Tang Band W3-881SI line array problems

Hey everyone, so I have built a few sets of speakers/amplifiers in the past, and recently got the urge to build a new set of speakers. This was all triggered after finding all the glowing reviews about the Tang Band W3-881SI drivers. I had always wanted to build a line array type speaker before, but they have always been too cost prohibitive; until I found the W3-881SI drivers. At $8 each making a set of line arrays becomes a lot more feasible. So I opened up WinISD and designed a set of ported enclosures that would house 9 x W3-881 Drivers each, wired in a series parallel configuration to maintain an 8 Ohm impedance with no corossover or filtering. The enclosures were designed to be 52.75" tall, 4.75" wide, and 12" deep and would use 0.75" thick MDF which made the internal volume about 1750 in^3. Tuning the ports to 70 Hz would give a decently extended and flat bass response. I was excited about these speakers and ordered everything online and built the enclosures and assembled them all (pic below) . However when I hooked them up I was very disappointed. :( The bass response was decent as I expected it to be, but the rest just made me sad. The speakers just sounded dead, and lifeless, and somewhat echoey. Not pleasant to listen to at all. The off axis response was awful as well. Going from a standing to sitting position or from left to right changed the sound drastically. Especially the highs which seem almost non existent sometimes. Needless to say I am pretty disappointed. For future reference I am wondering if there is something I did wrong, or could do differently in the future. I am thinking that maybe some internal dampening (stuffing/felt/foam) might help some, but am wondering if there is something else I could do in addition to that. Currently I don't even want to use them, so I figured I might get some input form all of the experts on the forum before I put them away in the basement to rot.

Any advice you can give would be much appreciated.

Thanks for the help,

Charlie

speakers.jpg
 

siruis815

Member
2011-12-23 12:54 pm
MO
From what I know - which is very much - but it seems like:
1) drivers are too far apart
2) line array isn't tall enough which is probably causing the vertical issue (I think if it were from floor to ceiling the issue would be resolved)
3) short line arrays should have some curvature and possibly a crossover to limit high frequency reproduction to the central-most drivers

Please feel free to correct me on any of this as I'm still learning.
 
Charlie,
First let me say the cabinets look great - really nice job.
I'm sorry to hear that they did not work out. These drivers should have nice mids and highs when used as single full rangers, so it is odd that you report their highs are lifeless when used in an array. How much damping/stuffing are you using inside? I suspect that what may be happening is that you have a long transmission line and the drivers are distributed at peaks/valleys of the waves and causing destructive interference. There may be a way to check to see if thus is what is going on.
1. Disconnect the bottom 5 drivers and run just the top 4 in 2x2 at 8 ohms. 2. Add stuffing to top 2/3 of cabinet. See if this improves the lifelessness. The bass will not have as good extension because port is too big, but sound quality should improve. If it does, I would go with just 4 drivers which will reduce effect of listening position in vertical. Seal off bottom 5 drivers and recalculate new port size. Now you have 10 more drivers for other projects. If you want to keep all 9 drivers, stuff entire cabinet densely to make it aperiodic. Bass will not be as good but it should sound better overall. The lots of stuffing is easy to try. Use cheap polyfill from pillows or open cell gray foam. In general, spacing of drivers should be as tight as possible for line arrays. If you go with 9 drivers, you will still have spatial up down sensitivity to listening position as array focuses sound into a vertically narrow beam - which is why line arrays are tall floor to ceiling deals or are curved.

If you want to use about same driver numbers convert back panel to have 4 drivers firing backwards in bipole config. Lose just one driver and seal bottom 5 front cutouts. The bass response of existing port should be ok still. Still need to stuff upper 2/3 of line.
 
Last edited:
The drivers are far apart. You are hearing comb filtering. Also you will need BSC if there isn't one already. Seeing that you have 9 drivers, you could do frequency tapering, and call it a point source. Just apply a first order x-over around 500Hz, having the middle one be the HF driver. This will take care of the baffle step loss as well. You might even get away with keeping the same port, it will just be tuned a little lower.
By the way, they look really nice.
 
Thanks for all the replies everyone. I tried "tweaking" the speakers some today by adding stuffing to the cabinets, as was suggested. This did seem to remove some of the muddiness from the sound, and it became much more clear sounding, but still it doesn't sound right. I read some more on line arrays, and I agree that I am getting some sort of combing effect with upper mid range frequencies, and reduced output in the high frequency range. That being said, if I sit with my ears at about the height of the center speaker they do sound fairly decent, but even moving a foot out of this sweet spot they sound poor again. Lesson learned on my part, I should have done more homework before rushing into a project like this.

I probably wont be using these speakers at all, especially since the "budget" speakers I made about a year back (pic below) sound much much better. Crisp highs, smooth mids, and deep lows, no matter where you are in the room. Plus they are pretty sensitive witch makes them great for parties. Oh well at least the line arrays look kinda cool. :) I don't suppose anyone wants to take them off my hands. I would be willing to part with them for a small price. :p

Charlie

speaker2.jpg


http://i193.photobucket.com/albums/z114/newcharliepuch/speaker2.jpg
 
Haha, you are right. I am glad I made them. I had a lot of fun building/designing them, and yes I did learn a thing or two. Just too bad they didn't turn out as good as I had hoped. Lesson learned though. Next time I will do more homework.

I may try wiring up the configuration with the 5 passive radiators tomorrow and see how it sounds. I doubt it will work any miracles, but hey its worth a try, and pretty easy to do. Thanks for the suggestion.

Wow those plots are pretty cool, and look like they paint the picture as to why the sound is so "off" sounding. The combing effect is very profound, and I would agree based on my listening tests that I have done. Especially the plot of the vertical off axis response. This explains why it sounds decent when when at that one perfect position, but as soon as you stand up or move slightly the sound changes so much. Very cool. What software did you use to model that? I might need to invest in a copy. It looks pretty neat.

Thanks for the replies,

Charlie
 
A wideband driver based array can work, but it needs to be very carefully implemented (including driver selection), and typically run with Eq since at higher frequencies the individual sources do not sum, the interference pattern / lobing / comb effect depending on the phase angles in question. For practical purposes, a basic ROT is that the top end output will have a similar SPL to that of a single example of the drive unit employed. The top end is never likely to match a good array of tweeters, but it should be adequate. Since your box is vented, active Eq is basically mandatory; passive is out. Other than that, the apparent comb effect seen in measurements / calcuations, providing it's not too severe (anything above 4in drivers is pushing your luck) is usually not much of an issue if at all, since the precedence effect kicks in.

The polar response of an array in the fresnel zone is largely cylindrical (bit of a fudge but it suffices) so as soon as you move vertically off-axis, the HF will fall off a cliff. That's why it's preferable for an array to be physically long, usually from floor to ceiling, or at least taller than the listener to ensure a reasonably uniform response if you're going to be moving about rather than remaining in roughly the same location / height.

As noted, the drivers in the boxes shown are spaced much too far apart (they need to be packed as tightly together as possible) which is one of the causes of the problems you mention in the mids & HF, particularly off-axis. This also depends on the off-axis response of the drive units employed; the 881 isn't too bad on this front, but it does drop off somewhat at the top end; using them in an array isn't going to improve this, and having them spaced that far apart will exacerbate the problem. You might be able to reduce this a little with some power-tapering, but it's not likely to help all that much with this particular box / configuration.

Jim Griffin & Tom Danley might have some more suggestions if they're around.
 
Last edited:
OK. my 2 cents worth.

It would have been better to build the box with the drivers on the wide side. Dunno why everyone insists on thin narrow boxes, that creates more problems than you can throw a stick at. (silly idiomatic American expression)

To "fix" this box, move the port to the side or rear. Router out or saw cut out the old front baffle (oh, hey consider moving to the wide side! - ur response will be better in the mids and flatter), assuming you want the thin side out, and make a new front baffle with more drivers, go floor up, as little gap on the bottom as possible.

Floor bounce
cause a big NULL based on the height of the lowest driver and the distance to the floor and the reflection back vs. the ear position. Add drivers down to the floor - no floor bounce in the lower mids... it also does the array lengthening effect due to the reflection...

Now, depending on the drivers and the effect of the distance between centers you may need to A) do frequency tapering or B) put a center tweeter in.

(another idea is to add a 10-12- 15" woofer on the side and partition off the "full range drivers" inside the enclosure so that they are now mid/high drivers, eliminating the floor bounce problem that way. Move the array that you have so the drivers are as close as possible to each other. This would require a bafflectomy as well. Put the woofer as close to the floor as you can... it can be sealed or ported...)

Inside the box, with a port, you do NOT want "stuffing". What you need is absorption on the walls. What to use for that? The ideal material is wool felt. Synthetic felt is not as good but how it is deployed and what frequencies you want it to work at (thickness counts) will alter the performance. Keep in mind the absorptive material will change the volume of the box, and alter the tuning a bit. Foam is not very good usually.

Did you pick the volume of the box based on T/S parameter and a simulation? Or is this a guess? It needs to be based on a simulation, otherwise the result is random.

Can you post the mfrs response curves and T/S curves?

ARE YOU SURE YOU HAVE WIRED YOUR DRIVERs IN PHASE?
How do you know?

Fwiw, if you keep the drivers spaced where they are, you can roll them off lower, and put a high quality tweeter into the mix. Base the rolloff point on the frequency where the effect of the distance between drivers starts to become an issue with comb filtering.

A measurement or two using any random mic and some FFT freeware would tell you some useful information about what the speaker are doing. The Panasonic WM mic is a few bucks and flat enough out of the box for this sort of general investigation. Calibrated mics are not really needed for most things.

No matter what, any array is no better than one speaker alone. So have you listened to ONE speaker in a box alone to see how it sounds?? If it sounds essentially the same, then you know that this just isn't a good speaker to start with.
 
Last edited:
Follow up:

A quick look at Parts Express' site says VAS= 0.0670979cuft. Qts = 0.8 from the mfr and 0.58 by a Parts Express test. Fs is >100Hz.

Bottom line is that this driver does not appear to be particularly suitable for a ported design, wants a vanishingly small box volume, and looks like a candidate for a sealed enclosure. With stuffing for absorption of rear mid and HF energy...

It will run from above 100Hz up in an enclosure. You need a woofer, imo.

However it is very flat - good.
 

Attachments

  • 299-113-factory-buyouts-w3-881si-specifications.pdf
    20 KB · Views: 34
Last edited:
Thanks again for all the input. I really like the idea of replacing the center driver with a tweeter and crossing it over at the frequency where the comb effect starts kicking in. I could even add one additional hole further down in the box to keep the original 9 drivers. I like this options since it doesn't require ripping apart the whole enclosure.

Thats interesting that you say that the drivers would work best in a sealed box. When I modeled the response with WinISD it calculated an EBP of 148, which from my understanding is very much suited for a ported enclosure. The enclosure volume and port dimensions were all modeled using the T/S parameters of the drivers. Bass performance is not the problem that I am hearing with anyway. They have plenty of low range and kick. Its just the mids and highs that need work.

Thanks for the advice about the felt BTW. I will try that. I know that stuffing is not good for ported enclosures, but I had some lying around, so it was easy to test out.


I am positive everything is wired correctly, so I'm pretty sure that improper wiring is not causing the issue I am having.

Thanks for all the advice everyone.

Charlie
 
I'd forget EBP. It's a 1st order approximation (i.e. it means very little). FWIW, these days I won't vent a driver < 4in, even in multiples. It's rarely worth the effort, & most of the time the tradeoffs are too severe.

Generally vented boxes are lined rather than stuffed, but it's not actually a hard and fast rule, and stuffing can be more useful in some cases, particularly of a high aspect ratio where you have standing waves (Helmholtz math, i.e. de rigure BR alignments assume a uniform air-particle density in the box & no standing waves). Since you don't appear to have any problems with the bottom end performance at present, I'd leave it alone.

I'm 99.9% that you've simply run into severe HF interference that is a direct result of the configuration (i.e. the drivers too far apart & no Eq). Unfortunately, there's nothing that can be done about this just with the current widebanders short of changing the box & adding Eq. Off the vertical axis, there's nothing that can be done, period, without a longer array. That's just the way they work. Shifting to a 2-way, with a central tweeter is probably going to be the best option.

I wouldn't rule arrays based on your current experience, but there are some things to keep in mind: if you're going to move about, you need more drivers for a longer array, you need to keep the drivers as close together as physically possible, and you need Eq for the top end. If you run the box sealed, this can be passive, at the expense of efficiency, so active is preferable since you'll preserve the dynamic range.
 
Last edited:
This is what I would do.

1. Block the port with a sock to make it an aperiodic cabinet.
2. Turn it into a 2.5 way system so you have 8 of the drivers running bass and 1 as a midrange. You won't need BSC but you will have to isolate the single driver from the rest but that won't be too hard. You may be able to do this with a simple 1st order XO to start or maybe 2nd order if that doesn't work well.
3. Add a tweeter to the top of the cabinet. Get one that looks nice or build a little cabinet for it on top so it doesn't ruin the effect of your woodwork.

Easy fix, problems solved, and all that hard work is now rewarding instead of frustrating.