I think I have a problem.

[Conrad.]

Hi all,

I've been reading this forum for a while now and I've found it hugely informative and excellently written by people who clearly know their stuff.

Between you all you've managed to convince me that my system could (and should) sound better. With this as my goal I've bought into the whole REW/BFD solution and I have some findings that I'd like your opinions on.

My set up is an HTPC which is passing a decoded signal out to my Yamaha RX-V1500's pre-ins. THe sub output from the multi-channel outputs on the Yamaha is being split to a pair of Velodyne SPL1000 subs. These subs aren't co-located, one is forward facing just inside the right main and the other is in a deep bay at the back of the room, firing diagonally across the back of the seating area. I've tried to illustate this in the attached plan. The room dimensions are about 22ft x 14ft at it's widest/longest. THe bay is just over 5ft deep by 8ft wide and the ceilings are about 11ft high.

My first pass with REW produced the attached results. Initially I thought that this was excellent and that REW couldn't change it much because it didn't need changing but that drop from 20-40Hz (or a huge hump at 40Hz, whichever way you want to look at it).

I was thinking of dropping my target level down to 64dB and seeing if I couldn't EQ the hump down rather than try and boost the low frequencies (a real no-no, apparently). Is that the right approach to take? Should I focus more on placing the subs differently? I don't have a lot of options on that front unfortunately. Or should I just sell both subs and buy a Monolith?

I know that 2 good subs don't equal 1 great sub and that I can't expect my house to shake with the pair that I have but they should go way below 40Hz, right?

Any help would be appreciated. I'm posting this on AVForums as well, and I'll try and transfer interesting points and comments between the two.

Thanks in advance.

Conrad

[Wayne A. Pflughaupt]

Welcome to the Forum, Conrad!

It’s going to be difficult to properly integrate separate subs in a non-symmetrical room like yours. As you can see, if you don’t do it right you can’t hope to achieve good response. I agree that these subs should be able to get below 40 Hz. It’s probably your placement that’s causing them to brick-wall at 40 Hz.

At this point I’d deem your current response as essentially unequalizable. Not even considering the narrow hole at ~95 Hz, you have a nearly 30 dB spread between your worst peaks and depressions. That needs to be narrowed to something like 15 dB or less if you hope to achieve smooth response.

I suggest experimenting with different locations, using only one sub, and find one that gets the best reading. Then you can add your second sub to the same location. The first place I’d try is that corner at the top right of your picture, followed by the one at the bottom right. We’ve seen this a lot here recently, people showing is an abysmal graph that’s beyond help, with subs placed in some weird location - center-wall, etc. They move it to a corner, and there’s a huge improvement – response we can work with. Here’s one – look what happened to his response when he moved the sub to the right (i.e. best) corner.

Regards,
Wayne

[Conrad.]

Hi Wayne, thanks for getting back to me so quickly. I'll try EQing just one of the subs first and I'll try it with it's phase setting in both the 0 and 180 positions. I could move the second sub into the bottom right corner of my layout diagram, if I can get one of them working first.

I do have another question though. Currently, when things are going through my BFD I'm barely getting one light lit up. Even on really heavy bass scenes I get the third green flickering at most. As that's the case would it be safe to add some boosting filters to bring that 30Hz dip up a bit?

Again, thanks for your time and knowledge.

[Wayne A. Pflughaupt]

Currently, when things are going through my BFD I'm barely getting one light lit up. Even on really heavy bass scenes I get the third green flickering at most. As that's the case would it be safe to add some boosting filters to bring that 30Hz dip up a bit?

I wouldn’t do it. “Some” boosting isn’t going to do much, since the drop-off is severe and deep. Go ahead and get some readings in those corners – I’m betting that will improve your extension considerably – for “free,” with no boosting required.

Re the BFD’s low meter reading, can you increase the sub output from your receiver? It would be good if you could get it up some more.

Regards,
Wayne

[Conrad.]

My subs are on +3.5 and the LFE setting is at 0 on a -10 to 0 range. I could crank it up a bit but, previously, It'd just mean turning down the gain on the subs themselves (at about 1 o'clock). Once I've got the BFD in the chain though I guess I'll be cutting frequencies and that are domainating or, probably, overpowering that nice, clean, defined bass.

Actually, hang on, I'm not sure if I know what I'm talking about! I'm putting the BFD into the HT chain, not the music chain. The HT chain is already decoded by the HTPC by the time it reaches the pre-ins. I can drop the volume of the other channels and increase the sub channel though, that'll just mean I need to turn the reciever up. It's odd, my reciever which has multi-channel ins and pre-outs is still able to affect the volume when it's, apparently, performing pass-through. But that's something for another day.

The short answer is that all channels are running at the same volumes from the soundcard.

[Conrad.]

Ok, some serious improvements have been made!

I started by turning sub2 (the one in the bay) off. This made an immediate improvement to the frequency response. That ~45Hz mountain almost disappeared. I tried one sub in various positions including the corners Wayne suggested and, surprisingly, the smoothest frequency response came from the position it was in originally.

I then decided to be a bit bold and hooked up the second sub, trying them together in different locations, co-locating and taking many measurements. The best I could get was with sub1 where it was and sub2 now sitting in the corner next to it. This produces a graph that looks like the uncorrected attached.

I think I should probably have gone through some recalibration at this stage as it's clearly running a bit hot. Instead I upped my target within REW and added some filters. I've managed to generate the attached corrected file which, I think, should sound pretty good.

Full range sweeps to ensure good integration are a bit beyond me at the moment. I ran a couple but couldn't decipher them so I'll use my ears for the moment. We're currently watching the LOTR trilogy; we did FOTR last night, it's Two Towers tonight and then ROTK on Saturday. That should give us a nice comparison.

Any comments or further suggestions? I'm going to play some of the scenes a couple of times tonight, with and without the BFD in place to see if it makes a noticable difference.

[brucek]

Any comments or further suggestions

You done a lot of aggressive cutting and have left quite a hump at ~150Hz. You need to back off on that cutting to keep a smooth drop. You need to add your mains and see what that 150Hz area looks like. Remember the mains will be overwhelming that area in their addition and you may have a noticeable peak.

I like the look of the pre BFD response better than the after. You're using way too many filters. Not neccessary. I would just lower some of the peaks and then take a stab at adding the mains and see if you can get a smooth transition...

brucek

[Conrad.]

Thanks Bruce.

I was kind of hoping someone would come back and say that I could EQ the subs by just dropping their volume, no real need for a BFD. What are your thoughts on that?

I've played some music I'm familiar with (Daft Punk's Aerodynamic) and the new location completely overwhelmed the mains. I've turned them down quite a lot (from about 1 o'clock to about 11 o'clock and they sound much smoother and more integrated. That kind of explains the shift in target level from 75 to 82dB I guess. I'm going to get another shot at running REW at the weekend and I'll try and involve the mains.

Now, to do that, I need to increase my sweep up to what, about 300Hz? I'm assuming that I don't need to go up to 10,000Hz or anything as the subs are well out of range by that point; right?

[Wayne A. Pflughaupt]

Great news there on your progress, Conrad, and happy to see the much improved results! Extension towards 25 Hz is much improved.

The best I could get was with sub1 where it was and sub2 now sitting in the corner next to it.

Actually, SVS’ Ed Mullen (who is also a fellow Shackster) has previously noted that a sub a few feet from a corner achieves the same coupling you’d get with it being in the corner (as for as output and extension goes) and that a second sub in the nearby corner can get smoother response than you’d get from putting them both in the corner. Naturally, every room is different, but once again it’s shown that Ed knows his stuff.

Yeah, and what’s up with that 150 Hz thing? Isn’t your crossover working?

I agree that you’re probably applying too many filters, and possibly cutting everything too much as well. I suggest raising your target level a few more dB; that’ll help the latter. Your proposed equalized curve looks too “perfect,” which you only get by equalizing out ever little nook and cranny. (That's one reason why I don't especially like our standard 45-105 dB window - it compels people to over-equalize.) As improved as your response is now, it shouldn’t take more than a few filters. It doesn’t have to look perfect to sound great. For instance, this is my in-room curve. Looks a bit ragged, as you can see, but sounds great.

I was kind of hoping someone would come back and say that I could EQ the subs by just dropping their volume, no real need for a BFD. What are your thoughts on that?

Mine are, you won’t get that house curve without the equalizer.

Regards,
Wayne

[brucek]

I was kind of hoping someone would come back and say that I could EQ the subs by just dropping their volume, no real need for a BFD. What are your thoughts on that?

Well, it certainly was my first thought. I really don't like these negative house curves that people use because they start at 75dB and then drop the upper end too much. To me, the standard target should rise unaltered until the crossover and then begin to raise down to about 30Hz and then flatten out. The simplest curve there is, is a (0db at 80Hz and 5dB at 30Hz). Yeah, some people like to make that rise linear , and that's fine if you like that - but stay away from negative house curves. If you take away your house curve completely, your response is actually quite good on its own and it would be fine if the mains integrated well. You'll have to take a look at that when you add the mains.

Now, to do that, I need to increase my sweep up to what, about 300Hz? I'm assuming that I don't need to go up to 10,000Hz or anything as the subs are well out of range by that point; right?

No, do nothing different than when you measure the sub alone - except readjust your Check Levels routine. The area of interest is at the crossover (80Hz) and then an octave above and below that, so 200Hz upper axis is fine. Try to get the transition around the crossover as smooth as possible with the sub phase and a bit of sub distance trim if needed. If the sub began to rise after and below the crossover a bit down to your lower end, then all the better...

brucek

[Wayne A. Pflughaupt]

I really don't like these negative house curves that people use because they start at 75dB and then drop the upper end too much.

What’s the difference between a negative house curve that starts at 75 dB target, and then you raise it 6 dB (as Conrad did), and a positive house curve boosted the same amount from a 75 target?

Regards,
Wayne

[Conrad.]

I'm currently running pure, uncrossed-over sweeps; is that wrong?
In my HT chain, only the LFE channel will be output to the subs, don't I want all of that? The crossover on my subs is turned off on both subs. Should I try with them on? Should both subs be set to the same? Should I start again with one sub and involve the crossover on just that one sub?

With regard to placement, one sub is inches from the corner, surrounded on two sides by walls and one side by a bookcase and it's firing out towards the door. The second sub is about a foot and a half away from the corner and firing out into the room.

Frustratingly, (and probably commonly) I was able to get close to the house curve by applying just two filters: one wide band and one narrow. This then prompted me to add enough to get the target line matched. My understanding of how it looks compared to how it sounds needs some refinement.

Bruce, I'm confused about the house curve thing you mention. Am I using a negative house curve? I've read the articles (although if there's something I've clearly overlooked then please, point me in the right direction) and I thought I'd understood them. It's all about higher frequencies sounding louder than lower frequencies, right? In Waynes Hard-Knee curve it drops off sooner an more linear-ly so to avoid the "swell" standard house curves generate? I guess I could try EQing to both and see which sounds better in my room.

I'll resweep this weekend with the mains (I have the luxury of being able to turn the mains off, they're through a seperate power amp) and post my results following your advice.

Tonight I've disconnected the BFD so we'll watch The Two Towers with no EQ; that should be interesting!

Franky, I'm quite humbled to have you two thinking about my problems. If only Sonny would come along and add his 2 cents then I'd have "the trilogy"!

[brucek]

Sorry I don't have any time right now as I'm going out, but here's a quicky.

I'm currently running pure, uncrossed-over sweeps; is that wrong?

Yes, use the the standard crossover you would normally use.

In my HT chain, only the LFE channel will be output to the subs, don't I want all of that?

You should have your receiver in stereo with no soundfields on to use REW. The crossover should be set to 80Hz (for example) and the mains on small (and disconnected) so all the bass below the cross will be redirected to the sub(s).

The crossover on my subs is turned off on both subs. Should I try with them on?

The subs own LPF should be bypassed or fully clockwise and the receiver crossover used.

Should both subs be set to the same?

Yes.

Should I start again with one sub and involve the crossover on just that one sub?

Yes...

brucek

[brucek]

What’s the difference between a negative house curve that starts at 75 dB target, and then you raise it 6 dB (as Conrad did), and a positive house curve boosted the same amount from a 75 target

Well, that's the rub isn't it. Using negative house curve demands that the user have knowledge of the Add to Data or target thumbwheel to offset the negative values entered into the housecurve.txt file to make the resultant target reasonable with respect to the mains. For the novice this is a confusing concept. I ask you why would you make the curve file negative in the first place and then compensate for it after (like Conrad did as you point out).

Take a look at the jpg below to see what I mean. I have overlayed a standard mains and standard subwoofer 75dB target (with no house curve) using an 80hz crossover. I think we all know that from those two standard responses that the result will be a nice flat response.

I have also overlayed a fairly aggressive negative method -15dB house curve. To test a theory in engineering we always go to extremes.

The housecurve file reads as:
30 0.0
80 -15.0

Do you see the problem? The standard 80Hz mains target and subwoofer negative housecurve target meet at a much lower crossover point that will basically combine to produce a nasty dip at 60Hz. Yes, the experienced user will increase the target to offset this problem. How many users here are experienced? The use of the negative housecurve is a bad idea. I think REW should dis-allow it...

It's a far better idea to use the positive method, so that the crossover point between the sub and mains without intervention occurs at the crossover frequency of 80Hz.

So the proper file would read as:
30 15.0
80 0.0

I haven't liked this negative method from the day someone suggested it. I still don't like it....



brucek

[Wayne A. Pflughaupt]

I ask you why would you make the curve file negative in the first place and then compensate for it after (like Conrad did as you point out).

Because you’re supposed to shift the Target Curve in range of your response measurement anyway. If this isn’t in the Help files, it should be.

Do you see the problem? The standard 80Hz mains target and subwoofer negative housecurve target meet at a much lower crossover point that will basically combine to produce a nasty dip at 60Hz.
It's a far better idea to use the positive method, so that the crossover point between the sub and mains without intervention occurs at the crossover frequency of 80Hz.

The Target Curve is merely a reference for the purposes of equalizing. Even if you don’t shift the Target Curve like you ‘re supposed to, when you re-calibrate the subwoofer level afterwards to blend with the mains (“intervention” after the fact, if you will) that hole at 60 Hz will disappear. Your graph doesn’t reflect re-calibrated levels, and that’s why things don’t look right. It has the mains’ Target Curve perfectly level with the sub’s – i.e., flat response – so it does not reflect the functioning house curve which will certainly be in place after re-calibration.

If you go back and re-do the graph with the sub’s Target several dB higher than the mains', as it should be, you’ll see that 60 Hz hole will disappear– guaranteed!

Regards,
Wayne

[Wayne A. Pflughaupt]

Conrad,

Frustratingly, (and probably commonly) I was able to get close to the house curve by applying just two filters: one wide band and one narrow. This then prompted me to add enough to get the target line matched. My understanding of how it looks compared to how it sounds needs some refinement.

I covered a lot of that in my Hard-Knee House Curve article, which I assume you’ve already seen. You’ll be able to better tell once you get an equalizer and are able to listen to what you’re equalizing, but basically you’ll probably find that adding a lot of extra filters to track the house curve target to the nth degree isn’t going to sound any better than the few that generally got you there.

Regards,
Wayne

[Conrad.]

My Mains are B&W 602 S3s and my subs are both velodyne SPL1000s.

Doing these calibration exercises has forced me to actually engage my brain and, you know what, I don't think I'm using a crossover at all. All of my processing/decoding (receiving?) is being done in software by my PC. This is being passed out as 6 discreet channels to my "receiver" but it's going into my "multi-channel in" inputs (on a Yamaha Rx-V1500). I'm then taking the sub out and splitting it at the sub end to the two subs. Both subs have their crossover turned to off. Currently, as I'm not EQing at all, there's no need to run two separate signals to the subs. If it turns out I need to EQ one then I can look into doing this.

With the receiver, I'm still confused. I know that no DSP is being applied but, by the same measure, the volume on the receiver affects the overall levels in the room. The only conclusion I can draw from this is that there's no crossover being used, other than the sub's limits. If I was to connect my HTPC's sub output directly to my sub, what crossover would be being used? I've never set a crossover in any of my media playing applications. If the receiver’s applying a crossover then it'll be set to 80Hz, but I don't think it is and REW seems to confirm that.

That said, should I set the crossover on the subs to 80Hz (or whatever works best)? That's probably what they're there for.

[brucek]

If you go back and re-do the graph with the sub’s Target several dB higher than the mains', as it should be, you’ll see that 60 Hz hole will disappear– guaranteed!

Yeah, I certainly realize that..., but you're missing my point I'm afraid - many don't realize that, and it causes a problem that I've seen over and over again. And no, it's not in the Help Files because it's counter-intuitive.

Why would you lower the house curve in the file itself and then subsequently raise the level the exact same amount to compensate..... It makes no sense. Simply don't lower the house curve level and then you don't need to raise the level after.....

brucek

[Conrad.]

I'm probably way out of my depth here but here goes! Isn't it the case that most people will EQ everything for a flat response (with a house curve considered) and then push the sub up by a few dB to their liking? Isn't the hard knee house curve doing that for you without having to alter anything afterwards?

[brucek]

Isn't it the case that most people will EQ everything for a flat response (with a house curve considered)

No. The proper method is to enter a house curve file as shown in the REW Help files where it states that the house curve would typically be used to define a boost for the subwoofer range.

Enter 0.0dB at the crossover frequency and then define your boost down to about 30Hz. The result on the REW response page will show your desired target. If you use a negative house curve, it doesn't show your target - it shows the slope of your target that is reduced by the final negative entry.

Seems reasonable to me that you'd want to see your target in relation to the 75dB mains target. Stick to a positive curve as shown in the Help files......

brucek

[Wayne A. Pflughaupt]

Yeah, I certainly realize that..., but you're missing my point I'm afraid - many don't realize that, and it causes a problem that I've seen over and over again.

The problem I’ve seen is people trying to equalize everything down to the Target line, which results in gross over-equalization. But I have to admit I never pay any attention to the house curve people use – I suppose it could well be all those cases were people using a negative curve. So I guess I can see where you’re coming from...

Conrad, sorry to sidetrack your thread...

Regards,
Wayne

[clubfoot]

Conrad this is how I have my HTPC setup:
M-Audio 7.1 digital coax out to coax in on receiver for all audio formats.
For REW measurements:
line out and in, 1/8" to RCA stereo cables and A-Audio switched to analog stereo mode. Receiver set to stereo, small, (+sub with mains in setup menu), pre out disconnected for mains. Once stereo with sub is selected in my receiver the xover is engaged and the sub xover can be set to disabled.

What sound card is in your HTPC?

[Conrad.]

Thanks for the reply.

I have an RME HDSP9362 http://www.rme-audio.de/en_products_hdsp_9632.php soundcard in my HTPC and I'm using the analogue outs to the pre-ins on the receiver. This shouldn't be applying any DSP or any crossover, effectively setting my speakers to large. However, I'm noticing that level adjustments in the reciever are having an effect, so it's certainly doing something. I think I need to play with this more and understand what's going on in my chain.

If we discount the reciever for a minute (just to help reduce the number of variables); imagine that I'm taking the multi-channel analogue outs from my sound card and running them all to my power amp with the sub connection going directly to the subs. In that situation I'd be sending the 5 channels out to the mains and the LFE straight to the subs. Can REW replicate this or does it need a crossover to work effectively? Can I change the parameters of the sweep to replicate the LFE channel?

Apologies if I appear to be moving the goalposts slightly but I'm trying to explore all of my options.

Another option would be use the S/PDIF out from the soundcard into the reciever and allow the reciever to do all the processing. The problem then is that I'd lose the quality of the RME DACs.

[clubfoot]

Ahhh, but no DAC is better than some DAC
If your sound card can drive the sub, it should work. Set the sound card xover to 80Hz as a starting point, disconnect the mains and surrounds and give it a shot.
The only reason to have the xover functioning is to see how the sub tracks and ideal target curve as a reference. Once the sub tracks the target as close as you can adjust, location, seating etc,....without filters, then you can tweak as you like to bring it closer, better or to taste.

[Conrad.]

I'm not sure I'm explaining myself very well. There's definately no crossover in the soundcard; it's actually designed as a studio sound card rather than an HT card. It's accepting a mutli channel digital input, converting it to analogue and then passing it straight out from a number of line-outs. The fact that the digital signal it's receiving is from PowerDVD and that it contains six seperate channels is really irrelevant to the soundcard, it doesn't know the difference between a centre channel and a rear right surround. Each of the inputs channels is assignable to an output port (and you can actually have any or all inputs going to one output as well as a single input going any or all outputs). It's very flexible.

Any crossover that I would apply would come either from the receiver (if it's applying any crossover at all; something I still need to find out) or from within the software passing the digital streams to the soundcard.

I think you're right though, I need to go from the soundcard to the sub directly. That's easy enough and I know that I can take the digital LFE signal from PowerDVD, have the soundcard convert it to a single analogue stream and send it out to the sub.

Does that make sense?

How would I EQ that? Would setting my maximum sweep frequency to be the same as the upper limit of an LFE channel work? Is there even a standard frequency range for LFE channels? I sense a trip to google coming up.