Exactly and this is the major problem with the ATC, it has very low distortion and a nice smooth frequency response, but the physical size of the thing means that your C2C distance is going to dictate how high up you should realistically take it.
The ideal thing would be to use a small, but very well designed neo tweeter. Scanspeak are apparently coming up with a Be version of their Illuminator neo's and at a significantly lower price to their other Be domes.
All this said Scan's newer lines of tweeters have moved in a different direction from their traditional designs. That is they've focused more on reducing the third and higher order harmonics at the expense of 2nd.
The original revelator, the D2905/990000 has exceptional measured performance. In terms of harmonic distortion it really is as good as it gets.
An externally hosted image should be here but it was not working when we last tested it.
Turn up the volume some and things look like this...
An externally hosted image should be here but it was not working when we last tested it.
Things look practically unchanged except for slightly higher levels. This is top notch performance, you cannot ask for more. The only problem with this tweeter is its wonky frequency response. This is the same as one of the Be domes, it's because of the slight waveguide loading of the face plate. This is actually a good thing but it does require a bit of crossover work to iron out. In the case of this design it is also going to increase the C2C spacing and thus want you to push the xover down. If I were to use the ATC and 9900 I'd probably cross them around 2kHz.
If you compare the 9900s distortion performance to that of the similar Be dome things look like this...
An externally hosted image should be here but it was not working when we last tested it.
This is the 2.83v measured distortion performance. For all intents and purposes the older tweeter has better performance and both require work in the crossover to sort things out.
The 6640?
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Isn't as linear as the 7140, which isn't as linear as the 9900.
Where the 6640 has the edge is in linear distortion as it has a nice flat and extended frequency response.
An externally hosted image should be here but it was not working when we last tested it.
vs the 7140
An externally hosted image should be here but it was not working when we last tested it.
and then vs the 9900
An externally hosted image should be here but it was not working when we last tested it.
The 6640 wins in terms of ease of use. The 7140 has a very smooth frequency response, it will actually be quite easy to work with but only if you can measure things properly so you can compensate it properly. The 9900 is even harder to work with.
The Be domes have an inherent droop to their top octave and usually a worse off axis response compared to soft domes, this is because soft domes breakup and the breakup improves their upper end frequency response. Metal domes usually employ a phase shield to keep their response attractive in the top octave but the 6640 does not. As Scott said, using them in a wave guide makes a lot of sense as this helps to sort that out, to this end the 7140 wins out due to its inherent wave guide loading.
Scan do have the good old 9700 though. This is very similar to the 9900 but it doesn't have the wave guide loading. It has nice low distortion.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
But comes with the added advantage of having a nice flat frequency response.
An externally hosted image should be here but it was not working when we last tested it.
Of their Illuminator domes though one of the better ones is the D3004/6020. It has the advantage of being small, it's flange is only 62mm.
It has nice low distortion.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
And a nice flat frequency response.
An externally hosted image should be here but it was not working when we last tested it.
The only problem is it's high qts on the low end, due to the smallish chamber, that causes a rising response.
Scan do have the D3004/602010 and this driver has a larger chamber that sorts this out, so could possibly be a very good option.
The other website does not have measurements of this driver for comparison, but Zaph has measured it.
Low distortion and a nice flat frequency response, it could be a very nice option as the small flange fits well with the huge size of the ATC. The upcoming Be dome of the same neo series should be very nice too.
"We live in the midrange" Paul Klipsch.
Designing your speaker around the ATC SM75-150 SPL+polars could help your tweeter selection process. Production ATC speakers like the ACM100 use 380Hz & 3.5kHz crossovers with the ATC dome. Studying the polar response of the SM75-150 around 3K-3.5KHz, and comparing this data with the polar response of different dome tweeters could allow you to design a crossover with a smooth polar directivity function.
The attached picture is my cut-and-paste of a SM75-150 SPL+polars with a Vifa tweeter SPLs at 3Khz to illustrate this polar matching. The ATC schematic also shows the high complexity of the stock passive Xover for the SM75-150 on the ACM100.
Designing your speaker around the ATC SM75-150 SPL+polars could help your tweeter selection process. Production ATC speakers like the ACM100 use 380Hz & 3.5kHz crossovers with the ATC dome. Studying the polar response of the SM75-150 around 3K-3.5KHz, and comparing this data with the polar response of different dome tweeters could allow you to design a crossover with a smooth polar directivity function.
The attached picture is my cut-and-paste of a SM75-150 SPL+polars with a Vifa tweeter SPLs at 3Khz to illustrate this polar matching. The ATC schematic also shows the high complexity of the stock passive Xover for the SM75-150 on the ACM100.
Attachments
Thanks that makes for very informative reading. That quite frankly explains the 9900's behaviour - spectacular, but clearly a bit uneven on occasion. As I thought earlier, the 'safest' option is to go for the BE 6640 Illuminator, but I'll certainly keep an eye on polar matching as a way to align matching drivers.
The tweeter in the picture you provided is a SEAS tweeter not a Vifa, although it appears there are several versions of the ACM100 that use different tweeters.
While it is true that a good directivity match is optimal, especially due to the ATC having a mild wave-guide loading of its own, I'd argue that the C2C distance with a standard flange tweeter would really mean that you'd be better off crossing lower 3-3.5k. More like 2k if you want your vertical off axis to look good too.
The ATC really does support crossovers much higher than that and naturally deserves one too, it's just that the gigantic ferrite magnet structure creates a driver with a huge flange relative to its actual dome diameter. ATC should really investigate using a neo magnet imo and shrink the flange down. Still if you're ignoring the C2C spacing limitations and recommendations towards the vertical off axis then something like the SEAS DXT tweeter would have a decent enough polar match too.
Which Vifa tweeter is it you are saying that the ATC would be a good match with anyway?
While it is true that a good directivity match is optimal, especially due to the ATC having a mild wave-guide loading of its own, I'd argue that the C2C distance with a standard flange tweeter would really mean that you'd be better off crossing lower 3-3.5k. More like 2k if you want your vertical off axis to look good too.
The ATC really does support crossovers much higher than that and naturally deserves one too, it's just that the gigantic ferrite magnet structure creates a driver with a huge flange relative to its actual dome diameter. ATC should really investigate using a neo magnet imo and shrink the flange down. Still if you're ignoring the C2C spacing limitations and recommendations towards the vertical off axis then something like the SEAS DXT tweeter would have a decent enough polar match too.
Which Vifa tweeter is it you are saying that the ATC would be a good match with anyway?
I'd have said something like 2.4 kHz (with either SS Be tweeter), but still quite close to my thoughts.
The nice thing here is that you can use a steep crossover (..unlike with a larger cone driver), because the off axis performance won't go to crap above 1.4-1.6 kHz:
Ex:
http://www.zaphaudio.com/temp/ATC-SM75-150S-offaxis-0-15-30-45-60.gif
So, you have been prepared for this problem. It means, the way you compensate for this will add a huge uncertainty to the real sound. But you cannot leave it as it is, because you will have worse vocal than the older Be.
IMO, you need to have good ears to make the new Be works. Because of the reason above. But I believe the new Be has "potential" that is not so clear.
The more detailed sound is obvious with its mechanical resistance. What is not so obvious imo is the difference in harmonic distortion. The problem is with the older Be. I doubt it is as good as what people think it is. The newer Be should be way better.
But with such an expensive budget on tweeter, why don't you study more options especially for the midrange unit.
Last edited:
Starting to think I should just accept the default ScanSpeak 9900s and review tweeters later, after all this. Especially if I can get their slight wonkyness sorted at crossover level.
The good news is, I've managed to find a pair complete with the 9900 units and a proper sorted crossover. The bad news is I now have a pair of Volt Rv3143's on the way that are going spare (already advertised in the swap area).
Thanks for all your feedback, and I certainly will keep an eye out on the new ScanSpeak BE units, amongst others.
Thanks for all your feedback, and I certainly will keep an eye out on the new ScanSpeak BE units, amongst others.
Received the speakers on Thursday. That 9900 may measure well, but its quite easy to bait into sibilance. Have tried EQ'ing down, as well as reversing phase of mid to see if that would make much joy. Also placed the grille covers back on and set the speakers back to 0 axis which reduced it a little, but still there.
I'm thinking therefore of changing the tweeter much sooner than planned. Wanted to wait for the new ScanSpeak BE but quite frankly the tweeter is causing me and my partner pain and angst, and ends up making us miss the Usher DMD in my previous speakers.
I'm also considering these : www.usheraudio.com.tw/wp-content/download/speaker/9980-20DMD.pdf as I have a pair to hand I can test with. Any thoughts? Guess it's going to be best just to try them...
I'm thinking therefore of changing the tweeter much sooner than planned. Wanted to wait for the new ScanSpeak BE but quite frankly the tweeter is causing me and my partner pain and angst, and ends up making us miss the Usher DMD in my previous speakers.
I'm also considering these : www.usheraudio.com.tw/wp-content/download/speaker/9980-20DMD.pdf as I have a pair to hand I can test with. Any thoughts? Guess it's going to be best just to try them...
Oh I realise that the 9980-20DMD is a good 2db lower than the remaining drivers, but I'm guessing reducing the other drivers by 2db wouldn't be too horrid a thing to do....
If not, ScanSpeak D3004 / 664000 it is...
If not, ScanSpeak D3004 / 664000 it is...
I hate to be "that guy", but maybe another factor to consider: if you go with a small form factor tweeter as 5th element touches on, you can really shrink the CTC distance.
Unfortunately not Be though.
http://www.scan-speak.dk/datasheet/pdf/d3004-602010.pdf (performed well per Zaph)
Unfortunately not Be though.
http://www.scan-speak.dk/datasheet/pdf/d3004-602010.pdf (performed well per Zaph)
I don't mind considering non-BE drivers, if they have similar characteristics to the BE/DMD (if not precisely the same).
The DMD and the BE are quite a bit smaller than the 9900 in place, so will be able to inch it a bit closer. I can see what 5th element is onto though - compared to my current Ushers, the ATC mid and ScanSpeak 9900 close up do sound less integrated, though from a seated position it doesn't appear to be too bad.
I've been using the Usher DMDs in a commercial speaker for a few years now. It's so refined, balanced and often integrates so well that I forget its a tweeter hence the obsession to try it or something of similar statue.
Does anyone have any idea when this new fangled ScanSpeak BE driver appears?
The DMD and the BE are quite a bit smaller than the 9900 in place, so will be able to inch it a bit closer. I can see what 5th element is onto though - compared to my current Ushers, the ATC mid and ScanSpeak 9900 close up do sound less integrated, though from a seated position it doesn't appear to be too bad.
I've been using the Usher DMDs in a commercial speaker for a few years now. It's so refined, balanced and often integrates so well that I forget its a tweeter hence the obsession to try it or something of similar statue.
Does anyone have any idea when this new fangled ScanSpeak BE driver appears?
Last edited:
Ah if its edgy, its staying well away from me. That's my major issue with the 9900 - hard edges at normal volumes, nevermind trying to shake the house down volumes.
If the 9900 is edgy - What causes it? If the distortion levels are "class leading" then surely it can only be the frequency response that could cause it?
I can measure this at some point using a UMIK-1. Prior to the speakers arriving to me, they were measured by a decent audio consultant (Audio Components) to be very flat and tailing off 15khz+. As mentioned tried to EQ down, but no good. Even at modest volumes you could hear the sibilance from a mile off.
I knew I'd be changing this tweeter at some point, just a little surprised I'm going to have to do it so soon. As you say its 'class leading' measurement wise, but our ears say otherwise in terms of behaviour elsewhere. Attached are the independent measurements from when the crossovers were last tuned - spot on.
Attachments
Of course they are edgy, manufacturer showed it. So, Scan Speak makes
cheaper drivers better than the more expensive ones? What's not to like?
The measurement cuts of far too early to see what's really going on, but It looks like there's 2.5dB peak from 1.5k to around 5k.
- Status
- Not open for further replies.