Coming back to life! Question about 10" woofer position in a 3 way speaker

[fjhuerta]

Hi all! I had a couple of things to take care of (ie, sort of another hobby, but one in which I think I'm all set for a couple of years), and I'm back!

I already have everything I need for my next project. It's nothing special, at least considering the designs in these pages. It'll be a Duelund WMTM tower.

Drivers will be the RS270, TB W4-1337s and either the Vifa D25 or the SEAS 27TAFN. I think I'll go with the SEAS, if only because I love the way they look next to the 1337s, and I'm not going to run them low.

The plan will be crossing the drivers over at 300 and ~4000 Hz, 2nd order (obviously). I want to take advantage of the very wide bandwidth of the 1337s as much as possible.

Here's the question I have, though. I don't have a clue as to how to measure the RS270s.

I know John Krutke mentioned it was a pain to get a decent measurement out of his big SB12 woofer - he said he had to do quite a couple of them, merge them, and then he decided not to use any BSC at all, taking advantage of corner loading. Quite frankly, ;x( I have no clue as to how to do it. I suppose I could place the speakers as high as I can and then take a free air, 2.5 measurement and work with no or 3 dB BSC, but that seems to be too much guesswork. I was wondering if any one of you has ever done such a measurement, and how you accomplished it.

Thanks for any help!

[5th element]

This is a very interesting question and something I'd love to see an answer on also.

[lhwidget]

I'm looking for corrections if I'm wrong here, because I'm about to start on an RS270 build also, thanks in advance for criticisms and advice.

With SoundEasy, my basic methodology is:
1) Impedance/TS parameter testing in free air (break in to taste before hand, or check values before/after to see if it matters)

After building the enclosure and mounting the drivers
2) A near field measurement of the driver
3) A near field measurement of the port
4) Add the port and driver near field responses (not merge, includes phase)
5) Move back to a meter and get the driver's far field response
Merge the driver's NF & FF responses.
6) Pull the phase back into alignment with a time delay adjustment to avoid the break caused by the merge

One of the tough parts is picking the frequency for the merge. If you are getting into the range where the enclosure's baffle step is effecting the drivers' response, it starts to be problematic. D'Appolito recommends a long window to extend the valid range of the mid's measurement, then pick a fairly level response interval in both drivers' responses below the baffle step influence. If I remember correctly, this is what drove Zaph to try ground plane measurements for his SB Acoustics design.

[NyxOne]

I don't have any pratical experience on this but the readings I've done would suggest using ground plane measurement. Most of the subwoofer test I've seen use this method and MJK (from quarter wave) use this method.

Good luck,
Chuck

[cjd]

Some of this depends what you intend to DO with the data.

I focus on having at least -24dB of range on known-good live data with exactly the same measurement point for all drivers involved. No merging. This means that the data is there to get the crossover cleared - comparing against measured net results has so far shown this to work for me.

The problem with this arrives when you do as you say: 300Hz 2nd order. Ahh, so now you need to go quite low with good data. 600Hz 8th order is easier.

Ground plane seems to be the most most consistently used method for lower frequency data acquisition, but it means you need to work in a minimum phase world, so accurate offset becomes critical. Even that data is within your reach though, so nothing to worry about specifically.

Please do share your results as you learn - it will be good to have more notes on it.

[fjhuerta]

Yeah, it's going to be a pain.

I suppose I could do what I usually do - place the speakers as high as possible, measure them in the biggest room in the house (basically an open hall) and measure them at 2.5m with the biggest window I can use (I forget what it is at the moment... 10 ms? Which should be accurate to 100 Hz. It'll be noisy, but at least I'll have a guideline about where BS starts (I think that due to the size of the box it should start at 250 Hz, or so). Then I'll do a normal 4 ms window sweep.

Then I'll do the usual close mic FR sweep, merge, and add BS to the 4 ms measurement using as a template my 10 ms measurement. I want to avoid ground plane measurements, because I've tried them in the past and I've never been able to properly integrate them to the rest of my measurements.

So far, so good ( I think ). BUT - and this is where this is all uncharted territory to me... if I place the woofer close to the floor, I'll get a ton of gain... and I have no clue as to how to account for it. Even if I place the woofer high up, I suppose there'll be some gain. From what I understood, Zaph decided no baffle step compensation was the way to go, but he added a design with an extra 3 dB of mid / high frequency power in case bass was overwhelming....

Maybe there's no real way of "guessing" how such a driver will behave in the real world, since it highly depends on where you'll be using it, anyway. A good starting point may be to consider no BSC at all, or 3dB at most, since I suppose that at 200 Hz room gain starts becoming a big factor, and then work from there.

[cjd]

Your usual practice sounds like my usual practice. However, I will run data with the woofer properly relative to the floor, because that always adds a totally different set of data to the picture.

I haven't done a 3-way design without running a crossover against different data sets to try to get a consistent picture (and to give myself a sanity check as well).

C

[ultrakaz]

Good to see that you are back and going to document your next design.

Yeah, it's going to be a pain.
Maybe there's no real way of "guessing" how such a driver will behave in the real world, since it highly depends on where you'll be using it, anyway. A good starting point may be to consider no BSC at all, or 3dB at most, since I suppose that at 200 Hz room gain starts becoming a big factor, and then work from there.

If you were evaluating a number of 10" drivers you would need some way to measure them uniformily (like Zaph), but since you have already chosen the RS270 and you more or less know where you are going to place it, I would follow C's method and measure how you will use the driver.

In any event, please keep us posted on your progress.

[fjhuerta]

Thanks

I just noticed Parts Express has a sale on the RS270's, so I bought another pair. After running initial simulations, I was surprised to see a single pair will run out of excursion pretty fast. I still haven't decided whether I'll use 2 of them per side (probably I'll do so, since the box will be at least 90L).

I'm wondering, if I was to glue two baffles in order to get a 38 mm one, can I use white glue to do so?

[Renovator]

Thanks

I'm wondering, if I was to glue two baffles in order to get a 38 mm one, can I use white glue to do so?

Should be fine. We do it all the time for production of veneer panels where we the required thickness doesn't exist. Laminate the two panels in a cold press over night and then the panel goes into the veneer press the next day.

Haven't had one fail yet! 8)

Lewis

[cjd]

I did 36mm baffles for the big WWMTM's (using a pair of RS270's...) that way. Works perfectly.

Are you trying to run them ported?

They go solidly into the lower 30's sealed, so... And they can take a BEATING too, let me say. It's fun to watch them dance.

[fjhuerta]

Hi C, I'd definitely use them sealed. I have been looking at your design, and the only thing that scares me away of dual RS270s is the impedance... 2.5 Ohm will need a hefty power amp. I do have it, but I want the speaker to be a bit more flexible regarding amplification...

Then again, 95 or so dB efficiency sounds too good to pass up!

[fjhuerta]

Should be fine. We do it all the time for production of veneer panels where we the required thickness doesn't exist. Laminate the two panels in a cold press over night and then the panel goes into the veneer press the next day.

Haven't had one fail yet! 8)

Lewis

Thanks! I just coated a panel with white glue, used a brush to coat it as smooth as possible, and joined both baffles. It seems it'll work just fine!

[cjd]

Hi C, I'd definitely use them sealed. I have been looking at your design, and the only thing that scares me away of dual RS270s is the impedance... 2.5 Ohm will need a hefty power amp. I do have it, but I want the speaker to be a bit more flexible regarding amplification...

Then again, 95 or so dB efficiency sounds too good to pass up!

I'm pretty sure 2.5ohm is partly due to the crossover? Maybe it isn't and I never paid that much attention.

Only ~90.5 or so after you take out baffle step, but that's still quite nice.

[fjhuerta]

Yeah, your big wwmtm's drop to 2.4 Ohm or so. Even without a crossover I believe the impedance for two RS270's is around 3.5 Ohm. Tricky.

Zaph didn't use any BSC on his big 12". But I do believe I'll need it - every speaker I've tried to build without it simply sounds strange. That's why I thought I could hit 95 dB - I was trying not to use BSC...

Still, 90 is good. REally good

[cjd]

2.8ohm

3.5 is quite within stable for 4ohm rated amps. And reasonable output levels for 8ohm stable.

I have used a chip-amp that's definitely NOT happy on a 4ohm load (the Ansonica trips its protection circuits if I crank it up) and it's stable at low listening levels, but goes out pretty fast if I push above ~80dB on anything with bass response (i.e. none of my solo violin albums )

C

[fjhuerta]

Yep. I remember reading about how much headroom you actually have with your towers. I want something like that, too.

Thing is, you have the RS180's so I'll need to run the RS270's a bit higher. And I'll probably use the SB29 dimple dome tweeter instead of the small SEAS - I still don't know whether the little 27TAFN can keep up with a pair of TB W41337s and another of RS270s. Sounds like too much for it.

I have time to think, while the baffle glue dries up

[fjhuerta]

OK, I've been playing with a 2nd order crossover, and the impedance is quite a bit more benign: 3.5 Ohm. It looks a lot better than a 4th order, so I may be going that way (I simulated the response using a simulated BSC).

I did simulate a ported box, and it's not huge - it's even bigger than that. I need around 250L tuned to 25 Hz to make it work, it's out of the question. The speakers have to be sealed if I'm to use 2 woofers. 1 woofer may work in a ported enclosure, but I'm concerned about the limited Xmax of the Daytons (6.6 mm). From my limited simulations, a single RS270 can't keep up with two properly filtered W4-1337s.

What can I say, I'm a fan of ported boxes. But this time I may go sealed.

[cjd]

Haha. Yeah, bigger than huge is about right.

I think you may be pleasantly surprised with the bass sealed. Get a 21" maelstrom and port it and cross at 40Hz if you want more.

C

[fjhuerta]

I'll be doing the baffle layout today.Two RS270's sealed it will be, and hey, the SB tweeter is also in. The star drivers will be the W4-1337s, though. I just love them to pieces, and it's the only metal mid driver I've never grown tired of.

[Silversmoky]

Nice driver choices Javier. Should be a very nice system, albeit a larger one. I agree that those Tang Bands are something special in the midrange. Looking forward to following your build.

[fjhuerta]

Thanks again!

The baffles are ready to be routed, and I have the design all worked out. Yep, they'll be big, but the space they'll be used in is quite large, so it'll all work out.

I played a bit with a WMTMW design on a small (6") base, but ripples and diffractions looked pretty ugly. So a WWMTM it'll be.

I've also been playing a bit with the Duelund Synkron Filter web site (http://pro.speakerbuilder.dk/synkron/). I want the midrange to cover as wide a range as possible. I'm considering a 300 Hz / 3,500 Hz bandwidth, with quasi 2nd-order filters. The RS270 seems to be adequately behaved up to 1,000 Hz, which is the upper limit it'd be used in this configuration.

I'm still considering, though,whether to go with the Seas 27TAFN, the SB29 or the Vifa D25Q. Reason being, if I cross over at 3.5 KHz I may not need a very beefy tweeter, but rather a small one... if I could get away with it, I'd definitely use the Seas because of its size.

[5th element]

I have just used the DQ25 in a design of my own, where I paired it with an RS125. I was extremely impressed with how the tweeter sounded, in fact the whole system sounds great.

[cjd]

With that kind of coverage I'd personally be looking to reach closer to the RS270's with the tweeter - particularly given the temper of the TB's, you could consider a 4th/2nd/1st order crossing strategy (or a hybrid on the RS270's, with a lower knee to knock the break-up out but an octave or more at 2nd order and then a knee).

[dlr]

I'm going through some of the same difficulty right now working with a dipole system design. The woofer (pair) is the most difficult and seems especially so with a dipole.

Right up front let me suggest reading the new manual for the SoundEasy ultimate Equalizer here . The basics of the procedure explained in it apply for any design.

The one aspect not being mentioned (unless I missed it in the thread) is that the baffle diffraction must be handled and especially for the midrange unit. A far-field measure spliced to a near-field (close-mic) measure that is adjusted to some selected point will have some error even if done below step unless that point is not much higher than about 100Hz for typical baffle dimensions. The procedure outlined in the SE UE manual shows one method to handle this. I've got a writeup of a similar procedure that I used that explains it in some detail here, you may have seen this.

So far, so good ( I think ). BUT - and this is where this is all uncharted territory to me... if I place the woofer close to the floor, I'll get a ton of gain... and I have no clue as to how to account for it.

With the woofer near the floor, you'll be largely operating in 2-pi. But even with a 200Hz crossover, that being the Fc, it will be in transition to 4-pi and above Fc it will be increasingly so. With two woofers, it's even more problematic (as I've recently found out), since the upper woofer will have more diffraction influence (operating more in 4-pi) due to the extra distance to the floor.

Maybe there's no real way of "guessing" how such a driver will behave in the real world, since it highly depends on where you'll be using it, anyway. A good starting point may be to consider no BSC at all, or 3dB at most, since I suppose that at 200 Hz room gain starts becoming a big factor, and then work from there.

It's seems that no matter what I've done with a woofer, I've had to use a bit of "guessing" to some degree. But with a 300-350Hz crossover and a woofer part-way between floor and mid (my preferred location for a single woofer), I've used the 1m measure from the box (or OB baffle) placed on a stand spliced with a close-mic response that was then adjusted for diffraction, then spliced. For the woofer, though, even without the diffraction splice, it may be sufficient since floor bounce alters the response in any case.

dlr

[cjd]

FWIW my crossover ultimately came together using measurements taken at 2M on the tweeter axis. This data gelled with all the data I tried pulling together in various ways, and if it didn't, some live testing almost always showed I messed something up in assembling data (splicing, etc... I don't think I thought about baffle step re: splicing in at the time, though it was a while ago so who knows).

Thus, it takes time.

just keep having fun.

[fjhuerta]

Thanks for your comments. It's good to know there's at least some amount of guessing no matter what you do. I always wondered if you could predict the behaviour of a woofer so close to the floor - I suppose that at 200 Hz or so it doesn't matter that much, since every design will have it, and we all listen on real rooms, and not anechoic chambers...

So... I glued the baffles, drew my design on pencil on top of them, and yesterday had a free day at work. I took out my Makita and my Black and Decker routers, did some things in order to be able to use the vaccum adaptor on the Makita, and went to my parents' home (they have a huge space to work in) and, 5 hours later (hey, I'm a n00b at routing!) had some stunning baffles.

I was really, really really happy, they turned out great!

And then... I measured the actual speakers on it. Not even ONE of them fit. Not one. The woofer came pretty close - 1 mm or so. The mids need an extra 2 mm or so. The tweeter is close to 2 mm.

I have no clue as to what happened. I measured 3 times and cut once. The only thing I didn't do was measure the cuts with a caliper after I did them, but not once had my Jasper Tools failed.

In retrospect, the only thing I can think of is that I used a 1/2" bit on the Makita. I did consider this when routing the holes - I used a hole 1/4" smaller than the one I needed. I'm thinking the bit wasn't really 1/2" - it may have been a bit smaller. I don't really know. All I know is, I'll need to buy another sheet of MDF, cut it, glue it, draw on it, and wait for another free day at work. ARGH!

A couple of quick thoughts...

1) Makita vs. Black and Decker. I had so many questions about routers back then, and using them for 5 minutes told me all I needed to know. The B&D is very precise on depth cuts, it's light, it has speed control, it's a joy to use - and it became hotter than hell after 2 or 3 minutes of use. It's obvious a 1 1/2" baffle is pretty close to its limits. I only cut the tweeter holes with it, and used the Makita everywhere else. The B&D barely could cope with the 7.2 cms hole.

2) The Makita is a beast. Simple, yet tremendously more powerful than the B&D. It ripped thru MDF as if it was paper. It was love at first sight (can someone fall in love with a router? I say yes!). It's very heavy, though.

3) Next time, I'm measuring three times, cut one time, then measure again, then cut again, then measure again.

4) I have the baffle diffraction file, the enclosure size and bass response, and the baffle design on PNG. I'll post them later.

5) A vaccum cleaner connected to the router is not an accesory - it's an absolute necessity. I'm never, ever routing without it. I used a respirator and at times it seemed unecessary - the vaccum cleaner is a very effective tool here. I was able to clean up my parents' garage in a couple of minutes afterwards, and my tools were fairly clean afterwards. I suppose the motors will last longer, too.

I can't wait to cut again.

[cjd]

Are your through-cuts centered and cleanly round?

I have a nifty little jig I made for a Bosch trim-router that worked perfectly when I needed to add 1/2mm or so of size due to a similar problem - under-sizing slightly. I can snap a pic - hopefully I remember NOT recommended with a larger router though.

Despite careful measuring, my latest baffles ended up with a pretty consistent 1mm extra. Probably "perfect" for most, but I like that snug fit that makes folks wonder how it was done. I'll just go a little heavy with the poly-coat.

C

[Silversmoky]

That is frustrating Javier. I had this happen on a project earlier this spring. I always use a 1/2 bit to do cutouts so I corrected for this on the Jasper Jig which has always been very accurate in the past. I was using an older bit I have had for a while and did the cutout for the tweeter. It came out about 1 mm to small which baffled me because I had cut this tweeter cutout before and knew the exact size. After a thorough run through I noticed the bit was worn just slightly at the cutting end. Almost beveled at a an angle just slightly, but definitely enough to throw tight measurements off. Now I always check for wear and use a new bit when appropriate.

Cjd - I would love to see a picture of your jig for enlarging just slightly.

[fjhuerta]

Yeah, I'd love to see the jig, C!

I'll cut this week-end again. I bought two new bits and I have my calipers ready. 1 extra mm is fine. I'd rather have that than what I have right now...

I'm wondering whether to get 30 mm MDF or sandwich two 18 mms together. I think I'll go for the 18 mms...

[Ray_D]

Solution to accurate and repeatable driver cutouts

The solution I use now to make driver cutouts, after years of jasper jig use, is to make a pattern for each driver and use that. The pattern is sized for the clearance hole based on a mounting flange size which matches a rabbeting bit. I actually usually use slot cutting bits since they are cheaper. Also, the size of the recess can be changed by changing the bearing on the bit.

I cut a recess on the pattern just to check the fitting of the driver. If the hole is a little small, some sanding and re-cutting the recess can fix it. If it is too far out I just re-make the pattern on some scrap.

You lay out the driver positions one the baffle and mark the cutout. I cut out most of the scrap with a jig saw, clamp the pattern in place and remove the rest with a pattern flush trim bit. Then with the pattern removed I cut the recess. For most drivers you can cut it a little shallow and finish it after veneering to get it perfect.

Ironically, I have a custom jig I designed and built which makes very accurate holes and has built-in dust collection; but, with this technique I use it very little.

You might be able to use this approach if your holes are undersized.

Regards

Ray

[fjhuerta]

Oops... well, I bought some more Trupan (I hadn't realized it wasn't MDF - or is it?) and I glued the baffles, so Saturday is routing day.

I'd love to hear some opinions regarding the lifelong debate - ported or sealed?

At first, I thought sealed was the way to go, since the box would be far too big by using a ported design. But I just realized I may gain a bit extra SPLs by doing the exact same box I was thinking about doing anyway (around 167 liters!) ported.

The box is huge because, quite frankly, anything smaller would have looked strange. I'm trying to mantain the "slim tower" design, and that calls for quite a bit of depth (around 114 x 37 x 70 cms).

Still - and I'm feeling strange about doing it - I don't know because BassBox insists sealed is the way to go and shows a bit of a hump at 140 Hz (1 db?) while Unibox shows no hump and a bit of an extended bass shelf when compared to sealed. As a matter of fact, ported holds an advantage right up to 17 Hz or so - more than enough, IMHO.

I'll post some pics later today.

[EdL]

other diameter options

options are good...

...you may choose to use either an over or undersized bit.

On the jasper jig...Move the pivot pin into an adjacent radius hole position to get intermediate fractions.

A 5/16" bit used instead of a 1/4" will change the radius by 1/32" (0.032").

...and then you can get into metric bits... :B

Some math will provide you with any number of sizing options.

[EdL]

I'm in the ported camp. I like the ability to "tune" the enclosure for the intended room.

I choose a simulated "drooping" response to offset the experienced room gain.

A benefit is an extended low end response.

I use MJK's worksheets to do this.

[cjd]

You can always plug the port to see.

IMHO you'll gain a negligible amount of bass response at the expense of some clarity, and excursion becomes a bit more of a concern. If you like to rock out, I think you'll find the response is solid sealed. If you like lots of range but don't listen too loud, you may prefer the extra extension.

[fjhuerta]

I don't know. I suppose, as you well said, I can always plug the port, but I can't port the box at any time. It's just that I sometimes find sealed designs to be a bit on the lean side. But the simulations I've done all point towards sealed as a better choice.

I'll post some numbers tomorrow or Friday, I need to really start documenting this build.

By the way, do you know what a "Garnacha" is?

[EdL]

The "Metro's" were in attendance at last weekend's CKDIY get-together.

They are a collaboration between Robert over on PE's Tech Talk & yours truly.

The two of us can't agree on the alignment issue. This is where we are:

As shown, the cabinet is sealed. It is made from 3/4" mdf which is heavily reinforced with ~3/4" hard maple struts and panel stiffeners.

The bass driver is a Genelec/Peerless 10" buy-out (nla). The cabinet internal dimensions are 13" x 15" x 26". It was designed as a MLQW enclosure in which Robert may install aperiodic vents. He is very pleased with the bass response from the cabinet as-is, so they may never see the vent in his set.

I plan to install the port in my set. Then I'll play with the length and stuffing.

So, to my above comment I should add that I don't know all that I will know...

[Silver1omo]

By the way, do you know what a "Garnacha" is?

You mean these?



Or these:

[fjhuerta]

Hah! First ones!

I'm actually tempted to just use a single RS270 and port it, but it'd kill one of my main design goals - do a huge speaker I'd never be able to buy on a store, with SPL capabilities far beyond what my ears could stand, and vanishingly low distortion levels. Just because it can be done.

Baffles are ready for cut-out session #2. I have two new router bits, my caliper ready, and patience... lots of it.

[fjhuerta]

So, here's where I'm at so far.

Baffle design.

I obviously tried to minimise diffraction by careful placement of the drivers. Following Zaph's big 12" tower, I´placed the mids with a slight offset, and did the same with the tweeter. Diffraction played a big part in me not building a smaller WMTMW placed on a base, actually. The ripples looked ugly.





The only ugly looking thing is the tweeter ripple at 1.5 KHz, but I'm crossing over higher than that, so it shouldn't be an issue.

[cjd]

Garnacha is a grape. At least, that's the meaning I'm using. AKA Grenache if you prefer French to Spanish wines...

[fjhuerta]

This is the baffle layout I'll be building. Sort of a very ugly drawing, but worked for me...

[fjhuerta]

Considering the following parameters:

1" thick walls
1.5" thick baffle
114 x 37 x 70 cms outside diameters

These are the preliminary sealed vs. ported responses. Ignore the other graphs...




I think I'll go ported...

[fjhuerta]

Garnacha is a grape. At least, that's the meaning I'm using. AKA Grenache if you prefer French to Spanish wines...

It's a nice name, although I always knew it as:

Garnacha, "street" foods, especially in Mexico[1]. In Oaxaca, it is an actual dish: a small corn tortilla fried with shredded meat, cheese, lettuce or a prepared type of cabbage topping.

Garnachas are pretty tasty, by the way...

[Silversmoky]

It is looking pretty good Javier. Just reading through I think I lean towards sealed with those 270's. Judging by CJD's comments on his big towers and your graphs they look to me like they would still have a very strong low end. But, lately my preference has kind of switched from ported to sealed a bit.

Did you decide on what tweeter you will be going with for sure?