M-Audio Transit and REW... a success!

[jagman]

So, I finally got sick and tired of SoundBlaster and ordered the M-Audio Transit. Although cheap, it was easy to implement, and I had REW plots in less than an hour. I spent probably 15 or so hours battling SoundBlaster... what a pain. Anyways... on to the good stuff.


This plot has 5 filters:


This plot has 9 filters:


That was just a quick attempt. I'll try and refine it this weekend!

[JohnM]

The target level is too low on those plots, should be at least 80dB and possibly higher. Is the measurement of the sub alone? If so, is there any crossover active? The Target Settings should be set to correspond to the measurement, so if you were making a measurement of both the sub and a main speaker together (for example) the target should be set for a full range speaker as that is what the combination of sub+main should ideally achieve, if it is just the sub then the target crossover should correspond to the crossover being applied to the sub - if you are connected through an AV receiver it should be the receivers sub crossover setting, if directly to the sub or to the sub through a BFD it should be the sub's own crossover setting.

[jagman]

It was set to subs but I hadn't pulled the wires from the mains. I realized that last night but it was late and I just wanted to get a couple readings before going to bed. The crossover within the AVR and REW are set to 80Hz. As you can see I have a massive null at 44Hz. I have the gain all the way up on the Samson S700. This way I don't have to boost frequencies with the BFD (except for the small amount in the mid 40's). I specifically chose to have the output calibrated to a little above the Target curve (that's my house curve). I like to have the lower bass around 5-10dB hot. I didn't get a chance to listen last night as it was a bit late, but I'll fine tune it to taste over the weekend. REW is a marvelous tool.

[jagman]

I played around with the filters and came up with this (with the wires pulled from the mains):



I added a 6th filter for the low end. It sounds pretty good. Not sure about the trough at 132Hz. I'll at some point do a Hz by Hz measurement with the Rat Shack meter to see if I pick it up with the mains attached, but I don't feel like doing it now. I think I'll watch a movie instead .

[jagman]

Guys... I've come across an epiphany.

I tried "refining" my plot. I added 3 more filters that kept the same basic shape but made it considerably smoother. It looked much better. I saved it to a second set of filters in the BFD and switched back and forth. One would think the smoother curve would have also sounded better... not even close! Dynamics in the sub were shot. It sounded lifeless. I was shocked. The only thing I can attribute this to is excessive filtering. I am now a firm beleiver in limited use of EQ. Use what you can to give your basic shape you want, but do not over use your BFD!. Limited use is phenomenal... it makes the lower frequencies sound much better. But excessive use sucks the life out of your sub. I didn't realise it until now... too bad it took about a year to figure that out. Don't make the same mistake I made!

[JohnM]

You need to get the target level set correctly, otherwise you are overcorrecting with your filters - they end up being a poor attempt at a level control instead of addressing the resonances they are aimed at. Try raising the target level to 85dB and generating some filters based on that. You are right about the filtering though, it shouldn't be overdone. Address the biggest few peaks and reassess.

[jagman]

John....

Here's a plot with Wayne's house curve aimed at 81Hz:



Not bad for 5 filters... I'd like to get down to less but it's just not possible.

[jagman]

Here's another with the house curve extended to 83dB:



I'm not sure this point whether I'll go with the 5 filter 81dB house curve or the 6 filter 83 dB house curve. The second may be a little risky with the low end boost. We'll see.

[JohnM]

Using a house curve offset is not really the right way to go yet. Raise the target level using the control in the Target Settings, or use the "Set Target Level" action to let REW adjust it for you, then you will have a better picture of where the response lies compared to its nominal level before starting to apply filters. When applying filters it is best to align them with peaks on your plot, that way you are directly countering the room resonance that is lifting the response at that frequency. Have you tried letting REW find the peaks and optimise the filters?

[jagman]

I haven't done it the way you suggested. I take you at your word because you obviously know what you're doing. I don't really get why raising the target level is better when REW is calibrated to pink noise at 75 dB. I would have thought it was best to set the target level to 75Hz since that would aim to make the full spectrum response flat, and from there, raise the low end with a house curve if desired. I don't understand how raising the target level changes the response within the room. Oh, and I have not allowed REW to set the curves. I'll try your suggestions and see what happens... I just don't understand why it is better.

[JohnM]

The level at which REW was calibrated doesn't really matter, that process just allows REW to relate the size of the signal on its input to a corresponding dB SPL level so that it can draw the measurements at the right level on the graph.

The target level adjustment is used to set a reference where the response would lie if there were no resonance contribution from the room. It is something of a best guess, as it is very difficult to separate the contrubutions of the resonances from the underlying output of the speaker being measured. The target level can be set manually, looking at the response and estimating where it should lie, or by letting REW make its best estimate using the "Set Target Level..." option in the Target Settings - REW does it by generating a test signal and measuring its level, which may or may not be more accurate than using your own judgement.

The target level is important because the gains and bandwidths of the correction filters are adjusted to bring the measurement on to the target line. The filters should be countering the effects of the room's resonances, if the target level is too high the filters would be taking out too little of the resonance, if it is too low they take out too much.

This process is a bit too manual at present, I'll be looking at providing further automation and more accurate determination of the contributions of the room's resonances and the best filter settings to counter them in a forthcoming release. That will also include features to automatically determine the settings for filters that shape the response to bring it to a preferred characteristic, such as a desired house curve.

[m-fine]

I think you are missing the point.

You raise or lower the target level to select the best filters to get the subwoofer SHAPE the way you want it with minimal EQ. Once the sub shape is as desired it is a simple matter to adjust the output level on the receiver, EQ, or the sub's amp to set the output level relative to the mains.

[jagman]

m-fine wrote:

I think you are missing the point.

You raise or lower the target level to select the best filters to get the subwoofer SHAPE the way you want it with minimal EQ. Once the sub shape is as desired it is a simple matter to adjust the output level on the receiver, EQ, or the sub's amp to set the output level relative to the mains.

How is that different than creating a house curve manually. In my case, I had the target curve set to 75dB, but added a house curve that deviated from the original. There is a linear 8dB increase (on the log chart) from 60Hz to 30Hz. It then stays flat at 30Hz below that. I've created the shape that I wanted while only using 5 filters. Isn't that the point?

Edit: I forgot to put in a new graph I did last night. It's a little different than the one above. I also moved the seating position back to try and alter the low end and trough at 44 Hz. It did a pretty good job at both:

[m-fine]

I think what Jonh was trying to tell you was to make one house curve and then adjust the level up and down with the target level adjustment. Yes you can change the house curve to achieve the same thing but that is a lot of work. The 75 db target starting point is completely arbitrary at this stage of calibration.

Once you have your filters set, you can put the target level back at 75 (and not have to adjust your house curve) and then adjust the subwoofer level on your electronics until the sweep best matches your target.

[brucek]

Others have tried to make this point, but you haven't really taken the advice to heart, but the fact is, you're using the BFD as a level control for your sub.

A big clue is that your target (even with an exaggerated house curve) doesn't touch the uncorrected response, and at its worst you require -30dB filtering.



brucek

[Wayne A. Pflughaupt]

Hi Krister,

It looks to me like there’s been something of a "breakdown in communication" here.

From what I can see, based on your posts above: (a) you’re trying to avoid any boosting filters, and (b) you have a highly-capable IB system with two 18" drivers, (which you mentioned in your PM to me, but I don’t think here, so this is news to John, brucek et. al.).

From what I understand this is a fairly typical response for an IB, being "weak" on the low end, coupled with excessive upper-bass energy (at least I think that’s correct – the guys at our DIY Subwoofer forum could qualify that).

So - no two ways about it, the levels below 30 Hz needs to be boosted, and the levels above 30 Hz need to be reduced.

Taking to heart all the stuff you’ve no doubt seen about "boosting is bad," you’ve cut everything above 30 Hz to avoid boosting below 30 Hz. But notice in your graph what the others have pointed out, the difference in overall level between your baseline vs. equalized response. Equalized response is nearly 15 dB lower at 35 Hz.

So, what’s going to happen when you fire your full system back up? You’ll find that the subs now sound really weak compared to the mains (you’ve noted that you like the subs 5-10 dB hot). So, you naturally raise the sub level up to compensate.

To get what I’m talking about, mentally move your equalized Target Curve line up to 95 dB, where the unequalized peak at 35 Hz is. That's about where your sub level will end up once you turn them back up to reclaim the lost volume. Of course, you can see that you’ve succeeded in boosting down low, even though you used only cutting filters.

So you see, in the end boosting or cutting is often academic: If you had sagging response below 30 Hz before, and now you don’t, then you’ve boosted up the sagging area. It’s as simple as that. It doesn’t matter which way you accomplished it, by boosting it outright, or by cutting everything else and then raising the sub volume, the end result is the same. Does that make sense?

Now, with that in mind, let’s take a look at REW and its role in equalization. As we’ve established, you’re going to have to re-adjust your sub level after you equalize to get the correct blend with the mains. So all the levels and stuff you set up in REW before generating the sweep tone, that’s merely for the program’s sake. Once you’re done equalizing, your system levels (including the subs) are going to be adjusted up or down depending on whether you want things quiet or blasting.

Therefore, the purpose of the Target Curve is merely for the sake of properly equalizing. That’s why John, brucek and the others have been encouraging you to re-align it up in range of your base response curve. By keeping it low, yes you’ve successfully flattened response, but you’ve also accomplished wholesale gain reduction.

Normally this isn’t a good thing (as I noted under the “Compelled to Excess” heading of the Hard Knee article, near the bottom of the first post). But in your case things may be different (I’ll get to that shortly). The reason brucek and John et. al. have been encouraging you not to cut gain so much with the equalizer is that we’ve seen cases where people cut so much that they didn’t have enough signal left to drive their sub amp! No kidding, they couldn’t even get a reading on the amp’s meters!

So – back to your graph: If flat response all the way down to 15 Hz is what you’re after, you’ve accomplished it, as long as you have enough of a signal left to drive your amps to max (or at least get the volume levels you need). It’s unfortunate that it took cutting 30 dB at 100 Hz to accomplish it, but if that’s the before-EQ response you have to work with with your crossover engaged (it is, right?), then so be it.

As noted, I think the significantly high levels of upper frequency energy you have is not uncommon for an IB. Raising the Target Curve (as would normally be done) will require some boosting below 35 Hz, but the lowest filter setting on the BFD is 20 Hz. So, equalizing “correctly” with a re-adjusted Target Curve may well result in response sagging below 20 Hz, instead of being flat down to 15 like it is now. It’s up to you to decide if that’s acceptable or not.

So, don’t worry about boosting, if you feel you need to raise the Target Curve and try again. It’s only an issue if your sub drivers and amplifiers don’t have enough headroom to handle it. You’ll find out soon enough if that’s the case. Anyway you cut it (no pun intended), when you’re done equalizing, the depressed areas will be raised, so any equalization will place additional demands on your system. Therefore, you have to have enough headroom going in.

Regards,
Wayne

[jagman]

Thank you for the very thorough reply. What you are saying makes sense. I am surprised how much upper bass the IB puts out. I'll try raising the target level and work with the amp gain to find a nice compromise. One of the issues I have is my integrated amp has a non-adjustable crossover at 80Hz. I plan on getting the Sherwood/Newcastle R-972 receiver after it's release, so I will be able to adjust the crossover down a bit. I think I'll end up crossing it over around 60Hz or so... that should help tame the upper bass a bit without the use of EQ. We'll see.

[m-fine]

You could also benefit from a step up in EQ that allows a low shelf filter such as the DCX2496 or a QSC DSP30. If you put a 6-8 db low shelf boost at 20 hz it will level out the region from 30 on down. You can't do much with positioning the IB to solve the 44hz dip but you did say moving the seats helped, and maybe a small boost in that region paired will smaller cuts on either side will flatten that area. You could also setup a low pass at 80-100 on the DCX until you get a receiver with a 60hz crossover option.

That should give you good results with just 3 PEQ's a low shelf and crossover. You may even be able to do it with just 2 PEQ's.

[Wayne A. Pflughaupt]

Another option would be to get an analog pro-audio electronic crossover and install it in front of the Behringer EQ. I think this would work better than shelving. A shelving filter will only drop and then flatten the level by the amount you cut it, not infinitely like a low pass filter will. The crossover will also have level controls to maintain a good signal for the EQ and amps downstream. An ouboard crossover would also allow you to keep an 80-90 Hz setting on your receiver for your main speakers, which you probably want.

The Behringer CX 2310 is probably ideal for you. It has a dedicated subwoofer output that you can set as low as 10 Hz. So you can set in anywhere you need to tame those upper frequencies. You could even cascade the mains low pass on top of it, if needed, for additional upper bass signal reduction. You can find the 2310 for pocket change on eBay.

Regards,
Wayne

[jagman]

Thanks guys... that may do the trick. Some major networking issues came up at work today so this will have to go on the back burner until next week. I'll think about it and keep all of the recommended equipment in mind. All of you have been very helpful.

[m-fine]

Wayne A. Pflughaupt wrote:

Another option would be to get an analog pro-audio electronic crossover and install it in front of the Behringer EQ. I think this would work better than shelving.

I think you misunderstood what I was saying. I get the impression the OP wants to boost the sub 30 hz output well below where the BFD can help with a PEQ because his IB should be able to offer more extension than he is seeing now. This is where the low shelf with a boost can help him. This is independent of his issues in the 60 hz plus range.

The second issue is he may benefit from a low pass crossover set higher up because the receiver does not seem to be doing it's job or his IB is way too sensitive in that range. I guess he could use the cx 2310 upstream of the BFD to solve that problem, but the DCX is a very good crossover (DCX = Digital Crossover) with PEQ and shelf filters all in one box.

If his receiver's low pass on the sub out is faulty, he can correct for that on the DCX assuming the high pass for the mains still works. Obviously the receiver is redirecting bass content to the sub or he would not have been able to get the graph he did. If the receiver is really toast and is sending full range signals to both the mains and the sub, he could also make use of all three inputs on the DCX and run the pre-outs for left and right into 2 channels and the LFE into the third. The bass from all 3 can be sent to the sub and the upper frequencies can be sent to the amp inputs for the mains (if his unit has amp inputs or he uses and outboard amp). Even though it could be done, and that is the type of thing the DCX is designed for (in pro use) I would still replace a receiver that is that faulty as soon as budget allows.

[Wayne A. Pflughaupt]

m-fine wrote:

I think you misunderstood what I was saying. I get the impression the OP wants to boost the sub 30 hz output well below where the BFD can help with a PEQ.

You’re right, somehow the fact that you suggested boosting below 30 Hz with the shelving filter got past me. Sorry.

Regards,
Wayne

[jagman]

The issue I'm fighting is the natural roll off of a sealed enclosure. The IB is a LARGE sealed enclosure, and due to the large displacement, it is fully capable of reproducing 10Hz and up. The problem is the lower bass is not nearly as loud as the upper bass.

[m-fine]

jagman wrote:

The issue I'm fighting is the natural roll off of a sealed enclosure.

I think that is one of your issues, but if you have an 80hz crossover in effect, the graph shows something is wrong in the 80-140 hz range and that is going to be WAY more audible than your 10-30hz levels. It could be that receiver is not filtering properly, the IB is too sensitive, or the room is adding a lot of gain at that spot, but no matter the cause you do need to correct it

BTW, I have a sealed sub with dual 15's in a 3800 cuft room that is fully capable down to 10 hz and below so I know what you are looking to do should definitely be achievable with the IB, you just need to get the signal levels adjusted properly without boosting or cutting to the point where you clip the amp input or have too weak of a signal to amplify.

One other point of caution. If you do low shelf the stuff below 30, it will be boosting all the way down to the 3-5 hz range where the response of your electronics will likely roll off very quickly. I am not an IB expert by any means but with such a large enclosure, I think you may have to be worried about supper low content in the 5-10hz rang bottoming out the drivers even if you are not getting much usefull output. You may want to put a steep 10 hz high pass in the loop somewhere to help protect them. Unfortunately I don't know of any DSP product that can do that for you that low so you may want to look into building a simple passive circuit. Better yet, ask the IB experts what they do.

[jagman]

Your point about what's going on in the 80-140 Hz range concerns me, too. The unalterred output in the most recent graph has peaks and dips that coincide with what is predicted given the room's dimensions (20' x 16' x 9'). The problem is it is behaving as if it is getting a full bandwidth signal. What I don't get is why it doesn't trail off above 80Hz. The response behaves as if the integrated amp is not applying low pass filter at 80 Hz.