Ask an Acoustic Engineer (me)

[mauian]

Not a dumb question.

I used to be super crazy careful about amplifier gain settings and I used an oscilloscope and everything. I realized a single sine tone can be hit or miss like you mention because there might be boosts or cuts all over the place. So then I did a sweep and looked for the hottest frequency (electrically at the speaker terminals) then I used that sine tone to set the gains.

Later I realized if you set gains, then tune, chances are you're going to make a lot of cuts and so your gain structure will not be what you thought it was. The system can wind up a lot more quiet than you intended.

How often has this happened: you set gains, you spend all day tuning until it's just right, you either share with a friend or you listen the next day, you realize it is not loud enough, you adjust the gains, you spend all day tuning until it's just right .... and round and round.

I like one of Andy's recent posts about setting gains last. I wound up with a very similar idea before I saw his post. In short it goes like this:
-if your system is not loud enough, increase gains.
-if you system has too much noise floor or hiss, reduce gains.

There are many ways to get fancy about finding the balance between loudness and quality but many of them are difficult or impossible with simple tools like REW. I find the ears are the best tool for the job unless you don't know what to listen for, then you borrow your friends' ears.



I'm not sure how the DD-1 works, but the C.L.E.A.N. gain setting lights on the Rockford amps I use are pretty reliable no matter what I'm doing. A linear sweep might not be a bad idea, but similar to above I think it's a bit overkill unless you have no other way to listen for it.



You could, but setting gains is usually hunting for a threshold of clipping the input or output of a device and then looking for the resulting distortion to see when you found the threshold. White noise or any continuous noise would be a poor choice since distortion is much harder to see unless you have really sophisticated (read expensive) tools.

Not a dumb question.

I used to be super crazy careful about amplifier gain settings and I used an oscilloscope and everything. I realized a single sine tone can be hit or miss like you mention because there might be boosts or cuts all over the place. So then I did a sweep and looked for the hottest frequency (electrically at the speaker terminals) then I used that sine tone to set the gains.

Later I realized if you set gains, then tune, chances are you're going to make a lot of cuts and so your gain structure will not be what you thought it was. The system can wind up a lot more quiet than you intended.

How often has this happened: you set gains, you spend all day tuning until it's just right, you either share with a friend or you listen the next day, you realize it is not loud enough, you adjust the gains, you spend all day tuning until it's just right .... and round and round.

I like one of Andy's recent posts about setting gains last. I wound up with a very similar idea before I saw his post. In short it goes like this:
-if your system is not loud enough, increase gains.
-if you system has too much noise floor or hiss, reduce gains.

There are many ways to get fancy about finding the balance between loudness and quality but many of them are difficult or impossible with simple tools like REW. I find the ears are the best tool for the job unless you don't know what to listen for, then you borrow your friends' ears.



I'm not sure how the DD-1 works, but the C.L.E.A.N. gain setting lights on the Rockford amps I use are pretty reliable no matter what I'm doing. A linear sweep might not be a bad idea, but similar to above I think it's a bit overkill unless you have no other way to listen for it.



You could, but setting gains is usually hunting for a threshold of clipping the input or output of a device and then looking for the resulting distortion to see when you found the threshold. White noise or any continuous noise would be a poor choice since distortion is much harder to see unless you have really sophisticated (read expensive) tools.

Thanks, Justin. I laughed when you described the scenario of tuning and then realizing it’s too soft (sometimes I have had it be too loud) and then having to adjust the gains. It’s funny (and reassuring) to hear that this is not an uncommon experience (as I have experienced it for sure, many times). Thanks!


Sent from my iPhone using Tapatalk

[Jdunk54nl]

Thanks, Justin. I laughed when you described the scenario of tuning and then realizing it’s too soft (sometimes I have had it be too loud) and then having to adjust the gains. It’s funny (and reassuring) to hear that this is not an uncommon experience (as I have experienced it for sure, many times). Thanks!


Sent from my iPhone using Tapatalk

After many conversations with Andy and with Justin, I've kind of modified Andy's (and Justin's) gain setting with Skizer's advice. I set my tweeter/mid gains at a maximum without noise. As in, if I turn up the gain on the amp any louder, I am going to get too much noise for my personal preference. I lower gains in the DSP like Andy suggests (-12db input, and -12db on output). I then use my RTA and adjust the other gains to match the curve to whichever speaker (tweeter or mid) that was the lowest output without noise.

I then tune and raise the input side on the dsp first and then the output side if I am not happy with the volume level. I know that I can't turn up my amp gains any more because that would just introduce more noise into my tweeters/mids.

Note: This is with a high pass on my midbass of around 100hz and if you read Andy's info on how much power speakers actually get with music/pink noise, how a crossover drops it either -3db (butterworth) or -6db (Linkwitz) MORE at the crossover point, it will help to understand why I am not worried about sending a 0db test tone to any of these speakers and checking for clipping the amps outputs. I am willing to bet that none of these speakers will EVER see any musical note that is 0db.

If we assume an 80hz high pass, the music will be down about -6db at that point naturally. If you throw in a Butterworth or Linkwitz filter you are looking at -9db with a Butterworth or -12db with a Linkwitz at 80hz.

Here is what pink noise (and most of music) does assuming we start with 0db at 20hz.
20hz 0db
40hz -3db
80hz -6d
160hz -9db
320hz -12db
640hz -15db
1280hz -18db
2560hz -21db
5120hz -24db
10240hz -27db
20480hz -30db

Here is a song analysis of one of the "worst" songs I could find that had the most 0db tones. It is Billie Eilish "Bad Guy".



Here is a good chart for power and music (same as the pink noise stuff above, just puts it in terms of power)

[mauian]

Hey Justin,

In the case of tweeters pointed up at a windshield, what does this do to the sound? Obviously it makes it reflected, but does it color it any way? Disperse it? How can it affect imaging? Any reason other than convenience or cost that an OEM would do this? Does a location like this require anything from a tweeter or other speaker? (In order to sound good)

Thanks!


Sent from my iPhone using Tapatalk

[dgage]

Hey Justin,

In the case of tweeters pointed up at a windshield, what does this do to the sound? Obviously it makes it reflected, but does it color it any way? Disperse it? How can it affect imaging? Any reason other than convenience or cost that an OEM would do this? Does a location like this require anything from a tweeter or other speaker? (In order to sound good)

Thanks!

Courtesy of ErinH. The main thing you need to handle is the angle of the speaker to the windshield so the sound comes off properly and not too shallow or steep.

http://www.jjracoustics.com/128-AES_2010.pdf

[mauian]

After many conversations with Andy and with Justin, I've kind of modified Andy's (and Justin's) gain setting with Skizer's advice. I set my tweeter/mid gains at a maximum without noise. As in, if I turn up the gain on the amp any louder, I am going to get too much noise for my personal preference. I lower gains in the DSP like Andy suggests (-12db input, and -12db on output). I then use my RTA and adjust the other gains to match the curve to whichever speaker (tweeter or mid) that was the lowest output without noise.

I then tune and raise the input side on the dsp first and then the output side if I am not happy with the volume level. I know that I can't turn up my amp gains any more because that would just introduce more noise into my tweeters/mids.

Note: This is with a high pass on my midbass of around 100hz and if you read Andy's info on how much power speakers actually get with music/pink noise, how a crossover drops it either -3db (butterworth) or -6db (Linkwitz) MORE at the crossover point, it will help to understand why I am not worried about sending a 0db test tone to any of these speakers and checking for clipping the amps outputs. I am willing to bet that none of these speakers will EVER see any musical note that is 0db.

If we assume an 80hz high pass, the music will be down about -6db at that point naturally. If you throw in a Butterworth or Linkwitz filter you are looking at -9db with a Butterworth or -12db with a Linkwitz at 80hz.

Here is what pink noise (and most of music) does assuming we start with 0db at 20hz.
20hz 0db
40hz -3db
80hz -6d
160hz -9db
320hz -12db
640hz -15db
1280hz -18db
2560hz -21db
5120hz -24db
10240hz -27db
20480hz -30db

Here is a song analysis of one of the "worst" songs I could find that had the most 0db tones. It is Billie Eilish "Bad Guy".



Here is a good chart for power and music (same as the pink noise stuff above, just puts it in terms of power)

Great info, jdunk. Thanks for sharing your process for setting gains. Having it laid out step by step like that helps it make a lot of sense (and of course helps me reproduce it). Much appreciated!


Sent from my iPhone using Tapatalk

[Justin Zazzi]

Hey Justin,

In the case of tweeters pointed up at a windshield, what does this do to the sound? Obviously it makes it reflected, but does it color it any way? Disperse it? How can it affect imaging? Any reason other than convenience or cost that an OEM would do this? Does a location like this require anything from a tweeter or other speaker? (In order to sound good)

I really like the paper that dgage linked. Can you take a look at that one and then come back with more questions?

[LumbermanSVO]

Or clipping has tremendous amounts of high-frequency energy and it will melt your tweeter.

Excuse the terrible drawing :-)

I played with a scope and amp once to see just what actually happens when the amp clips. Most people think the red line is what happens, but my testing showed that it's much more like the orange line. I think the short spikes could get to the tweeters through a passive crossover and damage them, but I never felt like blowing up gear to test that theory.

[Jdunk54nl]

Excuse the terrible drawing :-)

I played with a scope and amp once to see just what actually happens when the amp clips. Most people think the red line is what happens, but my testing showed that it's much more like the orange line. I think the short spikes could get to the tweeters through a passive crossover and damage them, but I never felt like blowing up gear to test that theory.

You will find this interesting about clipping..it's by Andy.
https://www.facebook.com/10000064042...685530618/?d=n

This is also from my oscope.

[LumbermanSVO]

You will find this interesting abiut clipping..it's by Andy.
https://www.facebook.com/10000064042...685530618/?d=n

This is also from my oscope.

Interesting, I'm curious if amp design has any effect on if you get those little spikes or not.

[mauian]

I really like the paper that dgage linked. Can you take a look at that one and then come back with more questions?

Great paper. Questions answered ;-)


Sent from my iPhone using Tapatalk