right now as I've a paper to finish writing, can you point me toward some of Rick's MTTM (twin tweeter) models? It would be very interesting to see how they were implemented. Or are you just using TT as a convenient synonym for 'ribbon'?
Be that as it may, since I'm not over-familiar with the Selah Audio range and I don't have time to look into their history right now as I've a paper to finish writing, can you point me toward some of Rick's MTTM (twin tweeter) models? It would be very interesting to see how they were implemented. Or are you just using TT as a convenient synonym for 'ribbon'?
Last edited:
I am probably verging more on the humourous than the sensible here.
Imagine the meeting at SEAS Headquarters in Moss, Norway.
Boss employer: "Why did we employ Joe d'Appolito to design our flagship speaker?"
Lowly student employee: "Because he was cheap?"
Boss employer: "Sounds reasonable."
But, TBH, a terrible over-lively room. A carpet would help.
Imagine the meeting at SEAS Headquarters in Moss, Norway.
Boss employer: "Why did we employ Joe d'Appolito to design our flagship speaker?"
Lowly student employee: "Because he was cheap?"
Boss employer: "Sounds reasonable."
But, TBH, a terrible over-lively room. A carpet would help.
Last edited:
Thanks for the link to the original article and rereading it the crossover looks to be very
well thought out. An in phase even order crossover should be 6 dB down at the crossover
point, and combined with full 6 dB of baffle step it should be 12 dB down, this design is
10 dB down at XO so it has about 4 dB of baffle step.
The design has about a 1 dB bump at the crossover point but I believe that a gentle dip
of about 1 dB sounds better, sometimes much better.
Rereading it my vague memory is that I simulated the published tweeter crossover and
the electrical response did not match what was published in the article. Perhaps I'll try it
again in Xsim, anyone have a ZMA file for the tweeter? I'm guessing that a component
value was incorrectly given in the schematic if I'm right about the match.
The crossover design intent is there to provide an outstanding sounding speaker.
well thought out. An in phase even order crossover should be 6 dB down at the crossover
point, and combined with full 6 dB of baffle step it should be 12 dB down, this design is
10 dB down at XO so it has about 4 dB of baffle step.
The design has about a 1 dB bump at the crossover point but I believe that a gentle dip
of about 1 dB sounds better, sometimes much better.
Rereading it my vague memory is that I simulated the published tweeter crossover and
the electrical response did not match what was published in the article. Perhaps I'll try it
again in Xsim, anyone have a ZMA file for the tweeter? I'm guessing that a component
value was incorrectly given in the schematic if I'm right about the match.
The crossover design intent is there to provide an outstanding sounding speaker.
Last edited:
That it is. It's the resonance amplified HD3 around 1.5KHz some hear that caused alternative filters to be developed. YMMV -as noted, from a purely theoretical POV I lean that way myself, but I haven't spent much time with the original, and you'd hardly rule out a filter design from the source in question without a plenty of first-hand time with it.
I'm not sure what you boys are talking about here.
But if it was me, I'd be thinking about a Fountek JP3 tweeter and a 2.5-2.7 kHz crossover.
http://www.troelsgravesen.dk/download/Acapella_all_files.pdf
I generally trust Troels Gravesen in these matters. If he says it sounds good, I tend to believe him.
But if it was me, I'd be thinking about a Fountek JP3 tweeter and a 2.5-2.7 kHz crossover.
http://www.troelsgravesen.dk/download/Acapella_all_files.pdf
I generally trust Troels Gravesen in these matters. If he says it sounds good, I tend to believe him.
No? Fair enough, I know these aren't the kind of drivers you like working with Steve, so you might not have run across it before.
Take a gander at the FR of the W18E001. No issue with spotting the c. 4.4KHz bell mode. If you look at an HD plot, you'll see this spike up at the same frequency. No surprise there. However it's not quite as straightforward as that, because that bell mode amplifies the higher order distortion modes (most notably HD3) lower down the range. Take the cone resonance mode & divide by 3; you'll see the spike in HD3.
The problem we have is that if you don't stamp on the cone's main resonant mode hard enough, that resonant-amplified distortion lower down the range (i.e. in the band you're using the driver) is still present, and potentially audible. A straightforward notch isn't always sufficient to prevent it -it depends where you're crossing, how severe the mode & the distortion it causes are. For example, you'll see in his tests of an early EX version of this unit, Zaph notes he wouldn't cross higher than about 1.6KHz, LR4 or higher order to ensure it would not be an issue. Speaking in in the abstract, it's an approach I tend to use myself, and some of the people who used the original Thor claim to hear it. As noted, I can't comment on that, and I take nothing for granted either way: D'Appolito is not exactly known for a lack of skill in filter design (I think that might be a contender for understatement of the year), but HD and its audibility are funny things, and some are more susceptible to it than others. To that you also have to balance off the need for a kit to keep the filter complexity to feasible levels, and also consider tweeter power-handling & IM distortion. The Millennium is a bit of a tank, but Seas woofers can handle a lot of power so going a bit conservative is sometimes preferred depending on where your priorities lie. As ever, YMMV.
Take a gander at the FR of the W18E001. No issue with spotting the c. 4.4KHz bell mode.
If you look at an HD plot, you'll see this spike up at the same frequency. No surprise there. However it's not quite as straightforward as that, because that bell mode amplifies the higher order distortion modes (most notably HD3) lower down the range. Take the cone resonance mode & divide by 3; you'll see the spike in HD3.The problem we have is that if you don't stamp on the cone's main resonant mode hard enough, that resonant-amplified distortion lower down the range (i.e. in the band you're using the driver) is still present, and potentially audible. A straightforward notch isn't always sufficient to prevent it -it depends where you're crossing, how severe the mode & the distortion it causes are. For example, you'll see in his tests of an early EX version of this unit, Zaph notes he wouldn't cross higher than about 1.6KHz, LR4 or higher order to ensure it would not be an issue. Speaking in in the abstract, it's an approach I tend to use myself, and some of the people who used the original Thor claim to hear it. As noted, I can't comment on that, and I take nothing for granted either way: D'Appolito is not exactly known for a lack of skill in filter design (I think that might be a contender for understatement of the year), but HD and its audibility are funny things, and some are more susceptible to it than others. To that you also have to balance off the need for a kit to keep the filter complexity to feasible levels, and also consider tweeter power-handling & IM distortion. The Millennium is a bit of a tank, but Seas woofers can handle a lot of power so going a bit conservative is sometimes preferred depending on where your priorities lie. As ever, YMMV.
This is a perfect example of how superposition _doesn't_ work in a non-linear system.
The motor is non-linear, the peak in the cone's transfer function amplifies the distortion
and a notch in the crossover may fix the frequency response but it does not correct
the distortion "amplification" because it is a non-linear system.
I pointed this out many years ago and I also prefer drivers without high Q peaks in the response.
The motor is non-linear, the peak in the cone's transfer function amplifies the distortion
and a notch in the crossover may fix the frequency response but it does not correct
the distortion "amplification" because it is a non-linear system.
I pointed this out many years ago and I also prefer drivers without high Q peaks in the response.
I wouldn't touch metal drivers with a bargepole.
But they do have a lack of colouration allegedly...
Troels Gravesen has done a huge amount of work with these sort of things:
TJL-2W
Also in the Acapella LWJ experiments. Where he used a ribbon sometimes. I have theoretical grounds for thinking a ribbon is better in an MTM.
DIY-Loudspeakers
Awful complex crossovers. Lot of work to cure that bell mode.
But they do have a lack of colouration allegedly...
Troels Gravesen has done a huge amount of work with these sort of things:
TJL-2W
Also in the Acapella LWJ experiments. Where he used a ribbon sometimes. I have theoretical grounds for thinking a ribbon is better in an MTM.
DIY-Loudspeakers
Awful complex crossovers. Lot of work to cure that bell mode.
It's also why you sometimes see people pulling their hair out over what they think is an over-bright tweeter & getting nowhere because it's not actually the tweeter that's causing the problem.
Drivers without high Q peaking certainly make life easier... I do like metal cones when used well though. Filtering doesn't necessarily have to be hyper complex -it depends on the baseline motor performance and where the main cone mode is. Seas's L18 is a good example. Doesn't have the Excel motor but they've got the main cone mode up to about 7KHz, so if you're crossing below about 2.5KHz & don't have major diffraction worries, you should be able to hit LR4 with 3 or 4 components without having issues with that HD amplification lower down the range.
Drivers without high Q peaking certainly make life easier...
I do like metal cones when used well though. Filtering doesn't necessarily have to be hyper complex -it depends on the baseline motor performance and where the main cone mode is. Seas's L18 is a good example. Doesn't have the Excel motor but they've got the main cone mode up to about 7KHz, so if you're crossing below about 2.5KHz & don't have major diffraction worries, you should be able to hit LR4 with 3 or 4 components without having issues with that HD amplification lower down the range.
Last edited:
I wouldn't touch metal drivers with a bargepole.
Metal cones have their place. I'm listening to a pair of Seas L22s at the moment, crossed around 950Hz to a pair of B&C DE250.
If, as some DIYers like to, you tried listening to the Seas L22 units full-range to see if they had any "potential", you'd assume them faulty. The kHz range is savage. It is possible, though, to hammer them into shape and IMO the results are worth it.
Chris
Thank you all for your advice. I’ve run the speakers for over 100 hours and they have vastly improved. I’m now doing a comparison to the PMC speakers and they are better in every way apart from the high frequencies, which compared to the PMC sound slightly muffled. I’ve currently got a 2.2ohm resistor and will try a lower value, but I’m not sure that will fix the problem as it doesn’t seem a ‘volume’ problem, perhaps it’s another part of the circuit that I should focus on? All advice welcome...
If you are looking for more air in the very top end put a cap across the 2.2 R but first
I'd try dropping it to 1 or even 0 ohms. If you drop the R and there is too much output
at the crossover point (bottom end of the tweeter) then the cap will tilt up the top end.
From memory the top end is tilted down, so the cap might just get the design closer to flat.
I'd try dropping it to 1 or even 0 ohms. If you drop the R and there is too much output
at the crossover point (bottom end of the tweeter) then the cap will tilt up the top end.
From memory the top end is tilted down, so the cap might just get the design closer to flat.
Here's a site with a file for the T25CF002 ZMA file:
Driver FRD & ZMA files
It would be very easy to simulate the XO filter response in Xsim and compare it
to what was published in the article.
Driver FRD & ZMA files
It would be very easy to simulate the XO filter response in Xsim and compare it
to what was published in the article.
Thanks PB2.
I have the top end sounding perfect now, the only thing that I need to fix is somewhere in the middle. Not sure where exactly but singing/voices are very quiet. I’m not sure if it is the woofer or tweeter that needs adjusting. I’ve got them bi-amped so have been able to do a direct comparison with the PMC’s when driving only the woofer or only the tweeter. The PMC woofer definitely crosses over much higher, but that’s what I was expecting as I know the XO that I have is fairly low, from memory I think it was about 1500Hz. Maybe I need to do some measurements to work out what’s missing?
I have the top end sounding perfect now, the only thing that I need to fix is somewhere in the middle. Not sure where exactly but singing/voices are very quiet. I’m not sure if it is the woofer or tweeter that needs adjusting. I’ve got them bi-amped so have been able to do a direct comparison with the PMC’s when driving only the woofer or only the tweeter. The PMC woofer definitely crosses over much higher, but that’s what I was expecting as I know the XO that I have is fairly low, from memory I think it was about 1500Hz. Maybe I need to do some measurements to work out what’s missing?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.