Optimal Subwoofer Port Orientation

[orbus]

Hi all.

A couple years ago I built a subwoofer using the Dayton 15" DVC driver and a 240 watt Dayton plate amp. I don't remember all the exact measurements now, but it's about a 225L enclosure with two 4" flared ports. I think tuning was to about 19Hz.

I've never really been entirely pleased with the performance of the sub. It just never hit the lows the way I was expecting it to - rolls off to nothing below about 30Hz. It's also got a pretty bad (like +10Hz) hump around the 50Hz point. Some of this could be due to room acoustics, which are not ideal, but testing with the SPL meter even right next to the box seems to indicate that it's really not hitting the lows like it should.

The only really unconventional thing I can think of that I did during the construction was orienting the ports to fire upwards. The box is taller than it is wide or deep. The driver is front firing, and the ports are oriented to fire up. Amp is in the back. I can post a pic if this description isn't clear enough.

It's been a couple years, but I find myself in need of a new project, so I'm thinking about starting over and building a new enclosure for the driver. Before I do that though, I want to ask if anyone has an opinion on upwards firing ports, and also if anyone has any other suggestions for my new enclosure. If I'm starting over from scratch, what would be the best port orientation? Forwards? Backwards? Down-firing? I don't want to have to build a third one. I'm hoping to get this one right, so I can start building my Natalie P's next. :P

[ThomasW]

Unless blocked by an obstacle the orientation of ports is of no consequence since the wavelengths involved are so long.

It defies logic that a box with a 19Hz Fb would have no output below 30Hz. Placing the SPL meter close to the ports should show significant output at 19Hz unless the port lengths are wrong.

A big hump at 50Hz is certainly a function of the room/placement not the design of the sub.

That said....

What's the budget for the new sub? How big is the room? What's the distance from the sub to the listening position? Finally what output levels are you looking for?

[---k---]

Does that plate amp have a highpass filter?

I seem to remember it does, but took a look at the manual and stuff and didn't see anything, but that doesn't mean much. If it does, that could be what was filtering out the low stuff.

Jeff B started a list of plate amp highpass filters, but doesn't have the 240W version listed. http://techtalk.parts-express.com/sh...ighpass+filter

Most likely the issue is the room, though having the meter right at the port should show bass.

Post a picture to show the end of the port with relationship to the back of the driver. I'm just curious.

[orbus]

Sorry for the delayed reply. Been kinda busy. I guess it's not entirely accurate to say there's "no" lowend output... It's just that it seems to be rolling off a lot earlier than I'd expect it should. I made this response graph for my system from 20Hz to 80 Hz. The other speakers were not disabled, but my receiver does have the speakers all set to small, so crossover is active. They might color the results a little bit at the upper end. I baselined the response at 70dB at 70Hz and then just went up the scale from 20, recording what I found.



The amp documentation doesn't mention a high pass filter, but from other things I've read online about it, it seems it has one with an F3 of about 18hz. I don't think that accounts for these results.

I can try to take a picture of the ports in relation to the driver, but I'll have to get the sub out away from the wall and take the amp off.

I'll try to hit some of the other questions: As far as building a new enclosure... I dunno - I figured budgetwise about a hundred bucks if I'm reusing the driver and amp. I was thinking maybe make it a little smaller - more in the 180L range. Make the ports fire out the back. I'd probably do some things differently after having built this one, since this was my first try. I'm not dead set on even building another one - I'd much rather build some Natalie P's. I just don't want to start on those without first getting this project right.

The room is fairly large... it's a townhouse, and the living room opens into the dining room and kitchen. High ceilings too. I'm not aiming for perfection - I mostly use the system for movies rather than music. I'd just like to get respectable output down to around 20-25 Hz and reasonably flat response above that. +10 dB peaks make some material sound pretty bad.

I've got a BFD that I've used at various times to try to tame some of the peaks and boost the low end up, but with only moderate success. If someone wanted to recommend some filter settings to try out, I'd be happy to give it a go.

[ThomasW]

Measure the output with the meter 1" from one of the ports

[---k---]

That graph doesn't look right. Do as Thomas says and measure the response 1" from the port and then another graph 1" from the driver.

If you can use a compute and Room Eq Wizard to do the graph, it would be much better. But if not, the old fashion way will probably be good. What tones are you using? And are you using a correction factor for your RS Meter?

[orbus]

I never managed to get Room EQ wizard working under linux, even though it's supposed to be cross platform, so that's pretty well out.

The tones are .wav files I generated using a program called sox. Just 10 second sine waves. I generated them from 15 Hz up to 100 Hz. Playing them from mplayer on a computer hooked up with an optical cable to the receiver.

I'm not doing any kind of correction for the RS meter at present.

1" from the port isn't really a problem since it's upward firing and I can just set the meter on top of the speaker, but I'll have to find something the right height to set it on to put it 1" from the driver.

Here's the graph with it about an inch from one of the ports, again baselined to 70 dB at 70 Hz (arbitrary value, but that kept the peaks from getting too crazy loud):



It does look a bit better. Maybe I just really need to rearrange the furniture and get the sub in the corner (it's up against a wall now, but not in a corner.) I'll try to get measurements close to the driver soon.

[orbus]

Here's the graph with the meter about 1" in front of the driver. The result is pretty interesting. Note that this one was baselined at 80dB at 70Hz instead of 70dB, so I could get some measurable output below 30Hz.

Note also that this one was done with the grill off. The others were done grill on. It's a cloth grill, made with speaker cloth that's supposed to be acoustically transparent, so it shouldn't matter. Still, thought it worth mentioning.

[ThomasW]

1" from the port isn't really a problem since it's upward firing and I can just set the meter on top of the speaker, but I'll have to find something the right height to set it on to put it 1" from the driver.

Here's the graph with it about an inch from one of the ports, again baselined to 70 dB at 70 Hz (arbitrary value, but that kept the peaks from getting too crazy loud):

If this measurement is accurate your sub is tuned to ~33Hz not 19Hz.

[orbus]

Hrm. Well the original model was done in Unibox. Since I've pretty well migrated off windows, and I can't get unibox working in openoffice (no matter what the unibox website claims), I can't go back and reference the model. I do know that the box is roughly 225L. I got WinISD running under Wine and modeled it up.

As the box stands now, it's ~225L with two flared 4" ports, both 18" long. WinISD says that should come out to a tuning of about 20Hz. I remember back when I first put this together unibox and winisd disagreed slightly, and the box volume's probably not exact. It's probably somewhere between 225 and 250. But either way, that's not anywhere close to 30Hz tuning. The model indicates to get a tuning of 30, the ports would have to be about 6 inches long (1/3 of what they are now).

EDIT: I'm pretty sure there are no leaks in the box either. I sealed all the interior edges with silicone. It should be pretty airtight. It'd probably hold water.

[Face]

Are you going to trust a program or actual measurements?

[JonMarsh]

Unibox is very reliable IF accurate driver and box parameters are entered. Since you don't have the windows system anymore, then I suppose you don't have the original file and cant post it for anyone to double check for you. My first suspicion is that your tuning parameters might have been correct using one flared port, to produce the box tuning you thought you have- two in parallel will nearly double the frequency of the box tuning- possibly a mistake interpreting Unibox or using it. I've seen people figure before that they could just add another port of the same dimension.

In this case, the giveaway is the notch in output of the woofer output at ~33-35 Hz; this is the minimum motion output point for the driver, and is pretty accurate in determining the actual box tuning. Possibly the box is a bit over stuffed, as usually this drop would be a bit more pronounced. But it could be leakage of port output into the mic, or lower box Q due to box leakage elsewhere- just thinking out loud about the possibilities to consider.

Stuff one of your ports as solid as you can with a foam or cloth plug and re-measure. There are commercial systems sold (SBS) with the idea that there are three ports in parallel and you can progressively stuff/block one or more ports to change the box tuning to your liking for the program material and room.

[orbus]

I may still have the file around somewhere. I'll take a look. And we do have windows at work, so I might be able to have another look at it there.

The box is not stuffed - the inside walls are lined with carpet pad (i.e. foam rubber), but other than that, no stuffing.

I have to go to work momentarily, so I didn't have time to do full measurements or make a graph, but I stuffed a blanket into one of the ports and took measurements at one inch from the remaining one:

20 Hz 80 dB
21 Hz 81 dB
22 Hz 81 dB
23 Hz 81 dB
24 Hz 82 dB
25 Hz 83 dB
26 Hz 84 dB
27 Hz 84 dB
28 Hz 85 dB
29 Hz 86 dB
30 Hz 86 dB
31 Hz 87 dB
32 Hz 88 dB
33 Hz 89 dB
34 Hz 86 dB
35 Hz 84 dB
36 Hz 82 dB
37 Hz 82 dB
38 Hz 82 dB
39 Hz 82 dB
40 Hz 82 dB


Also took a few spot measurements:

50 Hz 87 dB
55 Hz 80 dB
60 Hz 80 dB
65 Hz 78 dB
70 Hz 80 dB
75 Hz 79 dB
70 Hz 79 dB

[orbus]

On lunch break - thought I'd post these real quick. They're old pics, but this is the sub in question:

[ThomasW]

The consistent peak at 33Hz says there's a problem somewhere.

I used the online WinISD to do a quick look at the port tuning and it appears to be ok.

This means there's an issue with the measurement system, or there's a filter or setting somewhere that's rolling off the output. This could be in the plate amp, some setting (crossover or EQ) in the receiver/pre-pro effecting the output frequencies.

[orbus]

The chain is currently computer -> receiver -> BFD -> sub. But, the BFD is set to pass-through. It shouldn't be doing any filtering right now. I'll take it out of the chain and see if the result is any different. If that doesn't help, I'll try hooking the computer directly to the sub.

Got to go back to work now though.

[JonP]

Kind of overly obvious, but be sure to disconnect your speakers and run just the sub. How the speakers cross over to the sub could generate "issues". Might be that you're phased wrong and there's a cancellation between them. Getting them out of the picture will give you a cleaner look at the sub.

Later you can try with the speakers running and flip the polarity... that maybe you were phased wrong and they'd sum better by flipping the polarity between them. (or flipping the phase switch on the PE 240) If that isn't the source, it may be that your port tuning is higher than you thought it was...

Another thought... is it a new from PE amp, or did you buy it used, and could it be possible that it has been modded to change the highpass/bass boost filtering? If you were real careful not to smoke your soundcard, (using resistor probes, etc) you could use one of the testing programs to measure the amp response. Just be very gentle with that volume control! :E

[orbus]

It was a brand new amp from partsexpress (as of about 2 years ago). I'm the first owner, unless it was a return or something.

I just plugged the sound card directly into the sub amp and seem to be getting roughly the same response pattern: hump around 30, drops down around 40, another hump around 50. I was at a concert last night and didn't get to do any more testing till this morning. I'll try to get a graph up tonight.

Based on the spot-checking I've done, it looks like the issue is not at the receiver or with having the BFD (in pass through mode) in-line. That leaves the possibility that the computer is applying some sort of system-wide EQ (I can find no evidence that this is happening), or that my test tones are colored somehow. I'm not really sure what to do about the former, but if someone wants to suggest some known-good test tones, I can try those. Even better, if there are some I can burn to cd, maybe I can pop them in a cd player and take the computer out of the equation altogether.

JonP: When you talk about testing the amp response, did you mean by plugging directly into the sound card, or something else?

[ThomasW]

You're one of the very few people these days using test tones, since most are using the sweeps built into software measurement programs.

None of the stand-alone tone generators I'm aware of work with Linux.

Perhaps this software will work for you

http://gspeakers.sourceforge.net/screenshots.html

The PE amps usually have a ~18Hz hi-pass.

[---k---]

Ethan Winer has test tones on his realtraps.com website that are made to be burned to a CD. http://realtraps.com/test-cd.htm These aren't corrected for the meter though. Rives Audio sells a CD that has the test tones on it that are adjusted for the Radio Shack meter. http://rivesaudio.com/software/softframes.html

If you have access to a Woofer Tester, you can generate an Impedance graph very easily. This will show you the true tuning of the sub and might give some more clues. I believe that there are also methods to test impedance using Speaker Workshop and a computer.

[orbus]

I'm not going to worry too much about correcting for the meter at this point. It would be nice, as I know the radioshack meters aren't supposed to be terribly accurate in the low frequency ranges - but most of the information I've seen suggests it shouldn't be off by more than a couple dB even at the very low end. Also, I've read that trying to come up with a universal correction scheme for the radioshack meters is a fruitless exercise, since each one is different. Not sure how accurate that is.

If I could find a software measurement program that worked in linux, or even worked acceptably under wine, I'd be all for it. Room EQ wizard is written in java and is supposed to be cross platform, but I couldn't make it work last time I tried. I may try again tonight or tomorrow. Making graphs manually with test tones is a pain in the butt.

I'll burn the test tones you mentioned to cd and give that a go with a cd player - take the computer out of the equation.

[orbus]

Only cd player I had handy was an xbox 360. I burned the test tone files from realtraps.com to a cd and hooked the xbox directly to the sub inputs. Same response pattern. Note that I still have not unstuffed one of the ports - it's been stuffed since yesterday morning, when I first said I was doing it. Any information after that is with one port blocked.

Testing with the meter next to the remaining open port showed a response right at about 80dB at 20 Hz and as it went up through the frequencies got up to nearly 90dB in the low 30's. Then went back down. I didn't test any more extensively than that, as the xbox was in an inconvenient spot and I was having trouble seeing the SPL meter and still getting the xbox to see the remote.

I think from this I can conclude that the tones I'm using are probably okay, and the computer is not appreciably coloring the response.

That leaves the amp, or a problem with the structure of the enclosure itself. I've been trying to avoid taking the amp off as I don't like screwing and unscrewing from MDF more than necessary. Still, if I must, I must. I'll try to get some shots of where the ports are in relation to the driver. The port tubes are sticking down from the "ceiling" of the interior of the box, and the ends are probably about level with the back of the driver magnet, if that tells anyone anything. They're probably 3 or 4 inches from the back of the driver.

[JonMarsh]

If the port output peaks at 30 Hz, then that's the box Fb with your current construction and tuning. Block one port and the peak frequency should drop significantly.

[ThomasW]

RS meter corrections for the old analog unit

6Hz add 11.5dB
18Hz add 8.0dB
20Hz add 7.5dB
22Hz add 6.5dB
25Hz add 5.0dB
31.5Hz add 3.0dB
36Hz add 2.5dB
40Hz add 2.5dB
45Hz add 2.0dB
50Hz add 1.5dB
56Hz add 1.5dB
63Hz add 1.5dB
80Hz add 1.5dB
89Hz add 1.5dB
100Hz add 2.0dB
111Hz add 1.0dB
125Hz add 0.5dB

[BOBinGA]

What are the room dimensions? 33 Hz has a 34ft wavelength. If this is the long dimension of your room, I would suspect a standing wave is causing the increase at that frequency.

[orbus]

It's kind of an open floorplan, and there's furniture and things in the way. Still, I did some measuring and drew up a quick floorplan in sketchup.



The ceilings are about 20' tall. The sub is the box on the right. That's the back wall of the house. In the middle are stairs (14' from back wall). There's sort of a little hallway in front of the stairs, and the front door would be in the lower left. It's not a great drawing, but hopefully you get the idea. No dimension is anywhere near 34ft.

I'm going to try to get Room EQ wizard going in a bit here.

EDIT: I got Room EQ Wizard running, but I find myself short a few cables. I'll have to pick some up when I go out later. I'll see what it can do for me tonight or tomorrow.

[orbus]

Okay, I found an extra cable and I got Room EQ Wizard running. I'm not sure I got all the settings and everything perfect, but the graph looks about right. I don't believe it's applying any kind of correction for the RS Meter at this point. Looks like I'll have to find a correction file for it?

These are the first couple graphs I got out of it, with the meter an inch from the port.

One port stuffed:



Both ports open:



I think the results are pretty interesting. It does look like the tuning is in the right place. I'm going to experiment a bit more - if there's anything anyone wants me to try, let me know.

[ThomasW]

That's more like it.

You get the cal files the same place you got Room EQ Wiz.

[BOBinGA]

The long diagonal is 33.5 ft if I did the math correctly. If this is a continuous space, it could be the source of the peak. Also notice that you have a secondary peak at 45 Hz. 45 Hz wavelength is 25 feet which corresponds to your other major dimension. Nonetheless, it looks like Room EQ Wizard is helping a lot. Have you tried moving the Sub around and seeing if this affects the peak?

[orbus]

Not yet - that's the next step. There's a lot of furniture in the way, and the thing's pretty big. I'll see what I can do. I'm going to try moving it into a corner next.

[---k---]

If the measurement was taken 1" from the port or driver, the direct response should swamp the room reflections. TO be certain, maybe get the sub out into the middle of the room, away from the wall for the test measurements.

A measurement 1" in front of the driver should show the port tunning.

[---k---]

Unibox models:

225L
First = (2) 4"DIA, 18" long ports
Second = (1) 4"DIA, 18" long port.

The models look good. Don't know.

[orbus]

I still haven't gotten the Radioshack meter correction files (my hometheatershack registration is giving me trouble - may have to re-register). However, I pulled the sub out into the middle of the room and got these two graphs:

By port:


By driver:


Humps are essentially gone. It must be the room.

EDIT: This is with it in the middle of the room, same place, from my primary listening position on the couch. Not so pretty.

[BOBinGA]

Very interesting! Those graphs looks almost identical to Ryan's Unibox models. So now the question becomes: How do you rearrange the furnaiture to keep the sub where it is now?


Edit: Just saw your edit. The hump caused by the room diagonal is still there. That's going to be a problem unless you can soften it somehow. Is there something tall you can interupt it with like a china cabinet. Otherwise, move the couch to a location as far off this axis as possible. Try measuring at different locations and see where you get the least peak. Room nodes are hard to defeat, so you usually have to move things around until you find what works.
-Bob

[orbus]

Can anyone help me out with the radioshack meter calibration file? Hometheatershack is really, really picky about e-mail addresses. To the point where I don't have an e-mail at a domain they'll accept anymore. I used to have an account with them, but it's apparently been deleted, either from disuse or because it was registered via yahoo mail, which they no longer accept.

The meter I've got is the RS 33-4050. I have the .cal file from the 2050 (older style analog meter) but I gather the correction needed is different.

[BOBinGA]

I found this link in one of their forums:

http://www.audioxpress.com/magsdirx/ax/addenda/media/koya2811.pdf

It discusses the various models including your 33-4050. I use an old 33-2050, so I found this interensting as well.

[orbus]

Thanks for the PDF - interesting read.

I found correction values for the 33-4050 posted in another forum, so assuming they're right, I'm now correcting for the meter.

I re-measured at the port and driver, and then took an average:



The dip in average output around 20Hz is a bit discouraging, but the F3 on the average graph is still about 25 or 26 Hz, which is respectable, if not quite what I was aiming for. I'm going to start shoving furniture around and see if I can find some sort of workable solution. The couch has been up against the wall since I moved in - I'm thinking it may be time to move it to the middle of the room to face the tv more directly.


EDIT: Maybe I'm misunderstanding how the calibration file works. I see the line for SPL meter offset on the graph, but when I take measurements with and without the correction file applied, I'm not seeing any difference in what the graph looks like. If it's actually applying the correction values from the file, the lowend response should be much higher, since the file says at 18Hz the meter is down a whole 8dB.

EDIT2: Still not too sure on the cal file. I also think from looking at the unibox models that an average of port and driver response does not represent overall response.

I've moved the furniture all over the place and made a real mess. I've tried 4 different sub locations - they all have different responses, but none of them are anywhere close to what I'd call clean. That 30 Hz range hump is really hard to overcome. Starting to get a bit discouraged, so I think I'm going to pack it in for the time being and do something else for the afternoon. If anyone has any suggestions, let me know.

[---k---]

Averaging the responses is not correct. You just add them together. Visually, they look pretty flat and correct.

Some rooms are just don't like subs, mine included. One way to get around this is to use multiple subs. I know probably not the answer you're looking for.

Take a look at some of the Harmon Whitepapers

http://www.harman.com/wp/pdf/multsubs.pdf

http://www.harman.com/wp/pdf/LoudspeakersandRoomsPt3.pdf

Your BFD can help some problems, but not all.

[orbus]

Yeah, not the answer I'm hoping for, but.. I will say that moving the sub does seem to have helped the second hump around 50Hz. This also means that I didn't do anything terribly wrong in the construction, which means rebuilding wouldn't help. That's useful info. If I can tame it down to just the one big peak, maybe I can use the BFD to flatten that out - I haven't really gotten into experimenting with it and REQ Wizard yet.

The good news is that if I don't have to rebuild the sub, I can think more seriously about my next project - building a pair of Natalie P's. I'm hoping to eventually construct a full 5.1 set of speakers, but I'm going to start with some tower format Natalie P's for mains.

I'm going to keep playing with the positioning and stuff for the sub - just not any more today.