I mentioned many posts ago that I found comparable B with the same magnet setup in both the 3 and 6 gap version that I was simulating at the time. I think this was due to the double stack large-magnet-in-the-pole piece set up which I feel is far from optimal (but of course we have our constraints to consider).
To gain a general idea I enquired about the cost of some N45 24mm OD x 20mm magnets. The approx cost of £4.50 each means that we could get a lot more surface area (and thus eventually B) for a fraction of the price (consider 16 smallish mags used in the Parthenon).
Anyway, my hunch is that the phenomenon encountered by both Bill and I (if only my name were Ted) was mearly the result of non optimal magnet use in the first place.
Some more food for thought... For reference, I got a quote for 200mm long 25mm dia. 1008 steel rods (x20) and 25mm thick 200mm OD plates (x4). This worked out to approx. £130. Add 16 magnets mentioned above and you got the DIY beginnings of what looks like a Parthenon - and this for chump change comparatively speaking. Yes yes, I know about the ring magnets which cost a whole lot more, and the screw fixings/machining bla bla bla, but hopefully it gives a ball park figure to what we can achieve for the motor (I for one haven't the faintest what industrial "junk-yard" steel tubing costs?)
Steven, that link is a much appreciated resource. ("More input" - Johnny Five). You really can learn a lot from reading Dan's posts.
Bill, the non-flat BL curve was due to the inconsistent B between gaps right? I think this is somewhat resolved by cutting grooves in the pole piece. The B between gaps becomes a lot closer together IIRC. I think Dan mentions something related to it a few posts back, but I assume you left it out to reduce machining
It's not the size that matters it's what you do with it!
When else can you use that line???
I don't know if anyone has addressed this idea completly but I think that Dan eluded to the idea that a distribution of a number of smaller magnets with give a more evenly saturated piecs of metal. When I look at the wonderfull diagrams everyone seems stuck on a massive chunk of neo that will cost a fortune!!
Many small ones will probably work better and will definitly be cheaper.
Mark
When else can you use that line???
I don't know if anyone has addressed this idea completly but I think that Dan eluded to the idea that a distribution of a number of smaller magnets with give a more evenly saturated piecs of metal. When I look at the wonderfull diagrams everyone seems stuck on a massive chunk of neo that will cost a fortune!!
Many small ones will probably work better and will definitly be cheaper.
Mark
Dan--Yer right as rain. My off-the-cuffing was a tad paradoxical.
So, from your POV, pretending for a moment that we had an imaginary motor with a knob on it to turn BL up and down, should we crank the BL up and EQ the roll-off, or should we crank it down for a high Qts to induce a bump at Fs? We want efficiency and extension, so what's a good compromise Q in your opinion/experience?
Thanks for info on complimentary curves--makes sense.
Verdict...?
🙂Vikash,
I believe the pole channels are intended primarily to reduce field spreading between the plate and pole. If identically sized, I believe they have very little effect on flux strength variation from gap to gap. I suspect that has much more to do with the longer path length of the outer circuits. You could easily check by adding pole channels to the FEMM model I sent you, as you could directly compare it to the plots I posted earlier.
Dan,
This reminds me: I never really did figure out the importance of the pole channels, since I've been able to model flat multi-gap BL curves without them. I also still wonder abou the Le variations they introduce with excursion compared to a standard pole. Can you share the thinking behind them?
Many magnets versus one
Vikash, Mark, All,
I fully agree we should probably forsake the magnet-in-pole idea. See my last posted picture for my newest thinking on the matter.
Also, it occurred to me that stacking the rectangular tubing/plates like I've illustrated could at least partially correct the flux difference between gaps since the the magnets are now directly connected to the longest (outer) circuit element, and seperated by one and two slightly impeding plate boundaries respectively for the middle and bottom elements. Might be time for another model...
Vikash, Mark, All,
I fully agree we should probably forsake the magnet-in-pole idea. See my last posted picture for my newest thinking on the matter.
Also, it occurred to me that stacking the rectangular tubing/plates like I've illustrated could at least partially correct the flux difference between gaps since the the magnets are now directly connected to the longest (outer) circuit element, and seperated by one and two slightly impeding plate boundaries respectively for the middle and bottom elements. Might be time for another model...
I will, the next time I have them unbolted... which may be soon.Bill F. said:
They are currently inside of their little PVC housings, inside of "project servo", so I can't see their numbers.
I need a little motivation to get back working on that project again, anyway. 😎
My official vote is:
A) Use a single diaphragm, because while we can modify a dual-diphragm to a single-diaphragm, we can't, without compromising all our parameters.
Conversely, if we built it with the frame, and the sweep-walls (the "XMX-style" welded steel chassis), there's nothing preventing us from using it in a dipole-application.
B) Another benefit of the welded steel chassis and sweep walls.. attaching a spider front and rear becomes trivially easy. Front and rear of the sweep chassis... just attach. 😎
C) I personally wouldn't bother, as our primary benefit is reduced inductance - do we care? We're running this thing as a subwoofer... and with a ridiculously low Fs, at that.
It seems the more economical approach is to go straight-XBL^2 with it... which we've already got permission for.
Just my personal votes.
I'm not Dan, but I understand the grooves in the pole piece serve the same purpose as the grooves in the top plate... helping create three, distinct gaps.Bill F. said:
Think of them as serving the same role of having a T-shaped (undercut) pole piece on a standard motor... it focuses the flux right at the gap, rather than spreading the flux from the top plate's gap, to some larger adjacent area on a plain, smooth pole piece.
Yeah, that's actually what I meant by saying they prevent the field from spreading. They keep the flux zones more rectangular, while a standard pole allows them to be trapezoidal.
However, I'm wondering if the coil cares about the shape of the flux zone, as long as the sum of the flux lines crossing the coil remains roughly constant (assuming the flux average of the two topolgies is identical).
In other words, if I can get a standard pole to yeild a flat BL curve, what is the point of the pole channels? Is there some other performance parameter that they improve, and is it worth the added expense and tradeoff of (I assume) the increased Le fluctuation WRT excursion?
BTW, would you like a primer in the use of lua scripts? I printed one up for a guy I'm consulting for, and I could send you a copy if you like.
However, I'm wondering if the coil cares about the shape of the flux zone, as long as the sum of the flux lines crossing the coil remains roughly constant (assuming the flux average of the two topolgies is identical).
In other words, if I can get a standard pole to yeild a flat BL curve, what is the point of the pole channels? Is there some other performance parameter that they improve, and is it worth the added expense and tradeoff of (I assume) the increased Le fluctuation WRT excursion?
BTW, would you like a primer in the use of lua scripts? I printed one up for a guy I'm consulting for, and I could send you a copy if you like.
"T" shaped poles
They are primarily designed to focus the flux in a given area much like a lense focuses light. The more lines of flux across a wire the more potential motive force. ( Something that you knew allready but had to think about out loud 🙂 )
By the way great work Bill. It will fly and will be cool to build.
I did a search on the net for a program I used way back when for reverse engineering drivers. It's called the Driver parameter Calculator or DPC and I also found something called finemotor which appears to be available as a trial ( don't know for sure but I'll check when I get home tonight )
Mark
They are primarily designed to focus the flux in a given area much like a lense focuses light. The more lines of flux across a wire the more potential motive force. ( Something that you knew allready but had to think about out loud 🙂 )
By the way great work Bill. It will fly and will be cool to build.
I did a search on the net for a program I used way back when for reverse engineering drivers. It's called the Driver parameter Calculator or DPC and I also found something called finemotor which appears to be available as a trial ( don't know for sure but I'll check when I get home tonight )
Mark
I get that, but AFAIK, the pole channels don't increase the total flux across the coil, they just redistribute it. So if they don't noticeably impact BL flatness or total average flux, and if they take a toll in cost and Le modulaton, then why why WHY?? There's gotta be a good answer out there, and I'm gonna pitch a fit if I don't get it soon!
😉 Yeah, Mark, I'm excited about how this project is shaping up too. Seems like we might have something soon for you to start fabbing! It's gonna be very rewarding to break the proverbial champaign bottle across the bow of a penny-ante design that packs megabuck performance.
So... have we settled on an excursion target yet? If 6" p-p is sufficient, I'll take a look at available rectangular/square tubing sizes and costs, as well as flat plate stock cost to make a comparison.
Before we can get a grip on moving mass, we need to settle the one/two diaphram issue (the former being heavier) as well as decide on a VC diameter (now that it doesn't need to accomodate a pole magnet).
I think diaphram target size should be based off of desired performance in dipole applications... set our goals and see if we can meet them. "Reference level" down to, what... 15Hz? 20Hz? And do we want to take a multichannel reference level requirement of some 121dB? Adjust for nominal seating distance and say 125dB @ 20Hz or so? Other thoughts?
I can't seem to download a functional version of Linkwitz's "spl_max.xls" spreadsheet here at work, so I'm not sure what swept area would be required to produce the above performance figures with 6" p-p excursion. I'll take a look at that later today if no one gets to it before then.
Before we can get a grip on moving mass, we need to settle the one/two diaphram issue (the former being heavier) as well as decide on a VC diameter (now that it doesn't need to accomodate a pole magnet).
I think diaphram target size should be based off of desired performance in dipole applications... set our goals and see if we can meet them. "Reference level" down to, what... 15Hz? 20Hz? And do we want to take a multichannel reference level requirement of some 121dB? Adjust for nominal seating distance and say 125dB @ 20Hz or so? Other thoughts?
I can't seem to download a functional version of Linkwitz's "spl_max.xls" spreadsheet here at work, so I'm not sure what swept area would be required to produce the above performance figures with 6" p-p excursion. I'll take a look at that later today if no one gets to it before then.
It's my interpretation that a squared up field acts upon more of the coil vs a more pointed one typical of a non channeled pole. (But I hear what you're saying loud and clear Bill)
Dan, please put our minds at rest... 😴
Bill, Please email me the lua primer. Much appreciated.
Dan, please put our minds at rest... 😴
Bill, Please email me the lua primer. Much appreciated.
Right now, I'm writing an email to a forum member whose grasp of acoustic theory I've grown to admire. I'm hoping he'll chime in on the performance characteristics of a single OB diaphragm vs. two spaced ones.
so stay tuned on that point...🙂
Using Linkwitz's max_spl.xls spreadsheet and playing with path length (which I've taken as the diaphram "radius" for the moment), I get somewhere in the neighborhood of a 48" square diaphram to give 125dB output @ 20 Hz with a 6" excursion.
I need to take a look back at dipole models though and see what the path length of a raw driver with no baffle should be. It may well be effectively 2R (where R is the radius).
I need to take a look back at dipole models though and see what the path length of a raw driver with no baffle should be. It may well be effectively 2R (where R is the radius).
To get a handle on diaphram weight, after we have an approximate size, we have to set some acceptable limits on diaphram deflection. The acceleration for any given SPL and frequency can be easily enough calculated, and I'd be happy to model a diaphram under such a body load to see what deflections we have. These deflections will cause distortion, though perhaps we aren't that concerned? I'll also take a look at the diaphram modes to make sure they are well above the sub's operating range.
Hi all,
- As far as the two overlapping B curves of the DDD goes, it is the sum that matters from a BL-based standpoint, so if the sum is flat you're good to go.
Of course, placement/alignment of the voice coils is critical to get any decent excursion out of them. For 3" one way, the two sets of gaps must be at least 3" apart, and the voice coils even further (ideally you want more than 3" seperation of the gaps, because of fringe field).
The DDD approach is really quite a good idea, but it is cumbersome to build in production, because of the requirement for long formers, tall motors (widely spaced gaps), etc.
- Notches in the core are actually quite beneficial; Vikash is correct in that the square shape of the gaps is rather important. It does lots of good things for the BL curve, but that's all the more I'm going to say...😉
- Inductance variance from cuts in the core aren't really a problem; in fact, it helps lower the total inductance because the average distance of the core to the voice coil in increased, and that's one way to lower inductance (move steel away from the coil).
- For a dipole driver "in the buff", the baffle size is 2R radius (reference Fundamentals of Acoustics).
Dan Wiggins
Adire Audio
- As far as the two overlapping B curves of the DDD goes, it is the sum that matters from a BL-based standpoint, so if the sum is flat you're good to go.
Of course, placement/alignment of the voice coils is critical to get any decent excursion out of them. For 3" one way, the two sets of gaps must be at least 3" apart, and the voice coils even further (ideally you want more than 3" seperation of the gaps, because of fringe field).
The DDD approach is really quite a good idea, but it is cumbersome to build in production, because of the requirement for long formers, tall motors (widely spaced gaps), etc.
- Notches in the core are actually quite beneficial; Vikash is correct in that the square shape of the gaps is rather important. It does lots of good things for the BL curve, but that's all the more I'm going to say...😉
- Inductance variance from cuts in the core aren't really a problem; in fact, it helps lower the total inductance because the average distance of the core to the voice coil in increased, and that's one way to lower inductance (move steel away from the coil).
- For a dipole driver "in the buff", the baffle size is 2R radius (reference Fundamentals of Acoustics).
Dan Wiggins
Adire Audio
Thanks, Dan.
I was pretty confident about the BL summing question, but I like to defer to longer longbeards when I can. And I never forget there's always a chance I could be wrong...
Uh-oh, another mystery! With this single teasing hint, you've raised the stakes to a metaphysical level. Now we will absolutely have to include pole notches as a spiritual exercise, so the gods of the B field will grant us untold benefits...
Just teasing, of course. You have a right to withhold whatever info you want. It's good marketing, anyway--people need a little mystery to worship. 😉
Since it seems to me that the inductance peak would occur when the coil moves from the notched section of the pole to the unbroken section below the notches, I guess you could eliminate it simply by adding another notch below the bottom gap...
RH,
If you want to simulate and compare two diaphragms with one, until we hear different, I suggest you simply specify the baffle radius as 1/2 the path length from the center of the front diaphragm to the center of the rear one. You're probably already doing this anyway...🙂 I'll be curious to see what you come up with. I'd do some sims myself, but it's just starting to be crunch time here at the office.
I was pretty confident about the BL summing question, but I like to defer to longer longbeards when I can. And I never forget there's always a chance I could be wrong...
Uh-oh, another mystery! With this single teasing hint, you've raised the stakes to a metaphysical level. Now we will absolutely have to include pole notches as a spiritual exercise, so the gods of the B field will grant us untold benefits...
Just teasing, of course. You have a right to withhold whatever info you want. It's good marketing, anyway--people need a little mystery to worship. 😉
Since it seems to me that the inductance peak would occur when the coil moves from the notched section of the pole to the unbroken section below the notches, I guess you could eliminate it simply by adding another notch below the bottom gap...
RH,
If you want to simulate and compare two diaphragms with one, until we hear different, I suggest you simply specify the baffle radius as 1/2 the path length from the center of the front diaphragm to the center of the rear one. You're probably already doing this anyway...🙂 I'll be curious to see what you come up with. I'd do some sims myself, but it's just starting to be crunch time here at the office.
Actually, if a single diaphram dipole path length is 2R (thanks Dan... that was what I was beginning to suspect) then a spaced pair should be 2R+L where L is the spacing between them. I'll check this afternoon and see what impact a reasonable L has on excursion/Sd requirements.
Let's see... Taking the longest motor design we'd have 6" of gaps on each side of a DDD motor + 4" or so separating them + 3" clearance between each spider and the nearest gap + 3" between each diaphram and spider. Hmm... 6+6+4+3+3+3+3=28" spacing betwen diaphrams.
😀
Shortest design would use a single sided motor... 6" gaps + ~1" back plate + 3" spider/gap clearance + 3" diaphram/spider/backplate clearance. So perhaps 16" diaphram spacing on the short end. I think dual diaphrams goes hand in hand with the DDD topology... single sided I'd probably just stick to a single diaphram for a more "traditional" motor.
Bill... do the SPL targets sound reasonable to you?
Let's see... Taking the longest motor design we'd have 6" of gaps on each side of a DDD motor + 4" or so separating them + 3" clearance between each spider and the nearest gap + 3" between each diaphram and spider. Hmm... 6+6+4+3+3+3+3=28" spacing betwen diaphrams.
😀 Shortest design would use a single sided motor... 6" gaps + ~1" back plate + 3" spider/gap clearance + 3" diaphram/spider/backplate clearance. So perhaps 16" diaphram spacing on the short end. I think dual diaphrams goes hand in hand with the DDD topology... single sided I'd probably just stick to a single diaphram for a more "traditional" motor.
Bill... do the SPL targets sound reasonable to you?
121dB @ 20 Hz would be monumental! 😱 But heck, let's go ahead and shoot for the moon. Even if we miss that number, we'll still have an honest-to-God sonic weapon. Of course, we prolly oughtta set a budget pretty soon--that'll be the limiting factor.
OK, I think this might be an attainable goal. Here are my results using a modified version of the Linkwitz dipole model.
To reach a target of 125 dB @ 20 Hz @ 1 meter in free space with no baffle would require a diaphram of the following sizes, assuming a maximum peak-peak excursion of 6 inches:
(1) 34" x 34" for a single diaphram
(2) 29.5" x 29.5" for two diaphrams spaced 16"
(3) 27" x 27" for two diaphrams spaced 28"
My SWAG is that a single 34" square diaphram would be nearly twice the weight of a pair of 27" diaphrams of the same stiffness. However, perhaps the BL to move a larger diaphram won't be a problem.
Looks like the diaphram sizes required to achieve the performance target with 6" excursions are "reasonable." Let's settle on an arrangement and size and I'll run some basic structural FEA "what if's" when I get the chance.
To reach a target of 125 dB @ 20 Hz @ 1 meter in free space with no baffle would require a diaphram of the following sizes, assuming a maximum peak-peak excursion of 6 inches:
(1) 34" x 34" for a single diaphram
(2) 29.5" x 29.5" for two diaphrams spaced 16"
(3) 27" x 27" for two diaphrams spaced 28"
My SWAG is that a single 34" square diaphram would be nearly twice the weight of a pair of 27" diaphrams of the same stiffness. However, perhaps the BL to move a larger diaphram won't be a problem.
Looks like the diaphram sizes required to achieve the performance target with 6" excursions are "reasonable." Let's settle on an arrangement and size and I'll run some basic structural FEA "what if's" when I get the chance.
- Status
- Not open for further replies.