|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
173.52.247.27
Found these drivers at Zalytron- focal 7w4411 and the high frequency dropoff at 3khz is like a second order crossover. The tweeter can be a ribbon tweeter like the Hi-Vi RT1C whose specs are at madisound:
https://www.madisound.com/store/product_info.php?cPath=45_229_236&products_id=1486
This ribbon drops off at 3.5khz and has great off axis dispersion as does the midrange. The midrange has a 10 ohm nominal impedence making it perfect doubling as an mtm with a cap and potentiometer on the tweeter to control high frequency energy, it seems a perfect match and able to run without a crossover. As a sealed or even vented unit the mids should give tight bass down to 50 hz, covering the entire vocal range without the fog of a blocking cap and inductor.
Any builders out there want to put in their 2cents on whether they forsee problems with the high 3-3.5khz crossing point in an mtm rather than tmm format? Also, may be better to put in an acoustic suspension enclosure to get a quick drop off of the bass to easily be supplemented with a sub.
Follow Ups:
Even this crossoverless speaker will have 2nd order acoustic rolloffs that would require the tweeter to be wired reverse polarity if it meets the criterion for an LR2 design. Crossover or no crossover it's the final acoustic rolloffs that matter, and how they sum in the crossover region. You may as well add 2nd order electric filters to the natural acoustic rolloffs and aim for a 4th order acoustic design. Sure, the group delay is not as good, but the drivers polarities are the same, and the response is in phase at all frequencies. You just don't get perfect impulse response - but neither 2nd or 4th order have perfect impulse response, save for the 2nd order transient perfect design but that's a whole other enchilada. Also, you need to realize that even if both drivers have a 2nd order rolloff, they also need to have the same corner frequency if they are to sum correctly. If their corner frequencies differ, you'll either get a big hump or a big dip in the response. There is more to it than just finding drivers with good looking natural rolloffs. Further to that, all conventional tweeters have a natural resonant frequency. Even WITH crossovers, if you cross too close to this frequency or don't cross steep enough, the effects can be far from subtle. And I mean ranging from "too analytical" sounding to downright offensive. Tweeter Fs considerations and often impedance bump and rise compensation are a big part of the design process. Last but probably most important is distortion. When you try and run devices like tweeters full range and just let them roll off "naturally" you're going to be operating almost ever tweeter in existence (if not all) in a region where distortion is going to skyrocket. Not to mention with a ribbon, as you go lower in frequency, you also go lower in AC impedance. Ribbon transducers use transformers, which will result in a near dead-short when you try to apply a full range signal to them. This can be seen on their impedance plots IF the manufacturer wants to show you the impedance down low in the first place. Most do not - they don't want to advertise this feature of the transducers. So, even if you DID find a ribbon transducer that could handle full range, you'd still need at least ONE CAP (1st order electric filter) to protect not the tweeter per se, but the amp driving it!This "tweeter" is the closest thing on the planet to a true ribbon full range, considering electrostats are not, by definition, true ribbon transducers.
http://www.orcadesign.com/products/raven/ravenr3.htm#
In a nutshell, tweeter low end rolloff, by itself, is just not usable like the rolloff of certain well-behaved midbass drivers. Some designs DO have no crossover on the midbass but still use some sort of filtering and equalization (passive) to get the tweeter acoustic response to align with the midbass natural rolloff.
Read this:
http://www.musicanddesign.com/Speaker_Transient_Analysis.html
John Kreskovsky has written a lot of stuff about transient-perfect crossovers such as subtractive or filler-driver and quasi-transient perfect designs as well.
Another path is digital. You can use the Thuneau Allocator and Thuneau Phase Arbitrator to essentially eliminate the group delay of a 4th order LR crossover using forward-reverse processing. How this works is that digital filters are used to mangle phase in the exact opposite way that the desired acoustic response will mangle phase. Then, the speakers don't mangle phase, they actually unmangle the mangle, resulting in near-perfect group delay and a perfect impulse response (on the chosen design axis of course). And you still need to eq the drivers flat in their passbands and match the target acoustic response curves for the chosen fitler types. Think of it as "real time phase compensation" and it works. And it's not a "room correction / impulse EQ" Measurement is mandatory here. "Published" FRD data, bode plots, frequency response, impedance plots, etc. are generally good for theoretical work only - not really useful in the actual design process.
Go here: www.thuneau.com
But don't forget you're also going to need a measurement suite (software program) Speakerworkshop is free but a b*stard to figure out how to use. For a beginner I would high recommend something else. Soundeasy is the definitive "DIYers choice" for speaker design using passive crossovers, but still has a steep learning curve. ETF is good for measurements for active systems where you don't need to model impedance and passive elements like you do with passive designs. ETF can be used for passive designs too, but you would need a separate passive crossover design suite. Soundeasy is both a measurement suite and a crossover development tool.
I am currently working on a WMTMW with all digital crossovers right now that will be transient accurate, with no phase distortion. Essentially, a speaker system that is both time and phase coherent. The big trick with speaker design is not just getting a flat on axis-response, but understanding and being able to implement concepts of power response, dispersion, preventing lobing, and baffle step compensation are all pivotal in getting a speaker that sounds good. Making pretty graphs is easy. Making a great sounding speaker that also happens to conform to design ideals is not easy - and much experience is required to get the correct measurements, but more importantly, being able to interpret what you have measured.
If you do go the digital/active route be prepared to spend $250 - $500 for an entry level audio interface and as much as $2k or $3k if you want the best that's out there that will rival most high end components. You'd be surprised, however, what a decent $250 or $300 audio interface can do in the right hands.
No offense, but you've vastly over simplified the design process here. Your idea is neat in theory but in practice faces far too many barriers to be possible with real-world drivers at sensible listening levels. Also, 3-3.5K is quite high to cross over to a 7" midbass. Driver center spacing would indicate a much lower frequency, such as 1.8 - 2.5k, else you're going to run into lobing problems. Lobing is a necessary evil for ribbon designs that require high crossover points combined with the large center spacing distance of a large ribbon unit used with a larger midbass. You see a few 7"/ribbon designs out there and really, they start out with a few serious design drawbacks which might explain why they are not all that popular and not the be-all end-all. Getting a seamless integration between a dynamic 7" driver and a ribbon transducer is a very very tall order. Often, the drivers that are "fast" enough to keep up with a ribbon are not well known for low frequency extension. You need mass to get low Fs and low bass - and light quick drivers can't have a lot of mass.
Cheers,
Presto
Edits: 11/30/10
I mean, everything you said is true, but if somebody had told you or me, when we were just starting out, that "you can't POSSIBLY build a decent-sounding speaker without understanding all the advanced theory and spending thousands of dollars on measurement equipment and design software," would we ever have built ANYTHING?
A noob needs to start somewhere, and build a simple project or two WHILE he is learning. This guy wants to try a project with NO reactive components in series with the woofer, and a highpassed tweeter. There are MANY commercial designs like that out there, mostly mass-market junk, but a few with high-end pretensions, that even pander in their advertising to the superstitious fear that low-pass and response-shaping filters on the midbass driver somehow compromise sound quality. The baby Paradigms use this design approach, and sound OK for what they are. Let him build something with decent and potentially compatible drivers and textbook filters, hear for himself how it sounds, and maybe work up a "real" crossover for it down the road.
To address a few of your points:
"Even this crossoverless speaker will have 2nd order acoustic rolloffs that would require the tweeter to be wired reverse polarity if it meets the criterion for an LR2 design. Crossover or no crossover it's the final acoustic rolloffs that matter, and how they sum in the crossover region. You may as well add 2nd order electric filters to the natural acoustic rolloffs and aim for a 4th order acoustic design. Sure, the group delay is not as good, but the drivers polarities are the same, and the response is in phase at all frequencies. "
Absolutely true. But MANY well-reviewed "high-end" speakers have the tweeter and mid wired in reverse polarity, in order to sum flat through the crossover region -- lookit the past couple years of speaker reviews in Stereophile. If they also exhibit a deep null at crossover frequency when the polarity on the tweeter is switched to "normal," this means the drivers are in fact tracking in good phase with the reverse polarity connection. So what if they're out of phase two octaves either side of crossover? Is this truly audible? Even a single, unfiltered, fullrange driver "rotates" phase over its operating range, and will be out of phase with itself at some frequencies relative to others.
"You just don't get perfect impulse response - but neither 2nd or 4th order have perfect impulse response, save for the 2nd order transient perfect design but that's a whole other enchilada."
Yes. But very few speakers on earth have perfect impulse response. Maybe some fullrange planars, and some first-order, time-aligned designs like Thiel, Vandersteen, or Dunlavy approach it, but even then only on a VERY narrowly defined design axis. This does not seem to be a critical issue for most listeners to most speakers, and really shouldn't concern a noob.
"3-3.5K is quite high to cross over to a 7" midbass. Driver center spacing would indicate a much lower frequency, such as 1.8 - 2.5k, else you're going to run into lobing problems. Lobing is a necessary evil for ribbon designs that require high crossover points combined with the large center spacing distance of a large ribbon unit used with a larger midbass."
Well, lobing happens anytime the drivers are vertically/horizontally offset from one another -- only coaxials avoid it. Crossover frequency, order, and Q, along with driver spacing, can make it better or worse. If the primary response lobe is narrow, that means there's a narrowly defined vertical listening axis where you will hear the drivers sum properly. This may be more or less disturbing, depending on your listening habits.
As for tweeters -- all dome/cone types have an impedance bump at fs. This is somewhat mitigated if ferrofluid is used in the voice coil gap. The shunt resistance of an L-pad will mitigate it further, often to the point where if you cross at least second order electrical, at least an octave above fs, it causes at most a little ripple in the response in the stop band. If the impedance bump is severe, a conjugate filter may be needed to flatten it. Planar magnetic types (like the Bohlender Graebner "ribbons"), however, have an essentially flat impedance. Dunno about true ribbons, I've always been scared to mess with them.
I guess my point is that a noob has to start somewhere simple, and learn about the advanced stuff as he goes. My recommendations would be:
1) Read Ray Alden, and grasp the basics.
2) Build a proven, and well-documented design, like from Zaph Audio, and study the extensive documentation to understand why he did what he did.
3) Get some basic crossover design/ simulation software. Jeff Bagby's Passive Crossover Designer is free for download, as are a number of compatible and mutually supportive programs from FRD consortium. You can import published driver specs (which you HOPE are accurate) into the software, and demonstrate graphically the effects of various filters on the driver responses. If you run into problems, ask for help at PE Tech Talk -- people who know the software very well will be happy to help.
4) If you get really serious, invest in some basic test gear. The WT3 from PE will provide impedance curves and TS parameters on drivers. PE is apparently offering a new acoustical testing rig that includes a calibrated mic and RTA/FFTA software, so you can do quasi-anechoic response curves of actual drivers mounted in actual boxes. All of the above (not counting the computer to run them on) will set you back less than $400.
The proposed high mid/tweeter crosspoint might not be too bad IF you listen on-axis and IF your room is large/dead enough that the reflected off-axis response is significantly lower in level than the on-axis. I've built speakers intended for nearfield listening with quite a high crosspoint (6kHz) and they were OK in this regard. Do try to keep the center-to-center distance between tweeter and mid under one wavelength at crossover to reduce vertical lobing. (An MTM will have comb-filtering issues, due to the separation between the mids.
You WILL need to highpass the tweeter, even if you run the mid full range. Look at where the mid is down -6dB from its "average" sensitivity and make this your crossover point. Devise a second-order highpass filter for the tweeter that is also down -6dB at this frequency. You will definitely need a resistive L-pad on the tweeter to set its output level.
Unless the speakers are to be placed right against a wall, you have to consider the baffle diffraction step (look it up if you're not familiar with the concept). Since you are considering using two mids in an MTM or TMM, you can do a 2.5 way arrangement. Run one of them fullrange up to where it meets the tweeter, and lowpass the other to fill in the low frequencies. The low fill driver wants to be down -3dB where frequency = approximately three baffle widths. Or just divide your baffle width in inches into 4560 to get the approximate frequency. A series inductor plus impedance compensation (Zobel network) on the .5 way driver should suffice. OR you can build the cabinet as a bipole, with both mids wired in parallel and in phase, and one of them firing to the rear. Both could then be run full range, with no crossover components on either, and still take care of the baffle step. Bipoles need a lot of air around them (no near-wall placement) but can throw a huge soundstage.
There's more to it than this, of course, but this can get you started. Post (or PM me) your intended baffle dimensions and a link to the response curves of the Focal drivers, and I can crunch some numbers for you.
Recommended reading: Ray Alden, Speaker Building 201. This book has the clearest explanations of the basic concepts. Welcome to the addiction -- er, "hobby."
Thank you, thank you thank you. Its hard to find a lot of information of use although there is a lot of "information" online. I will look up the references you gave. I will continue to search for drivers for a good crossoverless system but have no taste for the Lowther style drivers that, to me, beam terribly and don't deliver weight in the lower frequencies given their limited excursion. Based on what I've seen so far, if I want a higher bandwidth covered by a single driver, I have to use a smaller driver to keep its center closer to the tweeter, or use a crossover with a lower frequency. Its all a list of uncomfortable compromises.
C.
That particular Focal looks like a VERY nice driver, but perhaps not the best suited for your intended purpose. The on-axis response peak right below rolloff definitely needs to be addressed, or it will sound "shouty" in the midrange. I would just notch it out with an LRC trap filter, but that's a no-go if you want to avoid any reactive components in series with the driver. Since it looks real good 30 degrees off axis, it could be mounted in a box with a "bicycle headlight" type tweeter (HiVi, Morel, and others make such tweeters) swivel-mounted on top of the , so you could have the woofers firing straignt ahead and the tweeters tilted in toward the listening position.
In the past, Peerless, Vifa, and others have offered polypropylene cone drivers that don't have that response spike before rolloff -- just ruler flat up to 6kHz or so, and then a smooth second order rolloff. Check current offerings from Madisound. Something like that could be the ticket.
Too many of the "fullrange" drivers out there have low efficiency, low power handling, and awful frequency response. Some of the Fostex models, and possibly the new 6" and 8" Dayton fullrangers from Parts express, are less compromised in this regard. Madisound offers several kit designs with the Fostex models that use back horn (or TL) loading to fill out the bass response, and a supertweeter mounted up top for air and sparkle.
HF beaming is as much a function of baffle width as driver design. Figure that with ANY driver mounted on a baffle of finite width, away from room boundaries, it will be radiating omnidirectionally where wavelength equals or exceeds 8 baffle widths, and its on-axis response will be down -6dB from its "rated" sensitivity. Where wavelength equals 3 baffle widths, on-axis response will be down =3dB from rated sensitivity, and where wavelength equals 1/2 baffle width or less, it will be beaming straight ahead at full rated sensitivity. This is the "baffle step" pehenomenon, which must be addressed unless the speaker is in or right against a room wall. To compensate for it, you will always be throwing away 6dB of sensitivity relative to the rated sensitivity of the driver. A 2.5 way design ADDS 6dB down low, where it is being lost off-axis, so the "upper" midbass driver can operate at full rated sensitivity up to where it meets the tweeter.
You'll probably get lots more good advice over at the PE Tech Talk forum than here, since it's ALL about DIY. Folks there are knowledgeable, friendly, and helpful. Check it out.
Without a xover a ribbon tweeter will die almost immediately, even 6dB slope xovers are not steep enough for most.
If it does not fry any bass/mid fed to it will stretch the ribbon and make it equally useless very soon.
I hear you. I expect to have to use something on the tweeter, and it may be that ribbon tweeters are wrong for this type of project. Maybe a supertweeter like a fostex horn will work better and also have a sharp dropoff below 3.5khz. If the crossover frequency goes down to where the ear is more sensative, I will have to either use a regular crossover or go to an active crossover and biamp. Difficult to design a really good first order crossover and difficult to match dissimilar drivers and amplification in a coherent point source like package.
I may need to look at smaller mids if I want to integrate with a higher frequency cutoff on the tweeter.
The peak in the frequency response at about 2.8kHz might be objectionable. Also, you'd have to figure dispersion at the highest frequencies for the Focal would be limited. Crossing any 7-8" driver at 3.5kHz isn't recommended. However, to minimize components in the signal path, it might be interesting to measure the response with a single inductor in series with the woofer. Some impedance correction might also help.
The response curves posted at zalytron on this midrange show its 30 degree off axis response to be impressively flat to cutoff. A lone inductor may tame the high frequency peak but to build a network to correct impedence takes us further down the road to what I was hoping to avoid, namely a complex network getting in the way of the music. Inductors, to my ears, are less offensive than capacitors in this regard. Two mids in close proximity would produce interference at frequencies an octave below 2.8khz so I guess It would not flatten out such a high frequency bump. Its difficult but I would like to find a way to cover all frequencies that the ear can distinguish direction with a single or dual crosoverless drivers and fill in high and low frequencies as needed. The sharp dropoff on the top end of this mid makes it a good candidate for no - or minimal crossover, I would think.
To my ears inductors are hugely more offensive than capacitors.
I only got to hear bass with real detail since I went active and thus eliminated the almost inevitable series inductor before the woofer.
BTW I use a ribbon supertweeter actively and it still needs a high pass to protect it from potential amplifier switch-on pops which would kill them in an instant. The hp is more than an octave below its active pass band and so its audible effect is minimized.
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: