Perhaps you should talk to the person who will be doing the xover work for you so that they can help advise you on driver choices. There is nothing wrong with using notches anywhere in a design if they are needed. It seems as if you're introducing severely limiting criteria without understanding exactly what they mean or understanding if they are necessary or could potentially, even with the perfect drivers, actually make things worse.
If you want a very sensitive midrange, that will extend low and high, then you need a big driver. The B&W is a case in point. It doesn't go low enough for you, the only way around this is to increase the surface area, which will mean an even poorer off axis response and the potential for an even worse case of breakup. Yet you also want to use a minimal crossover for probably no more reason then 'simple is better', which in reality isn't really true.
If you really want to have true 1st order acoustic crossovers then talk to your xover designer about what drivers you should pick. As an example the scan ring dome, 10f and a suitable 8" would probably be about as good as it gets, but it will still have compromises and will most likely need an xover at least as complicated as other higher order designs to work properly.
If you want a very sensitive midrange, that will extend low and high, then you need a big driver. The B&W is a case in point. It doesn't go low enough for you, the only way around this is to increase the surface area, which will mean an even poorer off axis response and the potential for an even worse case of breakup. Yet you also want to use a minimal crossover for probably no more reason then 'simple is better', which in reality isn't really true.
If you really want to have true 1st order acoustic crossovers then talk to your xover designer about what drivers you should pick. As an example the scan ring dome, 10f and a suitable 8" would probably be about as good as it gets, but it will still have compromises and will most likely need an xover at least as complicated as other higher order designs to work properly.
I'm not against notches, but problems in the 1-3k range are the most sensitive ones. I know because I had to fix one.
The compromise between poor off-axis response and cone size I think my choice is clear, although two 6" with powerful magnet may work decently.
I surely will ask for advice from the xo designer but I need to have all cards on the table before I commission the work (and I have 3-4 months ahead to define better the project specs).
About 1st order acoustic xo, I'm not inflexible at all, and it is not to KIS, but to have a phase response pretty linear at one side and not too many passive components in front of the speakers. The b&w driver requires more than a second order xo at 250hz, that could be a problem not only for a few caps and inductors on the signal but for lack of spl at those frequencies. This i learnt from direct experience with the last speakers I built (with digital xo so I could learn and tweak endlessly), so it's not just plain stubbornness.
A compromise I do not want to make is to cross within the voice range (250-3500 at the very minimum).
In the next weeks, I'll try to find a suitable driver from PHL, then a little more for other input, then I'll present a list of options to my designer.
The compromise between poor off-axis response and cone size I think my choice is clear, although two 6" with powerful magnet may work decently.
I surely will ask for advice from the xo designer but I need to have all cards on the table before I commission the work (and I have 3-4 months ahead to define better the project specs).
About 1st order acoustic xo, I'm not inflexible at all, and it is not to KIS, but to have a phase response pretty linear at one side and not too many passive components in front of the speakers. The b&w driver requires more than a second order xo at 250hz, that could be a problem not only for a few caps and inductors on the signal but for lack of spl at those frequencies. This i learnt from direct experience with the last speakers I built (with digital xo so I could learn and tweak endlessly), so it's not just plain stubbornness.
A compromise I do not want to make is to cross within the voice range (250-3500 at the very minimum).
In the next weeks, I'll try to find a suitable driver from PHL, then a little more for other input, then I'll present a list of options to my designer.
You probably wont need more then 2nd order electrical. Once inside a normal sized cabinet the fs will rise, putting it closer to 250Hz, yes this will be relatively high Q, but combine this with a suitable 2nd order electrical xover and you will end up with a 4th order acoustic. This the driver can just about handle, but you will most likely require a notch filter to flatten the impedance peak, due to fs, otherwise the 2nd order filter wont do what you'd expect. Any mid crossing close to fs will most likely require this anyway.
The reason I say talk with your xover designer is because if you tell them some of your priorities they will be able to advise you as to which drivers to use and to drivers they are happy working with.
The reason I say talk with your xover designer is because if you tell them some of your priorities they will be able to advise you as to which drivers to use and to drivers they are happy working with.
I didn't consider the ATC dome mid because I thought it was low efficiency, but I found out that is 94db. Yes like all small drivers, it doesn't go as low as I want, and costs probably more than the budget I set. But it would be the best to cross at 4khz.
Food for thought.
Btw, I also remembered that I had speakers with 380hz xo and it was a disaster to multiamp unless both amps were the same or had the same timbre. This solution (anything higher than 250hz) would force me to go 4-way, adding some 8-10" woofer to cover this area, maybe enlarged even more to something like 150-400hz.
Unless I fall in love with the ATC or the B&W mid, this will not happen for this project -it has to stay a 3-way with the xo points I defined.
Food for thought.
Btw, I also remembered that I had speakers with 380hz xo and it was a disaster to multiamp unless both amps were the same or had the same timbre. This solution (anything higher than 250hz) would force me to go 4-way, adding some 8-10" woofer to cover this area, maybe enlarged even more to something like 150-400hz.
Unless I fall in love with the ATC or the B&W mid, this will not happen for this project -it has to stay a 3-way with the xo points I defined.
Last edited:
As you're from Italy, I believe you can easily get this fine piece: Faital M5N12-80
I measured one piece of them just yesterday, see attached graphs.
Measurement conditions: anechoic room, baffle 60x50 cm open (that's where the drop at 900 Hz comes from), measurement distance 50 cm (level corrected). The levels are real! So these graphs are really at 90 dB and 110 dB! Because of the open baffle, there's a little gain between 300 and 600 Hz, so the distortion are a little lower than in a closed box.
This driver has 94 dB/W efficiency (8 Ohm), and I would not cross lower than 300 Hz, if you want high SPL 400 Hz.
I measured one piece of them just yesterday, see attached graphs.
Measurement conditions: anechoic room, baffle 60x50 cm open (that's where the drop at 900 Hz comes from), measurement distance 50 cm (level corrected). The levels are real! So these graphs are really at 90 dB and 110 dB! Because of the open baffle, there's a little gain between 300 and 600 Hz, so the distortion are a little lower than in a closed box.
This driver has 94 dB/W efficiency (8 Ohm), and I would not cross lower than 300 Hz, if you want high SPL 400 Hz.
Attachments
Yes, I am.
It looks good from 270hz to 4k (i dont need super-high SPL, let's say 105dB for the transients).
One more to add to the list of small good drivers to cross higher than ideal.
Just a note:
http://www.faitalpro.com/products/schede/ps.php?id=101010100
to show how the real sensitivity is often much lower than the declared one. You lost a couple db in your OB baffle, but the behaviour across FR is pretty similar to your measurements, the peak at 5k is less pronounced. No way this driver is 99db, though.
Last edited:
I confirmed the 94 dB/W with a TSP measurement. Haven't compared the parameters in detail, so I don't know what causes the difference. The measured efficiency is still very good. Even more as this driver has a quite high impedance (I measured Re=7.5 Ohm, Faital gives 7.2 Ohm).
can you elaborate a bit about the 250 hertz cross over.
Im planning on crossing my FR at 300 hz, but you seem to think its not ideal...
I meant that published specs are usually too optimistics, very few manufacturers provide realistic data. Moreover, production tolerance (and burn-in!) has to be considered.
I can report two cases.
First, XO at 380hz, biamping was very critical and I dont want to go that high ever. (they were commercial speakers).
Second, XO at 300hz. My current speakers have this, which i raised after initial 250hz due to lack of dynamics otherwise. The XO has a very high steep (it's done in digital with serious software). Here the amp matching is much better but still far from perfect. For instance, I get a more uniform timbre and rhythm if I use similar amplifiers, it is not smooth when using for instance a class AB and a SE-class A.
For these new speakers, I want less compromises in this matter, because I know for sure that I will biamp low and MT section, and that I like amp rolling.
The voice range starts at 200hz* and finish at 4khz. (*some male fundamentals are even lower) As you read from my first post, a suitable driver which is also highly efficient (which i require to use class A amplification) is very hard to find. If you dont care as much for efficiency, but want a wideband to push XO boundaries out of the voice range, go for a Volt as midrange, they are the smoothest. Of if space/budget permits an array of 4-6 per side
I see, you are using a fullrange, well, think about adding a midwoofer as third way (300 down to 100 maybe). I decided against a FR, and I dont want to go 4-way (read my post about the ATC midrange), but maybe you can. Scan-speak, Seas, Volt... there are many good midbass units around 90db efficiency, or higher (PHL).
BUT if you dont plan to biamp your speakers, it doesnt really matter, you can do even 800hz XO and you wont notice the transition if it's tuned well. (You will notice if the loading is different though, like sealed box + OB top) Most commercial speakers are not meant to be multiamped and when they are, the predefined amplifiers have similar technology and timbre (i.e. Linn, Naim), or they are active (ATC, PMC...).
Hope this helps.
Not bad, better than the 6" beymas I have in list (and that now gets scraped).
Not powerful enough for 250hz XO, though.
Of the two compromises (crossing higher in the low-range or crossing lower in the upper range), i definitely prefer the second option, so i'm more inclined in getting a (pair of) midwoofer than a pure 6" mid.
Finding one in smaller size than 8" is very hard, though. Physics cant be cheated
Finished checking B&C catalogue. They have a nice 8" midwoofer:
B&C 8PE21
and their graph seem to match with Zaph measurements, so this one has the capabilities to play well down to 250hz, with a little active padding. It's FR is quite similar to the Jantzen 8008.
I'd like to see a CSD, though.
Last edited:
Little update on PHL drivers.
While i could not find FR of all current models, it appears that the midbass ones have depressed LF reproduction, which is not acceptable for me.
Instead the extended midrange 6,5" such as 1120/1130 is pretty smooth (see attached test results) and most importantly has been used cut at 250hz with only 1st order, although the upper range will require more shelving.
Two 1130 in parallel I think that could be cut at 200hz whithout sacrifying SPL too much, which is ideal for me.
Moreover they are much cheaper than the Supravox 165gmf (140 vs 240€).
This road seems more satisfying than the single 8" because will allow me to try a MTM configuration and reduce beaming.
I'm still looking for data of other 6,5" PHL models, in particular of the 1060/1070 (which should be more extended in the HF) and 1660/1670 (which should have stronger bass).
While i could not find FR of all current models, it appears that the midbass ones have depressed LF reproduction, which is not acceptable for me.
Instead the extended midrange 6,5" such as 1120/1130 is pretty smooth (see attached test results) and most importantly has been used cut at 250hz with only 1st order, although the upper range will require more shelving.
Two 1130 in parallel I think that could be cut at 200hz whithout sacrifying SPL too much, which is ideal for me.
Moreover they are much cheaper than the Supravox 165gmf (140 vs 240€).
This road seems more satisfying than the single 8" because will allow me to try a MTM configuration and reduce beaming.
I'm still looking for data of other 6,5" PHL models, in particular of the 1060/1070 (which should be more extended in the HF) and 1660/1670 (which should have stronger bass).
Attachments
Your constraints drive you to build speakers that inherently require you to build more speakers.
Large light diaphragms prone to beaming and break up modes, and selecting bandwidth based on what a human voice does instead of what the driver can do without beaming, or breaking up.
You mention steep slopes and "serious software" in #31, and then 1st order in #38.
Whole point of multi-way speakers is to get away from deficiencies of lightweight drivers necessitated by premiums for power amplifiers fifty years ago. And likewise DSP allows techniques to improve upon limitations of analog filters.
What do you use for speaker measurements?
Does Tempest have references to bone conduction studies? I've looked and don't see much. His approach to assessing bandwidth of hearing perception is bone headed. It is obvious from ability to discern location and movement of broadband sources in the forward direction, where timing changes of 5 microsecond in path length cause perceptual change. The inverse of 5 microseconds, 200kHz is not present, or detected in perceptual process. Hair cells on bassilar membrane fire at zero crossing of membrane from its rest position in one direction only(corelated with negative going pressure). Parellel neural networking builds phase descriminator.
Regards,
Andrew
Large light diaphragms prone to beaming and break up modes, and selecting bandwidth based on what a human voice does instead of what the driver can do without beaming, or breaking up.
You mention steep slopes and "serious software" in #31, and then 1st order in #38.
Whole point of multi-way speakers is to get away from deficiencies of lightweight drivers necessitated by premiums for power amplifiers fifty years ago. And likewise DSP allows techniques to improve upon limitations of analog filters.
What do you use for speaker measurements?
Does Tempest have references to bone conduction studies? I've looked and don't see much. His approach to assessing bandwidth of hearing perception is bone headed. It is obvious from ability to discern location and movement of broadband sources in the forward direction, where timing changes of 5 microsecond in path length cause perceptual change. The inverse of 5 microseconds, 200kHz is not present, or detected in perceptual process. Hair cells on bassilar membrane fire at zero crossing of membrane from its rest position in one direction only(corelated with negative going pressure). Parellel neural networking builds phase descriminator.
Regards,
Andrew
Digital xo (by means of izotope Ozone) was used in my last speakers but i cannot do it in these because of WAF limitation, the only thing I can do is a low pass and highpass between woofer and MT with an active electronic xo such as Marchand.
Said that, I have been satisfied by passive multi-way speakers, so I do not have a prejudice on this matter.
The HF must be powered by a class a push-pull, no matter what you or others think of the improvements in amplifier technology. No class AB amp that I know of sounds good in the top end. If I can limit its power, the better, as I don't want to build a 100kg brick (but I have listened to one which sounded great). And don't even suggest pwm.
I build speakers that follow MY tastes, this is a DIY audio forum, isn't it?
Otherwise there are hundreds of designs I could replicate, 99% made more or less the same. Then there are fullrange monodrivers with or without reinforcement. BDDT.
Ofc I want to use the various drivers in their best range. But I want to employ them in a design that suits my needs and tastes.
I can end with a 2khz xo like Tony Gee did in a big 4-way, or maybe higher. I do not know until i have the drivers in my hands. but I'd like the midrange speakers be able to reproduce the full voice spectrum, that's all.
Said that, I have been satisfied by passive multi-way speakers, so I do not have a prejudice on this matter.
The HF must be powered by a class a push-pull, no matter what you or others think of the improvements in amplifier technology. No class AB amp that I know of sounds good in the top end. If I can limit its power, the better, as I don't want to build a 100kg brick (but I have listened to one which sounded great). And don't even suggest pwm.
I build speakers that follow MY tastes, this is a DIY audio forum, isn't it?
Otherwise there are hundreds of designs I could replicate, 99% made more or less the same. Then there are fullrange monodrivers with or without reinforcement. BDDT.
Ofc I want to use the various drivers in their best range. But I want to employ them in a design that suits my needs and tastes.
I can end with a 2khz xo like Tony Gee did in a big 4-way, or maybe higher. I do not know until i have the drivers in my hands. but I'd like the midrange speakers be able to reproduce the full voice spectrum, that's all.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.