How much lift on a BFD?

mightyoakbob

Established Member
Joined
Apr 22, 2004
Messages
317
Reaction score
21
Points
119
Location
Stourbridge
Hi guys,

I've been playing with REW and house curves and as I still have an acoustically horrible room I've come up with a question.

Because I have such massive room peaks after doing a measurement I lower the target line by 10dB, I pretty much have to if I'm to get anywhere near a decent curve. Unfortunately, I still have a very bad suck out in a couple of places but for now looking at the one at 46.4Hz this has filter 5 with 3dB of lift directly on it.

I stopped at 3dB because I've read many times how 3dB is the max otherwise something will reach the limit and clip. Sounds very reasonable at first. But my question is 3db above what and what is the "something" that will clip?

As I see it there are 4 possible overload (clip) mechanisms.

1. The electrical signal limit on the input of the BFD.
2. The maths algorithm for the filters.
3. The electrical signal limit on the output of the BFD.
4. The subwoofer itself.

I can, I'm sure eliminate (1) as this signal is not being changed. The sub (4) will surely pop or whatever if clipping. As for (3) well in my case with the filters on, the output here is a lot lower than it would be if the filters were off so I think we can (in THIS case eliminate) (3).

That leaves the maths algorithm. Is this what can overload?

If not, I can see no way for anything to clip even if I put another 5dB of lift at 46.4Hz.

What do yoiu guys think? Takes a bit of thinking about doesn't it?


Cheers,

Bob.
 

Attachments

  • lowres.jpg
    lowres.jpg
    24.9 KB · Views: 62
i think it's the fact that you're asking the sub to drive twice as much power into that particular frequency will lead the sub to reach the limits of driver excursion and possible other effects. I was under the impression that you should never boost, only cut. I've heard that from many sound engineers in general before I started playing with the bfd etc.

Maybe if you try some manual test tone readings for that frequency and the ones near by, you can determine the width of the frequencies affected. If you find it's actually only 46.4Hz pretty much on the nose that is affected then I'd guess you shouldn't worry too much as it's only making the graph look bad as it's drawn as a continuous line and you'll not notice the missing area in your response? Just a thought, sure other, more proficient, eq'ers will be along soon :D

Eric
 
I had some thing very similar to this in a long thin room with me sitting about 60% of the length of the room from the front but sorted the problem by moving the sub to about half way down the long side way. Not always easy to get the sub in the best place and it took me ages to find it but this site pointed me in the right direction

http://www.audioholics.com/tweaks/g...er-placement-the-place-for-bass-part-1-page-7
 
i think it's the fact that you're asking the sub to drive twice as much power into that particular frequency will lead the sub to reach the limits of driver excursion and possible other effects. I was under the impression that you should never boost, only cut. I've heard that from many sound engineers in general before I started playing with the bfd etc.

Maybe if you try some manual test tone readings for that frequency and the ones near by, you can determine the width of the frequencies affected. If you find it's actually only 46.4Hz pretty much on the nose that is affected then I'd guess you shouldn't worry too much as it's only making the graph look bad as it's drawn as a continuous line and you'll not notice the missing area in your response? Just a thought, sure other, more proficient, eq'ers will be along soon :D

Eric

This is the concern, but it's more about the power requirements than cone excursion, although that does come into it.
Overdriving the subs amp is the risk, in that, that is what will cause the clipping.
I've had this conversation before and the general consensus is that you shouldn't boost by more than two or three dB. Indeed that is exactly what the BFD manual says :smashin:
But IMO, and I freely admit I'm in the minority here, I think that is really a very general statement, that basically covers people arses, to prevent you overloading the amp and complaining that it was the BFD that caused it.
Let me explain my logic.

Seeing as the lower you go, the more power you need to produce a given volume, it makes sense that boosting the 46 hz area is not going to be such a strain on the amp as boosting at say 25hz.
Now let's say, and it's a reasonable assumption, you are running your sub gain at 50% of it's max (notice I didn't say 50% of the dial movement), it figures that you have an extra 50% power available, even at the lowest frequency the sub can produce (and probably then some to give the sub a bit of a safety factor). If that part is true, it figures that you could boost by 3 db at 25hz no problem and still be within the amps margins. It then follows to me, that this automatically means you can boost higher frequencies by more, because they take less power, right?
Now, I'm not exactly sure how much more power you require in relation to frequencies we're talking about, although I would have a suspicion that doubling the frequency would halve the power required to give a certain volume. If that were true, it figures that you could double the boost again and still be within the amps spec at say 50hz.
I'm not saying that is correct though, it's an assumption. Perhaps someone could come back on that bit?

Suffice to say, I boosted higher frequencies in the 60-70 hz area by a lot more than 3dB (I think it was nearer 10dB :eek:) with absolutely no problem whatsoever. But then that was with my old PC12 Ultra, which had IIRC 500W and probably a good bit of headroom before clipping was approached.

I'm not recommending you do the same, just putting across a viewpoint, that might interest you :smashin:
 
Guys,

Thanks everyone for your help and input it was most interesting.

I seems you all think that the +3dB limit is to prevent over driving the sub. I'll be honest and say I was hoping someone would come forward and say something that confirmed my thinking on the matter but no suck luck.

For the record, I *think* the advised +3dB limit came from the BFD manual and has nothing what so ever to do with the sub. After all, they don't know how loud you'll play it, or the sub's capabilities. It may be the size of a wardrobe for all they know.

No, I think its purely to do with the the voltage ratio between the input of the BFD and the output. If you drive the input of the BFD pretty well with a good signal as they advise and there is any significant gain within the BFD it will voltage clip on the output.

If I'm right (I'm anything but sure) then the implication is that it is also only slightly related to the gain of a filter because one filter doesn't act alone. In my case the output of the BFD is on average around 10dB lower than its input. This means (I think) that I can put a filter with a lift of 10dB in a dip provided by other filters and not add up to overall gain at all.

So in short I think the limit is thou shalt not have more than 3db of overall gain at any frequency and this is not the same thing as saying no filter should lift more than 3db.

Does that sound reasonable or should I book myself in somewhere for some much needed treatment? It would have been nice to get my idea confirmed
Ah well.


Bob.
 
Bob, if I read that graph right, you have the unEQd trace, the inverse trace of the applied filters and the actual in room response is the resulting dotted line?

If so and only if so, you're barely boosting anything, even with the 3dB boost applied, although the result of the boost is more likely visually beneficial, rather than audible..:thumbsup:

As a result of EQing, you've turned the sub down by 10dB and then added 3dB of boost back in, but that's still a net 7dB cut. A boost is only an issue, if you're boosting the response above that of the unequalised response. As it stands, you're still applying nett cuts and thus running the sub far less aggressively at all frequencies, than someone with the same sub unEQ'd might.

Further more, boosting only really increases as an issue with decreasing frequency especially if it has any effect below port tuning. The 3dB boost you've applied happens well within the subs passband and is unlikely to cause any issues, but I do notice one thing....

Call a boost something else and people forget that's what it is. Your house curve places the target curve 5dB hotter at 20Hz, than it does at 46Hz, where you're worrying about boosting 3dB. Guess which boost is more likely to clip the amp/bottom the driver first?;)

Russell

<EDIT: Started tying this quite while ago and I see you've worked it out for yourself>
 
No, I think its purely to do with the the voltage ratio between the input of the BFD and the output. If you drive the input of the BFD pretty well with a good signal as they advise and there is any significant gain within the BFD it will voltage clip on the output.
I think the mathematical clipping is as a result of inputing a signal voltage too high to be described by the dynamic range offered by the 16bit resolution of the ADCs of the BFD.

Of course the inverse of this situation is true in so far as, set the input volume too low (or indeed listen too quietly) and the resolution of the ADCs output is reduced.

Clipping the signal output of the BFD is less likely as thats an issue in the analogue domain after the D to A conversion of the EQ'd signal.

At least that's how I think is was explained to this resolutely analogue dullard.:)

Russell
 
As a result of EQing, you've turned the sub down by 10dB and then added 3dB of boost back in, but that's still a net 7dB cut. A boost is only an issue, if you're boosting the response above that of the unequalised response. As it stands, you're still applying nett cuts and thus running the sub far less aggressively at all frequencies, than someone with the same sub unEQ'd might.
is that always right? if you're looking at the in room response then the measured response is produced by the sub itself and the room and therefore you're not just measuring amplifier output at any given point. It's pretty obvious when that would be the case though (if it is indeed the case).

FWIW on the clipping issue then I would think that a boost could only tip DSP into clipping at the absolute extremes (i.e. input is close to clipping anyway and the boost tips it over the edge). The only way I could see this happening is if you set the levels too high anyway.

I agree on the other posts, if you're confident you have enough headroom then boost up into the stratosphere :) If you're not confident then increase the boost gently and test inbetween.

Cheers
Matt
 
Guys,

Thanks everyone for your help and input it was most interesting.

I seems you all think that the +3dB limit is to prevent over driving the sub. I'll be honest and say I was hoping someone would come forward and say something that confirmed my thinking on the matter but no suck luck.

For the record, I *think* the advised +3dB limit came from the BFD manual and has nothing what so ever to do with the sub. After all, they don't know how loud you'll play it, or the sub's capabilities. It may be the size of a wardrobe for all they know.

No, I think its purely to do with the the voltage ratio between the input of the BFD and the output. If you drive the input of the BFD pretty well with a good signal as they advise and there is any significant gain within the BFD it will voltage clip on the output.

If I'm right (I'm anything but sure) then the implication is that it is also only slightly related to the gain of a filter because one filter doesn't act alone. In my case the output of the BFD is on average around 10dB lower than its input. This means (I think) that I can put a filter with a lift of 10dB in a dip provided by other filters and not add up to overall gain at all.

So in short I think the limit is thou shalt not have more than 3db of overall gain at any frequency and this is not the same thing as saying no filter should lift more than 3db.

Does that sound reasonable or should I book myself in somewhere for some much needed treatment? It would have been nice to get my idea confirmed
Ah well.


Bob.

I think you're on the right lines.
It's basically a different but identical way to how I was trying to back up my argument.
My way of looking at it is, what if you upped every filter by 3dB?
This would be permissable, according to the BFD manual :smashin:
But you're basically asking double of the amp, all across the range :eek:
And if you turn the sub full up? :eek::eek::eek:

It's a no brainer, IMO.
Any decent sub with a bit of headroom will handle a 10dB boost, with one filter, no problem. Although I still wouldn't want to do it at the bottom end.
It's not as if a sub plays a particular frequency for sustained periods is it?
 
For the record, I *think* the advised +3dB limit came from the BFD manual and has nothing what so ever to do with the sub

IMO, you are totally wrong here.

You will get clipping via the BFD if you give it too much input, which is why you need to get the amps output level right.
The point is, if you clip the input, thus distorting the same and then boost it, you're asking for major trouble. As I said, the lights shouldn't hit red, more than momentarily.

Whatever, clipping at the BFDs input or output level is bad news.

By the same token, if the output from the BFD is linear, it will be safe to boost as much as you like, as long as the subs amp is still linear.

This is where the sub has everything to do with it.
 
is that always right? if you're looking at the in room response then the measured response is produced by the sub itself and the room and therefore you're not just measuring amplifier output at any given point. It's pretty obvious when that would be the case though (if it is indeed the case).
Unless my tired eyes read it wrong, then Bob has reduced the overall output by 10dB in order to hit the curve, which of course is the amplifier output. FWIW, given that cone excursion is the Monolith destroyer and even then, only with ham fisted treatment at, or below port tuning levels, I can't see the problem with a 3dB boost at 46Hz in Bob's case.

FWIW on the clipping issue then I would think that a boost could only tip DSP into clipping at the absolute extremes (i.e. input is close to clipping anyway and the boost tips it over the edge). The only way I could see this happening is if you set the levels too high anyway.
IIRC and it's a while since I've used a BFD, boosting is something that happens in the DSP stages of the BFD and is therefore post the input signal into the BFDs ADCs. These simply require the input level to remain unclipped at maximum listening level on big signal peaks. If the ADCs input is unclipped, then one assumes the signal into the DACs post DSP will remain unclipped. I don't recall any advice about adjusting input levels requiring adjusting to compensate for boosts applied in the DSP. Now you raise it, its an interesting thought as to how the DACs would output a greater signal than one input to the ADCs in the firstplace. Certainly a thought that requires more than my tiny mind!

I agree on the other posts, if you're confident you have enough headroom then boost up into the stratosphere :) If you're not confident then increase the boost gently and test inbetween.
Sound advice indeed. Always test any doubtful changes in a progressive fashion.

Russell
 
IIRC and it's a while since I've used a BFD, boosting is something that happens in the DSP stages of the BFD and is therefore post the input signal into the BFDs ADCs. These simply require the input level to remain unclipped at maximum listening level on big signal peaks. If the ADCs input is unclipped, then one assumes the signal into the DACs post DSP will remain unclipped. I don't recall any advice about adjusting input levels requiring adjusting to compensate for boosts applied in the DSP. Now you raise it, its an interesting thought as to how the DACs would output a greater signal than one input to the ADCs in the firstplace. Certainly a thought that requires more than my tiny mind!

Totally correct, IMO.
Linear in will be linear out, simple as, as far as the BFD is concerned.

I would expect any non-linearity in the ADC to be amplified by the back end.
That was certainly how it worked with my ham gear.
Distortion at any level is hard to kill, even with negative feedback.
 
Unless my tired eyes read it wrong, then Bob has reduced the overall output by 10dB in order to hit the curve, which of course is the amplifier output. FWIW, given that cone excursion is the Monolith destroyer and even then, only with ham fisted treatment at, or below port tuning levels, I can't see the problem with a 3dB boost at 46Hz in Bob's case.
I agree though I wasn't clear, I was querying the description of a boost in the general case as opposed to the specific example in the OP.

If the ADCs input is unclipped, then one assumes the signal into the DACs post DSP will remain unclipped. I don't recall any advice about adjusting input levels requiring adjusting to compensate for boosts applied in the DSP.
on reflection I'm not talking about the same clipping that you are :) I thinking of an analogue signal going into the ADC, undergoing transformation in the DSP and then back to analogue in the DAC. The DSP has 16bit resolution so the analogue signal will become a number, if that number is close to the maximum for a 16bit integer and it is subject to a boost then you could push the value to the max and you lose detail & resolution. My thinking is that the only way this is possible is if you've set the levels too high anyway though if you're boosting as opposed to just cutting then you might want to be slightly more careful about setting the levels. I suppose it depends how you set them, I rarely (if ever) hit the yellow markers myself (signal consistently in the mid green range whenever there is sub activity) so have plenty of headroom in the BFD.

Cheers
Matt
 
Bob, if I read that graph right, you have the unEQd trace, the inverse trace of the applied filters and the actual in room response is the resulting dotted line?

Pretty much correct though of course the inverse trace of the applied filters is more the inverse of what is normally displayed, which is itself inverted. In other words the filters are not upside down as they are normally displayed.

If so and only if so, you're barely boosting anything, even with the 3dB boost applied, although the result of the boost is more likely visually beneficial, rather than audible.

Fully agree. Thinking about it further, I think it should be possible to boost any point at least back up to where it would have been without the filters in place. This cannot cause overload of the BFD provided the maths can cope and i can't see why it shouldn't.

As a result of EQing, you've turned the sub down by 10dB and then added 3dB of boost back in, but that's still a net 7dB cut. A boost is only an issue, if you're boosting the response above that of the unequalised response. As it stands, you're still applying nett cuts and thus running the sub far less aggressively at all frequencies, than someone with the same sub unEQ'd might.

Yes, that's how I see it.

Your house curve places the target curve 5dB hotter at 20Hz, than it does at 46Hz, where you're worrying about boosting 3dB. Guess which boost is more likely to clip the amp/bottom the driver first?;)

Absolutely.

Bob.
 
Fully agree. Thinking about it further, I think it should be possible to boost any point at least back up to where it would have been without the filters in place. This cannot cause overload of the BFD provided the maths can cope and i can't see why it shouldn't.
That's exactly what I do.

The unEQd response is, by definition, subject to no boosts or cuts other than those affected by the rooms contribution. As long as your EQd response remains under or equal to the original response, I can't see what the problem is.

Russell
 
That's exactly what I do.

The unEQd response is, by definition, subject to no boosts or cuts other than those affected by the rooms contribution. As long as your EQd response remains under or equal to the original response, I can't see what the problem is.

Russell

wow what a head****! until you said this my brain was melting, then miracously it all made sense :rotfl:
 

The latest video from AVForums

TV Buying Guide - Which TV Is Best For You?
Subscribe to our YouTube channel
Back
Top Bottom