Terry Cain's BIB -why does it work and does anyone have those Fostex Craft Handbooks?

OK guys, thinking hats on, because, since the idea's recently come up again, I think it's about time we revisited this design and figured out why this monster TQWT simulates so poorly, yet usually performs well in-room -far better than conventionally accepted wisdom would suggest. I admit I have a liking for this box -I built a rough pair in response to a challenge from Terry a few months ago using the FE166E, and they worked pretty well -in fact, astonishingly so, considering all the apparant anomolies. We're looking at a driver ill-suited for TL loading with a low Q, a pipe with an So of 0 (so a theoretically high F3), and open at the other end, which isn't going to provide much any damping, and a driver position that seemingly dictates a massive hole circa 100Hz.

But it's never as bad as that. There's usually a dip at 100Hz[ish] but ony around 4 db in my experience. I got 28Hz loud and clear from the boxes I built -lower than the model says it should go. And while there's ripple, again, it's never as bad as predicted. So what's going on? Clearly, it's not the fault of Martin's software: his MathCad worksheets don't lie, though they don't take room-gain into account. So it must be what we (or I!) am / are inputting into them.

I understand that this is basically a Fostex factory design from their craft handbooks, which seem to be unavailable, or almost impossible to get hold of -anyone got them, and if so, could you send me an email? How is this thing it sized? I suspect / assume the line length is set to roughly 1/2 the wavelength of the Fs of the driver, but what about everything else? It's a pre Martin design, but what methodology, and how can we figure it out? Whatever it is, I suspect it's quite basic; perhpas it's a case of I can't see the wood for the trees.

That internal baffle particularly interests me for example. Terry's original is 14" deep, yes? Subtract 1 1/2" for the front and rear walls to give us 12 1/2". Now, unless I'm confusing things here (I'm not at my best at the moment, so if I make a stupid error do let me know), I understand that it finishes 5 1/2" from the base, and 5 1/2" from the rear wall. But half of 12 1/2" is not 5 1/2", so surely the expansion of this taper is not constant? If you look carefully at his CAD drawing, it doesn't even look it. It almost seems to be two differentaially tuned pipes, one like a Voigt pipe with a rear vent, venting into a second pipe with different proportions. Thoughts?

Terry commented to me that the FE166E is a great match, and I can't deny it worked pretty well (and I like a warm ballance, believe me), the FE168ESigma is even better, but a friend (you around Dan?) mentioned that in a discussion with Terry shortly afterward, he also suggested the best bet would be a higher Q driver like the FF165K -I assume he's referiing to its 10.92 Qes, because Qts is down at 0.2 which is the lowest of all the Fostex 6" drivers.

Any thoughts and coments are welcome!

Best
Scott
 
Interesting questions. I have no idea why it works, but it does. It's a pretty simple design. One thing that's nice is there really aren't any obsticals with ceiling propagation.

I'm using mine with the 168ESigma and it works well. I built them to his dimensions for a 5" driver I had, so the 168 is basically shoehorned in. If I were to build again for these speakers, I'd open up the interior dimensions a bit. I think that would help the driver breathe a little. I have some stuffing above the driver and a piece of an old seat cover behind the driver. I should try without all of this, but I have plans for something different anyway.

Who knows, I may come back to this design down the road.

Something to note, I did some emailing with Terry with regards to this design. One recommendation he had was to chope the back of the box down at an angle from the front to open up the mouth a bit. I haven't done it, but would design that into the box if I were to do it again.
 
Re: Terry's designs

SCD said:
Hello Scott:
I am also interested Terry's designs. I think you have to add a bit to your first note. I am note really sure just what the heck you are talking about. The Abby, or one of his horns

Neither. It's his 'Bigger Is Better' TQWT which you;ll find in the TQWT DIY projects section of the Single Driver Site. He's never produced it commercially, though I understand he made a few what I suppose can be termed as pre-production-prototypes a few years back using the FE168ESigma. As I understand it, those were slightly shorter, at 64" high, and deeper at 18" (external), which would give the driver more volume and, I suspect, was caused by him increasing the area of So, probably to Sd, though that's just a guess, and where the internal baffle would end or be positioned I can't quite work out. I suspect it would be approximately 8" from the floor and rear baffle, but without having the design method / calculations / equations that's only a guess again.

Increasing the mouth area is an interesting idea I hadn't thought of -I suppose this is where we're getting to the fringe of a design type where you can go in different directions -closer to a horn or remain in the TL camp by restricting the area by mass-loading. Were yours the dimesnions on the Single Driver Site?

Clearly, it uses room gain more, or differently to most other speakers, and driving the vertical mode of a room can often smooth things out which probably helps the in-room response. It's clean too, not having much in the way. As a rule I'm in the Linkwitz camp of avoiding if possible, but if not, get as much advantage as you can! I haven't emailed Terry for a while -I figured he's busy as I haven't seen him on any boards for some time, but I might fire one in his direction on this score soon.

Cheers for the replies guys, anyone else got some thoughts?
Scott
 
hi Scott

Well, I built the BIB for the 168Sig as well, and it certainly does work, and work well. In a 13X15 room I was getting strong bass down to around 30Hz, solid and fast. The amazing thing is that this driver is in no way optimal for this load, ideally a mid Q design is best. What I like about this design is how effortless and FAST the sound is, lending itself to fast, single stage digital amps, and the room gain is such that with the 168Sig you can rattle the cages with a couple tube watts as well. VERY impressive, I would say my favorite load for wide rangers.

The 8 inch mid Q drivers yield a load so big as to be untenable for any sort of home use around here, mostly ~18" deep, 70" high, the coming HempTones will be interesting here, ...but one driver that has caught my eye and I keep coming back to is the Supravox 165LB, Qts=.45, Fs=62, 95db, and a very healthy Xmax of 4mm linear. Far more suited for this type of thing, measures very well. Perhaps Scott you would like to apply your math skills to this one.
 
Scott,

Clearly, it's not the fault of Martin's software: his MathCad worksheets don't lie, though they don't take room-gain into account.

I think the last part of you sentence is the key. The MathCad worksheets you are using contain a significant number of simplifying assumptions. Once you start removing these assumptions, and calculating a better answer, the predicted response can improve (or get worse) if the real design is dependent on other factors like being close to the rear wall as shown in the article's photo. If the math improved, so does the understanding of the speaker performance possibly revealing why it works better then originally predicted.
 
I like that 4mm xmax. The 0.45 Q looks good too for these boxes. The Fs is a bit high, but with decent excursion I doubt that'd be an issue, especially as the box is tuned so low. Same question as above: were yours the original Single Driver Site Dimensions or did you modify them at all? The hempsters look and sound promising too from what I've seen so far. I'll see what gives in MathCad for those Supravox jobs as well and get back to you tomorrow (a day off! Well, OK, so I have to spend most of it re-drafting a chapter of my thesis, but I think I can find time ;-)

I've got a pair of the original spec boxes half complete at present and a pair of FFs waiting to go in them. Or perhaps not. I don't like the 0.3mm xmax, which seems way too low for my liking, even though I suspect that's Fostex being hyper-conservative, and the overall Q of 0.2 seems very low as well, despite the high 10.92 Qms. No harm in trying I suppose, and I can re-use the drivers if they don't work well, and get the Sigmas or those Supravoxs into them. I'd forgotton about the latter -you could be onto something there, though until I can get my paws on the Craft Manuals and figue out exactly how these things are sized, with the BIB box we'd be working on guesswork as I can't sim in MathCad without accurate dimensions to plug into it. Those books from Madisound might well have them in. I might email them to ask. Things are getting interesting!

Cheers for that
Scott
 
MJK said:
Scott,

I think the last part of you sentence is the key. The MathCad worksheets you are using contain a significant number of simplifying assumptions. Once you start removing these assumptions, and calculating a better answer, the predicted response can improve (or get worse) if the real design is dependent on other factors like being close to the rear wall as shown in the article's photo. If the math improved, so does the understanding of the speaker performance possibly revealing why it works better then originally predicted.


That sounds sensible and would explain it -I don't pretend to be an expert on math or acoustics; what I know I owe primarily to yourself, Siegfried Linkwitz and Bob Brines. I suspected that it might be the case that it's down to elements outside the software's remit, for want of a better phrase (it's midnight here!); that would account for the terrible predicted response of the pipes and their frequently solid in-room performance when shoved into corners or against a rear wall. Never neutral, and they've got a heavy character, but they do some things very well, I have to admit. I'll stick to your Project 2 for my normal listening, but I might finish the pair I have in construction as my 'party speaker' or for use if I ever join the SET brigade from my usually solid-state enclave. They should certainly rock the place down nicely!

Thanks Martin
Regards
Scott
 
Just got back from my evening walk, I think my wife is trying to kill me, and I was thinking some more about the Terry Cain pipes.

The location of the open end is very interesting and probably the reason the simulation looks so bad. But as you report the real response seems to be much better. Two points probably contribute to this difference.

First, with the open end backed up against the wall the effective mouth area is doubled, due to the reflection boundary condition at the wall, and this will allow lower frequencies to be more efficiently transferred into the room (the resistive portion of the acoustic impedance is doubled). At the same time, less energy is reflected back into the line reducing the magnitude of the standing waves. So more bass output and attenuated standing waves compared to the simulation results.

Second, by pointing the open end up you are listening over 90 degrees off axis. With the mouth output reflected off the wall, the depth dimension effectively doubles while the width remains the same. This shape is going to become very directional with increasing frequency so the ragged response predicted above say 300 Hz is probably grossly over estimated.

Two interesting properties not accounted for in the current MathCad simulations. My newer worksheets are starting to account for these issues, along with the baffle step, and include them in the final calculated results.
 
Greets!

Right, the room dominates down low and let's not forget that we are primarily amplitude based animals, so between a falling hearing acuity with decreasing frequency and basically keying off the peaks, the response can be quite ragged and still sound fine. Good thing, or we wouldn't be able to listen to a typical truncated horn design or wide BW audio in a typical room.

GM
 
Thanks, Greg, that makes all the sense in the world, and explains the balanced sound we are getting using a 168Sig in one of these. The 168 measures real ratty from 5KHz up, but, after the requisite thousand hour cook time, it is truly a music maker.

I will be very interested to hear a BIB using a driver far more suited to this load. Offhand, the Supra 165 appears best. Correction, linear Xmax is actually 6mm. Bigga id betta! I placed an order for a pair of 165LB tonight. SHould be here next week.
 
Would FE208 sigma work better? I was thinking of using it in a similar design (see http://www.diyaudio.com/forums/showthread.php?s=&threadid=66135). I thought about it without seeing anything similar before. The idea came to me by thinking of simply unfolding the back loaded horn recommended on Fostex website and place it in a tall enclosure. Then place it close to the ceiling and use the corner as the last section of the horn.
 
Michael Speaker said:
Would FE208 sigma work better? I was thinking of using it in a similar design (see http://www.diyaudio.com/forums/showthread.php?s=&threadid=66135). I thought about it without seeing anything similar before. The idea came to me by thinking of simply unfolding the back loaded horn recommended on Fostex website and place it in a tall enclosure. Then place it close to the ceiling and use the corner as the last section of the horn.

I doubt it. That Q of the 208ESigma is very low -too low for this kind of load I think, though you could artifically raise the Qms with a little series resistance. I think, if you went for an enlarged one, of the Fostex drivers, the 207 would be the best bet. Your reasoning seems correct in the unfolding of the horn, though I wouldn't angle the mouth forward, I'd either keep it pointing straight up, or perhaps angled slightly backward as Terry suggests.

Martin, that's all very interesting! Looks like you've cracked it (again)! I know you've been wary of adding updated worksheets to your site for a while due to the less-than honest sorts -I don't blame you! Any thoughts about a possible future public release (when they are ready of course)? I know you licence the technology / worksheets to Bob and probably some others -perhaps you could make some kind of charge for selling the sheets to the home user or something like that? Perhaps do something like Bob does via paypal or an equvelant, or even via snail-mail. You might be able to put up different packages for instance, similar to how you have them at present, but in a charged service -a set of sheets for TL, another for horns etc etc and perhaps an overall package price, while either leaving the existing sheets available as they stand on the site, or even removing them. I know this method has problems too, but it's just a thought -I've been trying to think how you could get some decent compensation for all your work over the past few years. I know I'd cheerfully pay for the lot, and I'm sure that many others here would do the same.

Dan -that's a 6mm Xmax for the Supravox? I like the sound of that -should be onto a winner there, and I understand those drivers have a good reputation amongst those who have heard them. I'm going to complete my boxs temporarily with the FFs I have (might as well try it out); I'd be very interested to hear how the Supravox's compare to the Sigmas. Whichever you reckon seems to work best, I might go for. My room is similar in size to yours, so this should be good. My bet is the Supravox with it's higher Q, though the 62Hz Fs might mean it's going to be tuned lower than it's comfortable with in these pipes. Oh yes, I'll let you know of any other design I can come up with for them later.

Regards to all
Scott
 
The one and only
Joined 2001
Paid Member
Re: Terry Cain's BIB -why does it work and does anyone have those Fostex Craft Handbooks?

Scottmoose said:
Clearly, it's not the fault of Martin's software: his MathCad worksheets don't lie, though they don't take room-gain into account. So it must be what we (or I!) am / are inputting into them.

If I understand correctly, they also don't handle dissipative
loss. After building maybe 50 rear loaded designs over the
years, I have concluded that undamped lines behave as
predicted, but not usually as I want. When you start damping
them out with stuffing or lining or whatever, the characteristics
relax and you have a lot more leeway in the specifics of the
dimensions and so on. Damping (or lack of) also can be used
to compensate for room character and placement.

That's the good news - that anybody can play with the damping
or anything else to improve the sound. In my conversations with
Terry I see that like myself he very much favors experiment -
he tries lots of things and some of them work out.

:cool:
 
If I understand correctly, they also don't handle dissipative

Nelson,

Actually the MathCad worksheets account for the damping provided by Dacron fiber. You can define the amount and location of fiber in the line. Calculated results correlate very well with measurements.

I also like to experiment and try a lot of things to see how they effect the response, but I choose to do it analytically and only build one speaker design. This is extremey fast and efficiently produces an optimum enclosure that meets my initial design goals. In one evening, I can look at many different designs and down select to the one or two most promising ideas. These ideas get refined even further to determine which design I want to finally build.

I learn more with each design and am continuously improving the calculation methods. Final tweaking is done on the finished speaker by listening, but this is typically only pulling the last little bit of performance out of the design.
 
scottw said:
Hey,

Could you elaborate on this a little for me?
You got 28ish Hz from a 166e?
Thanks,
scottw

Well, they do some things very well. Imaging is usually good, and they are subjectively a very fast sounding speaker, as you'd hope from single drivers anyway. With regard to the bass -yes and no. Say rather I got 28Hz out of my room. These things have to be pushed right against a back wall or into corners to work, for reasons Martin has explained in a previous post (and his reasoning seems likely to me -nobody else has ever explained how / why they frequently work well in practice). How low you go will rather depend on how large your room is. I got a usable 28Hz, anything over 30Hzish was loud and clear. The FE166E is not an ideal load though -its Q is too low. These were just an experiment I roughed up (Terry asked if I'd ever measured a pair, and I told him no, but I would rough some up and see what they were like).

I've got a better-built pair in the works currently, but I'm waiting on drivers. The best of the Fostex 6"-7" range for the load would be either the FE168ESgima, or the FE167E, which have a higher Qts than the FE166E. That said, the FF165K has the highest mechanical Q of the lot at 10.92 I believe, so as I have a pair, I might try those -however, that might be a dead end as its overall Q is very low however at 0.20. That might be fixable with some series resistance to boost the Qes though. The Supravox 165 with its higher Q Dan is trying out ought to be a the best match of all, though my only (slight) concern is that it's high Fs will mean the pipes as they stand are tuned way too low for it, and might push it outside of its operating parameters. it has a 6mm Xmax however, so it could be happen enough; a try it and see situation.

You ask about the character of the cabinets. Well, they will always have a dip at circa 100 - 150Hz -that's a consequence of the driver positioning, and there's nothing you can do about it. However, as they drive the room's vertical as well as horizontal modes, this is often flattened out by the boost you get from the vertical mode in this region, so the dip is not as pronounced as an anechoic simulation might suggest. (I'm guessing that's the reason, but I don't think I'm far off: 8' will give you gain at around 135Hz) However, this is really the key issue: the lower you go, the more you are listening to your room, activated by the cabinet, rather than the speaker driver itself, so what character you hear will be dependant on your own room, rather than the speaker itself -more so than any other design I can think of, off-hand.

As for the rest -there's ripple, more than other cabinets, like Martin's Project 2 ML TQWT or Bob Brines FT1600 Mk2, but they have somewhat different priorities; they're a cheaper and simpler construction (not that the other two are complicated), and they go lower. Without BSC or added series resistance, they're also more sensitive, so will match well to low-power SET amps, and allow you to play rock or complex orchestral pieces at window-rattle levels with such amps; not something you'd get very often.

Cheers
Scott