I already know the answer, but how long till you do something with seas excel line John. 😉 I know it's against your philosophy, but I need some inspiration for my new drivers.
Vikash said:
Don't do like I did and try to mate under damped with critically damped drivers. Ie. lightweight stiff material meets heavy soft material as is the case with the W22 and ATC mid dome.
From all the testing I've done and the experienced gained from doing the Perceive loudspeakers its almost certain that the best timbre/voice matching comes from mating drivers with almost identical material and sonic properties. Also be mindful of dispertion patterns and particularly at the XO points, matching here is critical IMO. Why is this important, well the relationship between direct(speaker) and reverberant(room) soundfields is critical IMO. Direct sound gives most of the phase related cues and also the low level detail which is quickly lost in the reverberant field. And the way we perceive magnitude band balance and the full energy of percussive or impulsive sounds, is dependent upon the power response of the loudspeaker or how evenly it excites the reverberant field with frequency.
Do this well and you'll be well on your way to a seamless transition between the drivers and a tonally even/accurate speaker. This all work in the favour of realism and believability.
Vikash said:
I was wondering the same thing! ... surely an Excel is actually cheap in comparison to that pile of drivers!
Vikash said:
My philosophy is equal parts high performance and high value. I'll start using big dollar drivers when I find one that I think is worth it. 🙂 The Excel line does come close though. The product lines that don't come close to falling on my value/performance curve consist of makes like Scan-Speak, Audio Technology, Accuton, etc. I'm not poor, but one way to *stay* not poor is to be wise with my money. 😀
paulspencer said:
Hmmm, not really. I had considered using W18's in my TMM system. One negative of using W18's in that design would have been that the crossover would then have to be 1.2kHz LR4. But the primary thing is the the system becomes $700 worth of drivers. I'm not interested in tripling the cost for something that sounds 2% better. Though I'll admit some would gladly pay for that extra 2%. To each his own.
I may test the Seas W18NX nextel cone someday though. I read somewhere that it's got some sort of new adaptive surround that provides greater damping in the upper frequencies and less damping for bass frequencies. If I can get the rigidity without the big 3rd harmonic from the breakup node, that could be interesting.
Jasper jig and ebay comment 😉 John's blog always has me smirking, plus good info. Have a read guys.
Do you have something to base your 2% theory on? Just wondering if you've made the comparisson and I've missed it.
Do you have something to base your 2% theory on? Just wondering if you've made the comparisson and I've missed it.
Vikash said:
2% is just a number I pulled out of my butt. It's pretty hard to truly quantify differences with just a single percentage number.
With regard to the test results I just posted, unless I also posted results of, for example, a Seas W18EX or a ScanSpeak 8545, a person could say I'm not qualified to comment. (and I wouldn't disagree) But drivers of that level do occasionally pass through my hands, and even if I've only tested them by themselves or under different conditions, it's enough for me to develop a general idea of the product line's value. So, I have opinions, but they may not be accurate or even based on reality. 🙂 Take everything I say with a grain of salt.
I think we can take your results very seriously. Your tweeter mishmash test contain high priced items and you were right, high price doesn't give more performance.
The little Seas is the best of the bunch, even versus Accuton. If you ever get the chance to test the C12, it would be great. Great way to kill their statement about it being the best tweeter in the world... kill it with a 30$ tweeter.
The little Seas is the best of the bunch, even versus Accuton. If you ever get the chance to test the C12, it would be great. Great way to kill their statement about it being the best tweeter in the world... kill it with a 30$ tweeter.
simon5 said:
A caution on these tweeter HD results, the results as measured are misleading. I had mentioned this before at another forum to Zaph. From Zaph's site:
"The critical range of harmonic distortion is the 1kHz to 3kHz range. This range determines how low you can cross over, and high order harmonics are easily heard this low in frequency. As you go higher in frequency, high order harmonics begin to be unimportant as they occur above audible frequencies. For example, a peak in the F5 at 5.3kHz actually occurs at 26.5kHz and is inaudible. A peak in the F2 at 5.3kHz would be audible at 10.6kHz however, possibly causing brightness, edgyness or "false air". "
About the 1-3Khz range, the measurement results are misleading because most tweeters are rolling off at this range. The tweeters that don't roll off as much as the others, will show higher HD in this range than the ones that roll off. But when used with a xover for a target acoustic FR, the roll offs will be compensated. The tweeter which is flatter in the 1-3Khz range will be attenuated more with the xover than the tweeter which started rolling off at 2Khz. An example: compare FR and HD results of RS28 and 27TBFC at http://www.zaphaudio.com/tweetermishmash/. At 2Khz both RS28 and 27TBFC are at 90db at their FR. And their HD levels are same at 2Khz. Now look at 1Khz FR plots, RS28 is at 89db, 27TBFC is at 84db. In the HD plots, at 1Khz RS28 has more HD than 27TBFC because it is playing 5db louder at 1Khz than 27TBFC. This invalidates the test there. For a fair test, RS28's drive level should have been attenuated 5db lower at 1Khz. The test has inherent unfairness towards tweeters that don't roll off early.
And about the HD numbers at high frequencies, IMHO they are important. Because the existence of them indicates these frequencies are likely to go into intermodulation distortion with each other and with lower frequency notes, which will result in distortion products that fall into the audiable range. 9K and 10Khz will generate IMD product at 1Khz, 8Khz and 11Khz, etc.
The logic is sound, but IMO the conclusion is wrong. I don't think I have my M18W0-09-08 measurements (lost when I switched from that vile virus-catcher to a Mac) or at least Spotlight can't find them right now. (Even then, I'd only have FR and impedance curves, or less than half the story.) However, both objectively and subjectively I find the M18 to be a considerably nicer sounding driver than the PL18. I'd go so far as to say that I prefer the M18 to the Scan 8545, too, because it doesn't impart that annoying lower-mid coloration that every 8545-based system I've heard is stuck with. I wonder if M18 vs. PL18 is a similar situation to the old Vifa D27 vs. newer D27's.
That said, decent as the M18 is for a tiny midwoofer, I think the Peerless HDS (old series; I have no experience with the new ones) or Seas G18 are better in most all aspects save perhaps bass extension. (The M18 can go down to the mid 30's at very decent levels in a Dan Wiggins-style ported alignment.) And all of them are little toys compared to a well-engineered large coax....
Feyz said:
Hi Feyz,
Right on all counts. I recall the discussion we had over on Madisound about this, and it's specifically why I added the comment "these results need careful interpretation" and changed the wording of a few other things. I wouldn't say it invalidates any results, but people should keep the rolloff of the fundamental in mind.
Like I said, I make a large number of concessions to brevity. The fact that I post the harmonics F2 thru F5 as a sweep is one of them. The other way to do it would be "Mark K style" where I do one frequency at a time. The benefit of doing it that way is that you get to see much more than just F2 thru F5. The problem with doing it that way is that you have to do many frequencies to get a full picture. Just a few isn't enough, and in fact I would not be happy with the quantity of information unless I had all the plots in 100hz steps between 1kHz and 3Khz. It's just too limiting for me, not to mention time consuming, bandwidth consuming and information overload for most website viewers. There's a lot more information in an HD sweep, but it has to be interpeted differently.
But, I probably should explain considerations of the low end rolloff a little better on the measurement pages. (and how they affect the CSDs too) It's all true for woofer tests too. I'll make it a point to update the pages sometime.
John
Pallas said:
I should probably not comment on woofers I have not tested. I have been told by others however, that the PL18 has a better motor than the M18. That's pretty much all I have to go on at this point.
John
Feyz said:
You're right.
The easy solution would be to show the HD curve in - xx dB versus the fundamental at 0 dB. That would be easy to do, just substract the HD curve from the frequency response curve and you get the attenuation curve in - xx dB versus the fundamental.
Compensating for rolloff with the amplifier would be a pain in the ***. I'm quite sure that the curve substraction method would be more than enough, even if it's not a perfect method because louder = more distortion.
simon5 said:
I wouldn't rely on that. Becuase even though louder = more distortion, the relationship is not linear most of the time. Cutting the input by 5db, will not necessarily cut the distortion by 5db. It is more likely that it would cut the distortion more than 5db, and we don't know how much. It may even cut it less than 5db but not likely. Think about this way 5db is about half excursion. Considering BL and Suspension vs excursion curves, this may make a big difference if before 5db reduction excursion was at the very nonlinear region of the BL and suspension curves.
It would be if it is done manually, but it could be done automatically on the software that does the measurement. Software could be written such that it will take in the FR of the driver and a target curve, like 2Khz 4th LR high pass, then will adjust its sweep amplitude accordingly as it plays it.
Yeah, that's why I said that sentence, I didn't explain myself correctly. I wanted to say that for example 5 dB louder might mean 10 dB higher distortion. I know it's not a perfect method, but it would be better than nothing.
You're right that a software method would be easy to play with but I don't know if it would be easy to program/design/implement. I'll leave that part to others hehe!
You're right that a software method would be easy to play with but I don't know if it would be easy to program/design/implement. I'll leave that part to others hehe!
What distortion measurements are best
I think that John's HD measurements, in a general sense, parallel my own. There are some differences though. John makes the point that you look at a wider variety of frequencies. While this is true, you also loose something to gain something.
The question is, how useful is HD testing? I think if you review the literature, the general consensus is that IM testing is a better marker of audibility. HD is very useful in looking at the low end limit, but I think a bit less useful as you go up. That is, it's helpful for excursion based distortion numbers. For midband distortion though, I don't think it's all that helpful.
I tend to use the SL style 3 tonebursts because they give a mix of HD and IM. But it might even be better to look at 5 tone or 10 tone IM patterns. Or, as Feyz did on his site, look at an SL style modified with a fourth, lower frequency. That is, if you're looking at a midrange driver covering 300 to 3k, then use a 4 tone signal with a 300 Hz, 2700, 3000, and 3300. That would be one single test. Or, use a multitone test like Jon Risch advocates.
Anyway, don't look at these tests in isolation. Look at them all together. Small differences are probably swamped by production variation anyway.
mark k
I think that John's HD measurements, in a general sense, parallel my own. There are some differences though. John makes the point that you look at a wider variety of frequencies. While this is true, you also loose something to gain something.
The question is, how useful is HD testing? I think if you review the literature, the general consensus is that IM testing is a better marker of audibility. HD is very useful in looking at the low end limit, but I think a bit less useful as you go up. That is, it's helpful for excursion based distortion numbers. For midband distortion though, I don't think it's all that helpful.
I tend to use the SL style 3 tonebursts because they give a mix of HD and IM. But it might even be better to look at 5 tone or 10 tone IM patterns. Or, as Feyz did on his site, look at an SL style modified with a fourth, lower frequency. That is, if you're looking at a midrange driver covering 300 to 3k, then use a 4 tone signal with a 300 Hz, 2700, 3000, and 3300. That would be one single test. Or, use a multitone test like Jon Risch advocates.
Anyway, don't look at these tests in isolation. Look at them all together. Small differences are probably swamped by production variation anyway.
mark k
Mr. Zaph has tested a new set 6,5-7" drivers
http://www.zaphaudio.com/6.5test/
http://www.zaphaudio.com/blog.html
Ascendant Audio Poly 6.5" despite of hardly the raised second harmonic looks very much and very attractively for the 35$
IMO
Great work John!
http://www.zaphaudio.com/6.5test/
http://www.zaphaudio.com/blog.html
Ascendant Audio Poly 6.5" despite of hardly the raised second harmonic looks very much and very attractively for the 35$
IMO
Great work John!
Well here's an ancient thread that popped up again. 🙂
The AA poly 6.5" did indeed measure well. The mildly highish 2nd order in the midbass is directly related to suspension non-linearities. Head over to DIY Mobile Audio for npdang's Klippel test of this driver and you can see it there too. It's funny how all the non-linear tests show the same thing in different ways.
To answer Mark's year old question, "how useful is HD testing?" To me the farfield infinite baffle HD sweeps are the 2nd most important measurement closely behind the far field infinite baffle response curve. That's why I do them. I've done a lot of actual listening following testing, and given these two measurements alone, the driver's "personality" comes through exactly as the measurements would lead me to expect. That boosts my confidence level about the usefulness of my measurements.
It may come as a surprise to some that I actually listen to speakers from time to time rather than just testing them. 😀 Currently, I can't stop listening to Katie Melua's Piece by Piece. Step aside Nora Jones and Diana Krall. The bottom of the CD pile for you. 🙂
The AA poly 6.5" did indeed measure well. The mildly highish 2nd order in the midbass is directly related to suspension non-linearities. Head over to DIY Mobile Audio for npdang's Klippel test of this driver and you can see it there too. It's funny how all the non-linear tests show the same thing in different ways.
To answer Mark's year old question, "how useful is HD testing?" To me the farfield infinite baffle HD sweeps are the 2nd most important measurement closely behind the far field infinite baffle response curve. That's why I do them. I've done a lot of actual listening following testing, and given these two measurements alone, the driver's "personality" comes through exactly as the measurements would lead me to expect. That boosts my confidence level about the usefulness of my measurements.
It may come as a surprise to some that I actually listen to speakers from time to time rather than just testing them. 😀 Currently, I can't stop listening to Katie Melua's Piece by Piece. Step aside Nora Jones and Diana Krall. The bottom of the CD pile for you. 🙂
zaph, you did a good and bad thing this time by not giving any comments on the drivers 😀 the good thing is getting better reading of your distortion plots, the bad thing is i don't have anyone to tell me what is best 😉
from what i see the usher 8945P is having quite an outstanding performance too, am i right thinking that ?
also i think from the pictures that it do have frame venting under the spider, but im not sure, could you confirm ?
if my reading of the graph is right, the usher driver do offer quite a good alternative to the more expensive high end drivers of seas and scanspeak, and also a little step above peerless exclusive offering.
i also found that the two drivers (peerless vs usher phase plug) were also having some similarities in the measurements, wich makes sense, both being made with a copper sleeved motor, semi - hard cone and phase plug.
the nomex appears to be an harder material than carbon paper, and the usher phase plug appears to be better conceived than the peerless one.
oh yeah, and the AA poly 6.5 HD measurements also amazed me, 11mm xmax, good sensivity, cast frame, good distortion numbers except for the 2nd order issue in low mids, for 35 bucks, i don't see where is the trick !
anyway thank you very much for the outstanding work !
now tell me if i didn't understand a thing about measurements 🙂
from what i see the usher 8945P is having quite an outstanding performance too, am i right thinking that ?
also i think from the pictures that it do have frame venting under the spider, but im not sure, could you confirm ?
if my reading of the graph is right, the usher driver do offer quite a good alternative to the more expensive high end drivers of seas and scanspeak, and also a little step above peerless exclusive offering.
i also found that the two drivers (peerless vs usher phase plug) were also having some similarities in the measurements, wich makes sense, both being made with a copper sleeved motor, semi - hard cone and phase plug.
the nomex appears to be an harder material than carbon paper, and the usher phase plug appears to be better conceived than the peerless one.
oh yeah, and the AA poly 6.5 HD measurements also amazed me, 11mm xmax, good sensivity, cast frame, good distortion numbers except for the 2nd order issue in low mids, for 35 bucks, i don't see where is the trick !
anyway thank you very much for the outstanding work !
now tell me if i didn't understand a thing about measurements 🙂
Here's the $140 question:
Could we put four of the AA 6.5 mids into a small sealed box to create a stealth and/or art statement subwoofer?
They appear to have nearly the same swept volume as the Extremis. If I'm interpreting those distortion plots and frequency response plots correctly, the AA 6.5 has distortion comparable to (or better than) the Extremis in the extreme low bass.
If that's the case, I'm getting excited to build something along this line of thinking:
http://cambridgesoundworks.com/store/category.cgi?category=spk_subwoofer&item=c1p300hd
with a PE cabinet and a 500 watt BASH amp. I have a Behringer measurement mic, so I think a linkwitz transform might be within my grasp.
Thanks for the tests, Zaph! I've been looking forward to this batch for a long time.
Could we put four of the AA 6.5 mids into a small sealed box to create a stealth and/or art statement subwoofer?
They appear to have nearly the same swept volume as the Extremis. If I'm interpreting those distortion plots and frequency response plots correctly, the AA 6.5 has distortion comparable to (or better than) the Extremis in the extreme low bass.
If that's the case, I'm getting excited to build something along this line of thinking:
http://cambridgesoundworks.com/store/category.cgi?category=spk_subwoofer&item=c1p300hd
with a PE cabinet and a 500 watt BASH amp. I have a Behringer measurement mic, so I think a linkwitz transform might be within my grasp.
Thanks for the tests, Zaph! I've been looking forward to this batch for a long time.
- Status
- Not open for further replies.