Subwoofer doesn't sound right at the high end (>50Hz)

[neRok]

My DIY subwoofer isn't sounding the best and I'm not sure what the problem is exactly (how to identify it via tests/measurements). The problem is in the higher bass region, above 50Hz, and is affected by where I set my DSP LPF. The problem is it has a sort of thump sound to it that doesn't sound right, and it only occurs on "sudden" bass notes like drum beats, but not on "sustained" bass notes. The thump sound stands out and makes it easy for me to localise the sub's sound even when using 90Hz 8th order LPF, which seems a bizarre thing to be occurring.

To pre-empt some possible responses: I designed the sub in WinISD as a vented box with an extended bass shelf tuned to ~24Hz. It's a 120W driver with only a 90W amp, and WinISD calcs the max excursion at 6mm (with 90W) vs manufacturer spec'd Xmax of 9mm - so I don't think it's a problem there. Additionally, max excursion is at 37Hz, and at 60Hz it's down to 3mm, and 1mm at 80Hz - so excursion is even less of a potential problem in the upper bass region. Port velocity is max 10m/s at 24Hz, but only 1m/s at 60Hz. Also port gain is like -16dB at 60Hz, so I don't think it's a problem there either. Putting my ear near the port also confirms no problem from it (and the port is on a different face to the driver). So the problem definitely seems to be coming from the driver.
Edit 3: Forgot to mention, I don't think there is anything wrong with the driver physically/mechanically. It moves fine when I press it by hand. I'm going to drag it out soon and look over it for issues like loose screws and tinsel slap.

A possible source of the problem could be group delay introduced by the LPF? In the WinISD screenshot below, I can see 2nd order ButterWorth @ 90Hz is a rather smooth curve, 4o puts a little hump in it, and 8o is quite a big hump. Could this be the problem I am hearing? Because when I change these settings in DSP on the fly (and I'm listening to the sub only - no speakers), yes the problem becomes a little better, but it doesn't become perfect, and I wonder if it only seems a bit better because the sub is now playing more sounds around 200Hz? Because when I stick it on 8o@60Hz the problem is reduced even further without hearing anything in that 200Hz area, and yet the group delay is apparently even worse. So what gives?
Edit 2: FYI the song sample I am testing against is from 3:20 to 3:40 on this song. Using 2o@85 I can hear heaps of the jingles and stuff from the song. With 4o I can only just hear some non-bass notes, and with 8o I can only hear bass but the thump issue is more pronounced.


I have used UMIK-1 and REW to EQ this sub in the past. Looking back at some measurements I've done, the following seems to be the last one I did with EQ applied and LP 8oBW@85Hz. I'll do some more soon to see what the current state of things are, but nothing distortion wise stands out to me on that plot. But maybe the peak at 140Hz is a problem? Perhaps I need a PEQ just to knock that down? If that is the problem I am hearing, perhaps reducing the LPF to ~60Hz has been enough to move away from that peak, and alternatively reducing the order increases the sound around that peak so that it blends in? So maybe that peak is the problem?
Edit 1: Tried -10dB + 8-12Q at ~150Hz but it didn't seem to make a difference.


Anyway, any advice on what/how to test would be appreciated. One thing I wonder about is doing measurements in REW with sweeps, those seem more like the sustained bass notes I mentioned at the start which aren't a problem, and not so much like the sudden and impactful bass notes like kick drums that are the problem. Is there any way to better simulate quick notes in REW?

And further to this, whilst excursion doesn't seem to be a problem in WinISD, could "dynamic" bass be causing a problem (eg a sustained bass note being played with a drum hit over the top)? I think probably not because this is happening at rather moderate volume levels (the music I'm playing from PC is at like -20dBFS after DSP and considering the volume setting), so I should be miles away from xmax. But perhaps there is some other property of the driver that is related to dynamics?

 

[BR52]

For me it sounds your cut of ( low pass ) is not hard enough. I had a similar problem and could solve it with a harder one. Because some woofer can have a much higher output in higher frequency than expected.

 

[sigbergaudio]

This is a very weak subwoofer, I suspect your are bottoming out the driver. A harder high pass filter around the tuning frequency of the port (you have that right?) may help a bit.

 

[neRok]

This is a very weak subwoofer, I suspect your are bottoming out the driver. A harder high pass filter around the tuning frequency of the port (you have that right?) may help a bit.

It is just a little 10", but according to WinISD it does 103dB above 30Hz @ 1m and full power, which should be more than enough for near field listening in a small room (and it does go good otherwise).

I'm not sure exactly what HPF the amp has, but it apparently does have one, so I presumed 20Hz 2nd order. I then have an additional 2nd order HPF at 25Hz in DSP, to get the curve shown below. Cone excursion below the tuning frequency is well under control (according to WinISD). And actually, when I am playing the song mentioned at moderate volume, there is very little cone movement evident.
Edit: In DSP I actually had 8o HPF @ 24Hz. I changed it on the fly to 40Hz and 60Hz and heard no difference to the problem.
Additionally, I quickly played that portion of music with REW RTA open, and even with 24Hz HPF there was very little picked up below 50Hz, which suggests the song doesn't even go that low.

For me it sounds your cut of ( low pass ) is not hard enough.

If by harder you mean 2nd vs 4th vs 8th order filter, then I tried that.

 

[BR52]

It is just a little 10", but according to WinISD it does 103dB above 30Hz @ 1m and full power, which should be more than enough for near field listening in a small room (and it does go good otherwise).

I'm not sure exactly what HPF the amp has, but it apparently does have one, so I presumed 20Hz 2nd order. I then have an additional 2nd order HPF at 25Hz in DSP, to get the curve shown below. Cone excursion below the tuning frequency is well under control (according to WinISD). And actually, when I am playing the song mentioned at moderate volume, there is very little cone movement evident.

View attachment 291269


If by harder you mean 2nd vs 4th vs 8th order filter, then I tried that.

Yes was only an idea. But I guess you will find more help here with more experienced guys

 

[neRok]

Just did a quick sweep with the mic sitting in front of me on desk (so not perfect position, but close enough I would presume to see major problems). "HP_8o_24 LP_8o_85". Distortion and overall SPL still looks about the same as I measured in the past. Also I screenshot the SPL and phase tab. There is a bit of "plateau" in the phase at 100Hz. Could this be an issue?

 

[voodooless]

Can you do a sweep of the whole system. Likely the issue is with the integration.

 

[BR52]

And a sweep subwoofer only 20Hz to 1.5 kHz with high level

 

[neRok]

This is a measurement from the last time I mucked around with it (vs harman target curve I think that is). The speakers are causing me the most grief because there's something wrong with the low mids (and the speakers should be capable as they are Kali In-8 V2). It might be because I have them sitting on some old tower speakers because I don't have stands, and I wonder if they are causing more problems than just the SBIR issues I am also contending with (and possibly this issue is showing up on the spectogram from 150-500Hz). I'm trying to whip up some basic timber stands now, but it's 5pm, so I better get to it.


But also, I can hear this subwoofer noise with the speakers off. It is something specifically with the subwoofer IMO.

And also just to address another potential problem - the couple of EQ adjustments I have made to the sub (via REW) are all -gain, no boosts, so there should be no issues there either.

 

[alex-z]

You generally want to keep group delay at 1 cycle or less, those steep low-pass filters are definitely not helping you out. The Kali IN-8v2 start rolling off around 45Hz, so I would probably select a 2nd order 60Hz high-pass.

How is the time alignment between your subwoofer and speakers? Worth asking, because if they are off by a few milliseconds that can degrade transient response without being obvious in the frequency response.

 

[sigbergaudio]

And you are not able to reproduce with sine sweeps, only music?

 

[neRok]

And you are not able to reproduce with sine sweeps, only music?

So I did some sweeps with only the sub on and my ear up close. With 4oLP at like 80Hz there isn't anything tremendously obvious, but there may be a sort of simultaneous ringing sound above midway of the sweep. Trying a 100Hz LP and sweep, and actually there was a vibrating sort of sound near the end of the sweep, so I will have to investigate tomorrow (didn't get time to tear the sub down today).

I finished off some temporary speaker stands and they have improved the speakers bass response. I'll get onto that next post. But it seems they are putting out bass down to 50Hz now, which sounded pretty good on the song from OP (which evidently doesnt have much deep bass). So then I had the idea to apply 40Hz HP and ~80Hz LP to the speakers and run the same sweep, because maybe the sub doesnt have a problem and actually I just dont know what it is supposed to sound like?! But no, the sweep sounded rather smooth like I expected. No "ringing" effect.

BUT doing this has revealed a new problem. With the speakers set to this "bass only" mode, there is a noise coming from the woofers of both speakers that sounds like if your nail was tapping on the cone/dust-cap. When playing a song, it seems consitent, as in it happens on the same notes time and time again, and it would happen a couple of times during a bass sweep. Very strange. I am now wondering if my interface has issues? It is PreSonus Studio 68c and I have it connected to the speakers with balanced TRS cables (sub also connected to it with TS to RCA cables). I'm going to have to investigate tomorrow, but I only have a Topping RCA DAC to test with, and previously I found it made the speakers hum louder than they do otherwise, so it might not prove anything.

How is the time alignment between your subwoofer and speakers? Worth asking, because if they are off by a few milliseconds that can degrade transient response without being obvious in the frequency response.

This is something I need to work on. The problem is I don't know where to start (or really what to do lol). The sub has an analogue plate amp, whereas the speakers have DSP amps. That DSP must be adding delay to the speakers? But how much? Because if it is considerable, could it actually be more than the subs group delay? Or is that over thinking it? It probably is, and so I imagine the problem would appear in a full sweep? I did see a video that said to go to GD tab and then "generate minimum phase", and doing that does reveal a large step up where the sub comes in. So then I delay the speakers by roughly that amount, a tweak it slightly until it gets loud. Then I have measured a sub sweep and a seperate speaker sweep, added those sweeps, then done a combined sweep, and comparing the two is close. So I think that means the timing is ballpark good?

And 1 other thing I don't understand is if it is necessary to measure with "acoustic timing reference" or not, and if I am measuring the sub only (speakers muted), then how does it even do the acoustic timing reference, because that seems to be a rather high pitched tone that wouldn't even make it through the subs filter?

 

[neRok]

To explain why I've been crossing where I have and why my speakers have more bass just now, it's probably easiest to quickly give my speaker adventure story from the beginning. So I started with some B&W bookshelfs, then got a UMIK and saw pretty terrible results below 1kHz. At this time I didn't understand how the room and speaker positions play such an important role in that area. But the measurement on the net for those speakers wasn't so good, and at the volumes I was listening they were distorting. So I decided upon Kali IN-8 because their measurements have far better distortion characteristics, and being 3 way and stuff I thought it would be a winner. But when I plonked them in the same place as the B&W (just a bit lower to be at ear height) and did a measurement, they weren't good. In fact, they are probably worse, because below 100Hz they had more cancellation. So I was quite disappointed.


At this time I was working on my DIY subwoofer, so I put the Kali aside and used my old 3 way floor standing speakers in the corners of the room, on bricks to bring up their height. It wasn't half bad. I think they punch harder with 2x8" each than even these Kali are right now. I tried to measure them "quasi-anechoic" in the middle of the room, and they were actually quite flat in the straight on gated response (so above 250Hz). The Kali coax is superior though.

Anyway, once the sub was done (+ a few weeks because I'm lazy), I went back to the Kali, but because I knew they weren't too good on the desk and I didn't have stands, I just put them on top of the tower speakers. I tried them in a few different locations over the months (like narrow and close, wide and close, narrow and far, etc). Earlier today I had them wide+far, which are the old measurements you see above. Always in the measurements I could see that huge peak at ~120Hz, a null at 100Hz, a variable amount of response in the 50-80Hz range, and then another peak at 40Hz. I am of the understanding you can't EQ a null, so I figured EQ'ing the peak at 120Hz down was a good idea, then whatever slope you end up with down towards that 100Hz null, I would use that as my HP. And so that's why I'd kind of given up on getting bass out of the speakers, because I presumed the room width vs speaker width just meant there was going to be a problem in that area.

But even after this, something hasn't been right with something in the lower midrange area I think it might be. Lots of the frequency response coming out of these Kali sounds great, but somethings not right in a particular spot. I had wondered if the tower speakers could be the issue, causing some kind of baffle effect, or absorption effect? My other thought was it could be the speakers height in the room, because the mid point of the drivers is basically the mid height of the room. So for the last week I've actually had the speakers on the side, like this (taken earlier).

But here are the test stands I whipped up just before;

I might have made them too tall, because they are basically exactly the height they were when sitting on the tower speakers lol.

And running a quick measurement with the speakers in this position shows an improvement in that 50-80Hz range (also they sound better listening to them right now). There is still something wrong higher up in that low-mid range, but I think it is slightly improved. Also I have the same EQ on as before, so that might need some work too (back to square 1 tomorrow I guess). But here is the measurement taken now;

So blue is with no EQ at all, when they were sitting on the tower speakers, roughly in the same corner positions. Orange is a measurement from now, also with no EQ. I'm glad the 50Hz peak has come down and that 50-80Hz is basically flat. There are some other changes when comparing blue and orange too, which I guess confirms that having the Kali sitting on the tower speakers was affecting them...? But also the changes aren't huge, so maybe I just have the mic in a slightly different spot. Who knows?!

 

[neRok]

Keen observers might notice the curtain I added to the brick wall on the left, lol. I haven't got around to doing anything with actual room treatments, but one youtube vid I watched said it is good to have symmetry in a room, and so I put that curtain there (had the bits on hand). I'm not sure it made a difference.

But also keeping with my cheap and lazy theme, I happened to have some large TV boxes at my disposal the other week. So I mucked about with them and some blankets trying to improve the rooms reverb, because it was bad. Here's an example of that;

In the end I settled on stacked boxes with dooner over the top on the back wall (behind me). It has made a huge difference!

It has removed all the problematic echo/reverb from the room. The difference is so stark that the first few times I walked in to the room I was like "WTF". Everything sounded different including my footsteps, chair noise, keyboard noise, etc.

And the measurements show it. You can scroll up and see the current spectogram in post #9. Notice how above 1kHz that the blue tips barely go over 400ms. Well here is what they were, going well over 500ms!

 

[Spkrdctr]

Usually if you hear a tapping sound from a sub at you normal listening position, you have a woofer problem. I would suggest you not put your ear up to the sub driver. It doesn't really tell you anything useful, unless the woofer is damaged somehow. The sub is a very weak sub. I see you have a nice boost in the 25hz area. Doing that will often times give you a thump which is really a one note bass effect. If you can, I would try to lower that 25hz area a lot and see if your problems go away. Just giving you some off the cuff advice. I hope you get it sorted out.

 

[neRok]

I decided to go for the YOLO fix and have hacked the box up to fit a 12" woofer I had laying around. TLDR is that it basically sounds the same, so I think my problem is just that I don't like hearing those noises at 100Hz and above coming from my subwoofer. Possibly those noises would be more pleasant if I put some stuffing in the box? But AFAIK that would be to the detriment of the low bass, so I will just have to cross as low as possible (and put more effort in to speaker location for their optimal deep bass response).

The long story is that I dragged the sub out to sit it next to me and test. I did some measures with the mic about 0.5m away at 92dB. I turned it up another 6dB and did some measures, and whilst the distortion did rise a bit, nothing "jumped out". I turned it up another 6dB but must have caused the amp to clip. So I turned it back down and then use tone generator at the various distortion peaks, to see what was what. No obvious problems with the woofer or box presented themselves, but at 50Hz the plate amp was making some terrible noises. I couldn't work out if it was the actual plate vibrating, or the electronics buzzing horribly.

That plate amp has given me all sorts of trouble, so I have removed it and temporarily fitted some offcut wood over the hole. Then I connected up an old 2chan emotiva amp I had laying around, and retested the sub at the same volumes. The sweeps and sound were unfortunately more or less the same. So even though that plate amp has problems, it wasn't evident in sweeps. Possibly the woofer could have needed tightening, but I wasn't able to access the screws due to an ill-conceived decision whilst assembling the sub. [But actually, after cutting it out and testing the screws, they all did easily tighten a little more (~1/4 turn), so I guess the gasket did compress a little more over time, but actually I don't think it was enough to matter because it felt secure.]

So the only thing left to do was to cut out the 10in woofer and replace it with a 12in. WinISD shows the box is a bit too small to get a good "extended bass shelf" effect going, but with the extra SPL afforded due to its larger size and power handling (300wrms), it can be EQ'd the same. Measuring at the same SPL, the response is basically the same, and the bass in my song sample is about the same. Distortion however has dropped off considerably above 50Hz, where previously it was 1% min with peaks up to 5%, now it is rather consistent at 1%. There is however a large distortion peak either side of 40Hz, which was on both the 10in and 12in, but has gotten worse. I'm not sure if this is something to do with the box or the way the bracing or port is fixed within the box, or if it is the room (the ceiling does make noises around this frequency). It is below the HPF frequency I've been testing at too, so it could be "nosie". So it's a big of a head scratcher that I'll suss out when I get the sub installed proper, but also it's not really relevant to the "problem" I've been looking at in this thread (the sound up around 100Hz).

Measurements;


Out with the old;

And "on" with the new

 

[ozzy9832001]

I don't think the problem is with the sub, but with the lower bass causing reinforcement on the high frequencies (160 to 300). Your peak SPL from 123 to 350ish is significantly higher than the rest. There is a lot of excess energy present. The 200hz frequencies are awful and for certain genres of music can cause a sound that is similar to a vibration or a chair being drug across a floor. EQ that region down or you'll probably have to move speaker position. I had a similar problem and the only fix was moving the position.

 

[neRok]

I had a quick attempt at EQ'ing the system last weekend, and I've had a few listening sessions since. The speakers are now pretty flat to ~50Hz, so I EQ'd them to the Harman room curve using their natural roll-off as their crossover point, and then I EQ'd the sub response/LPF to match ensuring that the slopes added up properly. Initial impressions were that the problem was pretty much gone, but also the sub is doing very little now on much of my music, so maybe it's not really fixed? And actually, I did get the impression it was still occurring on certain portions of music, but maybe I was "hearing things".

I had a thought though, whereby it is actually quite cold here (relatively) at the moment, so I wonder if it is the cold dense air causing the "problem"? I don't have heating, so it's ~15°C in my PC room. I actually built this DIY sub ~Aug last year (and didn't use it much for the first month or so), and so by then the weather was already warming up for spring, so this is my first time experiencing a good subwoofer in the cold. So could the temp be making a difference?

Lastly, I intend to try EQ the system with a higher crossover point, any different slops (I think I'm on 4th order at the moment), to see what that is like. That may or may not increase the problem. But I've had an experience that raised questions whilst EQ'ing the speakers, so I'm going to make a thread about that first.

 

[ozzy9832001]

Use REW's RTA feature and place the mic close to your listen position. Play a part of a song that doesn't sound right. See where the peaks are. That will give you a real idea for how the sound is completely interacting with the room. Sometimes it's a small area of frequencies which are causing the issue. If everything is about 80dB, but say 220 hz jumps to 92 dB, then probably 220 needs to be EQ'd down.