To Boost or to Cut ? (1 Viewer)

[Doug Fraser]

I have read with great interest the discussions on the board about amplifier headroom and whether, when using an EQ for a Sub, to cut frequency responses to a desired curve or to boost holes in the response. The arguments presented to date center around amplifier headroom. I have done some thinking about this and would like to present my thoughts for discussion.

For the purpose of this discussion I define amplifier headroom as the difference between the maximum output voltage of the amplifier can deliver and the instantaneous output voltage of the amplifier with a given input signal level (this could be converted to db however for the purpose of this discussion I will leave it in volts).

For this example lets assume that the amplifier has a ± 70v power supply. Then the maximum output voltage of the amplifier (before clipping) would be a signal of ± 70v (Note: Actually you would not get exactly ± 70v signal due to the voltage drop across the forward biased active devices (transistors) in the output stage of the amplifier, however for this discussion I am ignoring this small voltage drop).

Amplifiers can be thought of as voltage multipliers. For example an amplifier with a constant gain of say 50, then a ±1.3 volt input signal would be outputted as a 1.3 x 50 = ±65v signal. This is good, as we have not exceeded the maximum potential output signal of ± 70v. At this point it should be noted that the assumption here is that we are putting pesudo-random noise into the amplifier that has the same voltage level across the frequency band in question (say 15 Hz to 100 Hz). In this case we would have 5v of headroom (70v maximum – 65 v output signal = 5v headroom).

This is where it get interesting….

Now lets assume that choose to boost a small frequency band around say 50Hz by say 0.1 volts. Therefore at 50Hz we would be asking the amplifier to amplify 50 times a 1.4v signal. 1.4 x 50 = ±70v (remember our original input signal was 1.3 volts). All is still good as the amp can deliver ±70 V however what is important is that now that all headroom is lost. We have lost 5v of headroom by boosting the signal around 50 Hz. While this is only for one small frequency band, never the less, headroom still lost. Any additional rise in the input signal (around 50 Hz) would cause the amplifier to start clipping the signal.

Most amplifiers vary the output volume level by varying the amount of gain. Therefore lets assume that the gain of our amplifier can be changed from 0 to 60 by simply turning the volume control. (Note: above I assumed the gain had been preset to 50. However it is possible to get more gain out of the amp).

So now lets assume that the gain (volume control) has been set to 60. With our non EQed signal of 1.3 volt we would clearly be driving the amplifier into clipping. 1.3v x 60 = ±78v when the amplifier can only deliver ±70v. This approximates the real world as you can turn up the volume control and cause an amplifier to clip.

Now lets assume instead of boosting the signal by 0.1 volts at 50 Hz we decided to cut all the frequencies from 15 to 100 Hz by 0.1v. Now we have an input signal over the frequencies in question of 1.3 – 0.1 = 1.2 volts. With the original amplifier gain of 50 the amplifier would output ±60v. Again this is good, no clipping, however there is less signal out of the amp and the Subwoofer would need to be recalibrated. OK, so we recalibrate our Sub by turning up the volume control on the amplifier so we get the original ±65v out. This is done by setting the gain of the amp (via the volume control) to 54.17. This is still below the maximum gain of 60.

We now have a ±65v output signal with 5v of headroom. Which is exactly where we started except we now have a flat in room frequency response whereas before we had a flat in room response with no headroom.

Therefore given the above discussion is seems that it is appropriate to focus on cutting peaks when EQing provided that the combination of the input signal level and the gain of the amp allows one to get back to the original output level. If one had to cut the signal (in the frequency band in question) to say 1v over the frequency band, then, the maximum output level of the amp would be 1 x 60 = ±60v (remember 60 is maximum amount of gain we can get out of our hypothetical amp with the volume control turned all the way up). There would be a loss of ±5v output signal.

Conversely, if I had picked different numbers in my example it could be shown that with small amount of boost that it is possible to retain some headroom and not drive the amp into clipping. However the amount of headroom would still be reduced.

Therefore the conventional wisdom of focusing on reducing peaks and using small amount of boost is probably the best way to go when equalizing a Sub.

regards,

Doug

 

[brucek]

Doug,

Therefore lets assume that the gain of our amplifier can be changed from 0 to 60 by simply turning the volume control................
is seems that it is appropriate to focus on cutting peaks when EQing provided that the combination of the input signal level and the gain of the amp allows one to get back to the original output level.

Therin lies the rub.. I agree with your assessment, but as the quote indicates there's usually more variables involved that have to be given consideration. Indeed, in a perfect world if there was an unlimited range of supplied voltage for the sub amplifier to work with it would lead us to never boost, but to cut everything until a flat response was reached and then wholesale increase the voltage and feed it to the sub amp.
But with most EQ's, they are a unity gain device and so the maximum voltage range is determined by the processor sub out line amplifier. It has a limit and therefore so is your required compensation after a large EQ reduction of level.
A further limitation is placed on the level fed from the processor by the maximum allowed input level to the EQ's ADC system. We don't want to over drive it into a clipping situation.
A further limitation is that many sub amps are "dialed up" to maximum just to have enough level to drive them to sufficient power given the levels from many processors sub outputs.
In that regard, you can simply add a line amp between the EQ and subwoofer amplifier, but there are limits and of course this adds to the entire SNR of the sub. Sometimes line amplification is necessary though, particularly when mixing commercial and pro level equipment.
But, I agree, it is wiser to reduce peaks. I'm just not generally a promoter of wholesale shelf reducing large ranges of frequencies to compensate for one narrow valley that could have simply been given a boost or even left alone to preserve the overall voltage level and avoiding further amplification.
Good post....
brucek

 

[Doug Fraser]

Brucek:
Thanks for the response. Below are my comments...
Your Quote:
"But with most EQ's, they are a unity gain device and so the maximum voltage range is determined by the processor sub out line amplifier. It has a limit and therefore so is your required compensation after a large EQ reduction of level."
I would be surprised if there was not enough adjustment range in the output analog circuitry of most EQ's to not provide for 3 to 5 db of overall boost and thus be able to drive an sub amp into clipping.
Your Quote:
"A further limitation is placed on the level fed from the processor by the maximum allowed input level to the EQ's ADC system. We don't want to over drive it into a clipping situation."
The level fed from the processor to the EQ should be set independent of whether there is overall boost or cut. Here the key factor is not overdrive the input A to D converters that are infront of the Digital Signal Processing stage.
For the BFD I would suggest a set up as follows....
One should use AVIA test tones which are recorded at -20 dbfs (db relative to full scale digital). We also know that with Dolby 5.1 that reference level is 105 db in room SPL with the volume control turned up to Ref level. AVIA instructs us to adjust the in room SPL to -85 db. The reasoning here is that the test signal is recorded on the DVD at 20 db below digital maximum so therefore we would expect the audio volume to be 20 db below the reference volume of 105 db (which is 85db).
Further, the LFE channel reference level is 115 db or 10 db higher than the other speakers. Home Dolby Digital equipment is pre-set to play LFE data 10 dB higher than a main channel (or 10dB higher than the bass from a main channel). It is only necessary to set the subwoofer relative to a main channel and the LFE level will be correct.
This is further explained at the following link:
Link Removed
Therefore, with an AVIA test tone, speakers set to small so that bass is redirected from the F,C,L to the LFE channel and processor set at Ref Volume level the BFD should be adjusted so that the -30 db light comes on.
The reasoning for this is best described by the use of a signal budget as follows:
-20 db input signal from AVIA (from digital Ref level), plus -10 db which is automatically added by the Dolby Decoder for the LFE channel indicates that we will be -30 from full scale digital. This is why the BFD should be set so the -30 dB light comes on. Now when we get a signal at 0 db full scale digital from a DVD we will not overload the front end of the BFD.
Your Quote:
"But, I agree, it is wiser to reduce peaks. I'm just not generally a promoter of wholesale shelf reducing large ranges of frequencies to compensate for one narrow valley that could have simply been given a boost or even left alone to preserve the overall voltage level and avoiding further amplification."
My question is why are you not a believer of wholesale shelf reducing large ranges of frequencies....?
Provided that one can still drive the sub amp into clipping it seems to me that EQ reduction is the only way to go as one retains the amp headroom.
Regards,
Doug

 

[brucek]

Greetings Doug,

My question is why are you not a believer of wholesale shelf reducing large ranges of frequencies....?

I'll answer that one again by using your response of:
"Provided that one can still drive the sub amp into clipping it seems to me that EQ reduction is the only way to go as one retains the amp headroom."
Often, it just ain't so. I site the many posts regarding this very problem from those that use pro amps for their subs fed from commercial equipment. But I agree with you if there is indeed any room left.
brucek

 

[Steve Stogel]

Man, you guys are speaking almost entirely Greek to me. I understand the little words (if, the, and, etc...), but the bulk of it is waaaay over my head. I do have a question, though, and maybe someone can give a simple answer (explain as you would to a child ). I have read that boosting eats up headroom (got that), but that lowering peaks does not create or give you any headroom (not so clear on this). In my room, I had a pretty big peak between 36 and 56Hz or so, I believe. My EQ for 40Hz is -16dB (the max min ...max min is right, right?).
So if you have a ton of reductions and just a few boosts, why does this kill your headroom so badly? As an example, say you've got 20Hz at +3, 25 at 0, 31.5 at -5, 40 at -10, 45 at -16, and so on. Why would that 20Hz boost kill your headroom? I know it would halve your power at 20Hz, but doesn't your amp work much, much less at 31, 40, 45, et cetera?
It may be that my mind won't be able to grasp this, but any explanation would be appreciated. As it is now, I'm just dropping my peaks since it seems to be the safest thing, but I still don't understand it clearly. Thanks.
Steve

 

[brucek]

Steve,

Yep, you've got it pretty good.

The loss or gain in headroom is specific to the frequency that had the boost or cut applied.

For this explaination, "very simply", lets assume it took 4 volts to drive a sub amp to full power at all frequencies of interest. The room I have to work with on its input is from zero to 4 volts (AC).

The 4 volts will drive the amplifier such that it has reached its maximum voltage "headroom" supplied by its +/- DC power supply.

Let's assume you placed an EQ between your processor and sub amp and boosted only 50Hz such that it doubled the voltage going to the sub amp at 50Hz. If I fed 2 volt tones from my processor at 100Hz, then 50Hz, you can see that the 100 Hz tone would not overdrive the amplifier and you would have lots of "headroom" in the amplifier to still turn the input voltage up higher from 2 volts all the way to 4 volts.

But, the 2 volts from the processor at 50Hz would reach the amplifier (after EQ) at 4 volts. You now have no "headroom" left in the amplifier. So by boosting the 50Hz signal you reduced your headroom - but only at 50Hz........

But here's the rub - music will contain lots of 50Hz signal along with the other frequencies, so by boosting at 50Hz you've reduced your headroom. You can't really play your music in this example greater than 2 volts can you? Or you'll overdrive and likely compress the 50Hz signal when it gets into the amp.

brucek

 

[Bob Christensen]

The problem with trying to boost the nulls (the low spots on the curve) through EQ is that the nulls are there for a reason. Think of these as “black holes” in the frequency spectrum, or think of the room as “swallowing up” the frequencies that are in the null. Now, simply trying to boost the nulls by increasing the gain in those frequencies is simply throwing more of the same frequencies into the “black hole”. You would need a tremendous amount of headroom to overcome a 6-10 db null, since you are pumping the wattage into a “void”, and you get very little return for the amp watts devoted to this task. Much more effective to simply attenuate (cut) the gain on the peaks. This is especially true if you have a single noticeable “peak” on your frequency response curve.

It CAN be done, EQing a large proportion of the curve, but it takes a TREMENDOUS amount of amplifier headroom to do so. Many of the commercially available tiny sealed box subs like the Sunfire line rely on this approach, and that is why their amps are rated for astronomical figures.

 

[brucek]

Bob,

Good post. What I find most interesting is when you have a couple of valleys that appear about the same bandwidth and amplitude and one of them EQ's with a boost very easily (i.e. a 5dB valley that raises 5dB when you EQ by +5dB), and then the second peak that won't budge no matter how much you try to pump it up. What's your take on this?

brucek

 

[Bob Christensen]

BruceK: I would probably first suspect a "problem" in the electronic path of the null that EQ's out easily. Most likley a XO point that does not jive with the drivers frequency response curves very well. I find most low frequency nulls are very hard to overcome by boosting gain. Some higher frequency nulls (say over 2k) are much easier. May be room harmonics issues.

 

[Doug Fraser]

Brucek:
Again thanks for the reply....(btw how do you get the Quotes indented?)
It sounds like we are in violent agreement.
Your Quote:
"I can't imagine you're implying to wholesale raise and EQ flat a 3 to 5dB increase for the entire range of frequencies that pass through an EQ. Consider a BFD, there is no "volume" control on it. It has unity gain. You may set 12 parametric filters that indeed have cut and gain. These work well to remove peaks and valleys. They make a rather poor volume control."
You’re right I am not implying a wholesale rise. I am assuming that there is an input and output gain adjustment on the EQ separate from the processor or the sub amp.
Your Quote
"Why rob yourself of dynamic range because you've set your input level using exact AVIA reference levels. You need to use the highest dynamic range device, which would be a DVD and play it as loud as you normally would and when an explosion or loud bass scene occurs, check the LED level on the BFD."
I agree but I would still advocate the previously proposed setup method with the exception of setting the volume control (on the front of your processor) to the max listening level you would ever turn it up to instead to the reference level.
In conclusion I think we both agree that the conventional wisdom of focusing on reducing peaks and using small amounts of boost is probably the best way to go when equalizing a Sub.
BTW thanks for engaging me in this discussion. It is wonderful to have a location to test ideas in a suportive environement.
regards,
Doug

 

[brucek]

Hey Doug,

To encase any statement in quotes, the command is, quote surrounded by square brackets.

Then when you want the quote area to end the commend is, /quote surrounded by square brackets.

Yep, we're probably pretty close in our thinking on this topic. I'm always interested in throwing around ideas about EQ and sound in general. I have a fair understanding of electronics since it's my profession, but I am pitifully weak in my knowledge of sound, sound pressure, acoustics etc. So I'm always sticking my nose in those threads that deal with that subject to try and learn a bit more.

brucek

 

[Doug Fraser]

BruceK:

yea me to...

I have a MSEE, so like you, I am very comfortable with the electronics, but I have never really pursued audio to the extent that I do now. What I find really cool is with digital audio (specifically home theater) there are guidelines (like THX, Test Disc's etc) that have supportable scientific roots. This stuff makes sense to me, whereas, the "golden ear" age of audio left me cold. Don't get me wrong, good sound is the goal but I always had difficulty with "tweeko things" like, outrageously expensive cables, green felt pen on CD edges, special speaker stands etc. Now I get way better sound in my family room than I do by going to a THX certified commercial theater.

I am also an amateur woodworker and I just love combining this hobby with my electronics training. I have built all my own speakers L,C,R,RS,LS,LR,RR including a Stryke 15.2 Sub. The drivers are made by AVI a company based in Richmond BC. What blows me away is that no one ever talks about them in home audio circles as the company has been focused in car audio. At first I was skeptical but after realizing that there are more people putting expensive sound systems into cars than into their homes I became a believer. The real clincher for me was when I compared them to my B&W 802's (about 10 years old). They sounded close. A local dealer here in Vancouver, Speaker City, that specializes in the amateur market provided plans and even had a custom x over designed for me as I implemented dipole surround speakers.

I have paid for the ETF software from Doug Plumb and just took delivery of a BFD. I have yet to get the time to EQ my room but it’s coming.

regards,

Doug

 

[Doug Fraser]

Brucek:

I just check the RANE web site and their parametric and graphic EQ's have gain controls with up to 6 db of overall gain in the output stage.

These would certianly take care of the issue of not enough drive for a Sub amp.

Doug

 

[brucek]

Doug,

Interesting stuff. A Stryke 15.2 - yikes. No lack of output there. If I built a sub, that would be the one.

Yeah, a few people around here use the Rane PE17 or the Symetrix 551, which is another popular one. The complain I hear about these is the difficulty fine tuning the dial pots. There's also the problem of little fingers changing the dials and of course they don't have any memory and so your stuck setting the system back up after an accident.

Myself, I use a Paradigm X-30 control unit between my BFD and Sub amp which provides me with about 13dB voltage gain if I need it. It's acts as a very nice line amp among other things, allowing front panel control directly beside my processor.

Since you have the ETF software and a BFD, you'll have to post your thoughts about the setup and effectiveness of the two together when you get around to playing with them. I'll be interested and so will others. I've used the ETF myself - bit of a learning curve, but works well.

brucek

 

[Doug Fraser]

Brucek:

When I get to EQing my room, I will post my results and thoughts.

I have also read with great interest the subject of house curves (pro's and con's). I have also examined the Fletcher Munson curves. This should be fun to play with. It would seem that for a pleasing sound that the start, stop frequencies and gain of a house curve should be a function of ones average listening level..... plus of course, other parameters that have been discussed in other threads.

I have drawn this conclusion because from Fletcher Munson, as the volume level decreases so does one's sensitivity to low frequencies. Therefore as the average volume drops the house curve should (TBD) cover a broader frequency band and have more boost.

WRT to ETF has there been any discussion regarding the windowing / gating functions that are available on the EFT and the virtues of including or not including room refelections in the measurement?

regards,

Doug

 

[brucek]

Doug,
There doesn't seem to be a lot of discussion with regard to those kinds of details about ETF on any of the forums. I posted a little review of my experience a week ago, you may have read it. Basic stuff only.
http://www.hometheaterforum.com/htfo...threadid=52284
When I did my measurements I left the gate out at 100ms to get a good return of the room response. So for me, it was pretty basic useage. If you find anything interesting, maybe you can post it.
I agree with your assertion with regard to bass levels. Well, I guess that was the theory behind the "loudness" controls you use to find on receivers, (and maybe still do).
I listen to music at fairly low levels and have my "mains to sub" level differential set quite high because of this. If I turn my volume up for a test and listen to music, there is way too much bass.
brucek

 

[Sonnie Parker]

Good book here fellows.

Lots of good stuff.

As far as the boost and cut issues, don't we draw a bottom line conclusion that if the boost clips your amp then obviously a boost ain't good and if it get's rid of the dip and the sub and amp perform properly then all is well?

This is not intended to take away from your fine opinions and logic above, please don't take it the wrong way. I thuroughly enjoyed reading it. I didn't understand every bit of it either, but I got most of it. Maybe it will absorb. It was a lot to read at this time of the night.

 

[brucek]

Sonnie,

Yep, that certainly is the bottom line. But, the problem is trying to decide what is best for all those situations where a frequency response graph shows both peaks and valleys. How do you solve the problem. You have to weigh off the loss of headroom from a boost against the loss of level from a cut.

In most commercial subs, it also may be hard to actually decide when you have entered into a loss of headroom situation. Clipping usually isn't allowed, there's some form of gain control at that level which causes compression of the sound. I wonder how easy it is to hear a slight compression at a specific overboosted frequency?

brucek

 

[Bob Christensen]

bruce: It would probably be fairly difficult to detect such compression in the lower frequency ranges we are talking about with quality subs. Especially considering that the vast majority of subs equipped with such circuitry also contain onboard amps designed to work in concert with the XO and boost circuitry. But in inexpensive ones.......

 

[Sonnie Parker]

I've been wondering about something concerning this issue of cutting vs. boosting.

If we had say a 10db dip that we were having trouble with and decided to bring all frequencies down to the level of the dip bottom we would then have a smooth line (whether it be flat or house). But then of course we have to increase the volume as suggested because we've cut a lot of our volume out with the massive cuts. Doesn't the dip come right back when we increase the volume?