DIYMobileAudio.com Car Stereo Forum banner

A tutorial on gain setting and amplifier power usage

90756 Views 40 Replies 18 Participants Last post by  chad
Let's start with some simple concepts. Your source, whether it be a tape player, dvd, or a cd player has a maximum rated output voltage. That output is too weak to directly power a speaker, so in turn it's sent to an amplifer, which well... amplifies that signal and sends that along to your speaker.

What the "gain knob" does is matches the amplifier's input sensitivity to the maximum output of your source. Simply put, it prevents your headunit from overdriving your amplifier by telling your amp what is the max voltage/signal that it can expect from your headunit. At that point, your amp will achieve maximum power output without being overdriven into distortion/damage.

So if your headunit is capable of 4v, and you set the amp's gains to 4v then your amp will make max power when your headunit sends it a 4v signal. Sending it anything more than 4v will cause excess distortion and heat, since you're now exceeding the maximum power rating for your amp. On the other hand, setting the gains to 8v when you're headunit's max output is only 4v will cause your amp to never reach maximum power output.

Now, take a look at the following song:



This shows you the amplitude levels of the song over time. Notice that it fluctuates quite a bit, with some parts obviously much louder than others.

The ratio between the average level of the entire song and it's loudest point is called the "crest factor".

You can see that there's a problem here. If we were to set our amp's gain to the max output of our headunit, the only time our amplifier would be making it's max power output would be during those few, brief moments in the song when the amplitude is at it's peak. We can clearly see that for most portions of the song, the average amplitude is considerably lower than it's peak resulting in a large portion of our amp's power output being wasted. Remember that a doubling of power is equivalent to 3db of gain. Imagine that your typical song has a crest factor of 12db or more, and you can easily imagine the considerable waste of power if your gains were to be set in such a manner. For a 16w amp, you would only be outputting an average 1w... with volume set to max on your headunit.

So what can we do? Generally, I prefer to set my gains by ear. I set the gains at a lower output level than what my headunit is capable of producing, up until the point I can "hear" the amplifier distorting. At that point I back off a bit. Now the amplifier is still going to distort, but only on those very brief high amplitude peaks in the music. Everyone's ears are different. It's best to choose a song with a high crest factor to ensure the least amount of amplifier distortion with the widest range of music, and to let your ear be the judge of what distortion levels are audible to you. This way, you can be sure to utilize as much of our amp's power output as possible.

Also consider the practical fact that as you turn up the gains on your amp, you generally begin to pickup more hiss/noise. Having a more powerful amp will allow you to keep the gains down, and thus reduce noise.

Regardless, with music it's fairly obvious that your amp does not constantly output it's maximum continuous power output. Looking at the example song above, the amplitude fluctuates quite a bit. So when people claim their speakers are taking the full rated output of their amp's on a continuous basis, it's probably not true.

Also, bear in mind that we rarely use one speaker in our setups. We have mids, tweeters, and subs. Generally, the power output to tweeters and mids is fairly low. Those huge spikes you see in the music are bass tones. This is because our ears are less sensitive to bass frequencies, so when recording a low frequency and a higher frequency tone of the same spl, the low frequency tone is going to have much higher amplitude.

For anyone curious to get a more realistic measure of how much power is needed or used, I suggest using 0db sine waves to test your system. Hook up a voltmeter across the terminals of your tweeter or mid and play the sine wave. You'll be surprised to find how little power it takes to get your speaker extremely loud above 200hz. Usually a few watts at 1m on your typical tweeter at 3khz is enough to cause pain.
See less See more
1 - 20 of 41 Posts
so when you are testing your tweets by ear do you only listen to the tweets? Same for the mids and subs?
npdang said:
...our ears are less sensitive to bass frequencies, so when recording a low frequency and a higher frequency tone of the same spl, the low frequency tone is going to have much higher amplitude.
Wait, aren't amplitude and SPL the same? Loudness is a subjective measurement, but amplitude and SPL are scientific, and are a measure of the same thing (wave pressure).

Dan
amplitude is a measure of electrical current, and sound pressure level is a measure of acustical energy. completely different.
I usually listen to each pair of drivers separately, before level matching. Once the gains are set as high as they can be, I use my processor to level match. You can tolerate quite a bit of clipping from a bass amp, but not too much on your tweeters. Tweeters tend to pick up audible hiss if the gains are raised too high though in my experience.
So Each set of drivers should be set first by ear using the amplifier gains....Then put together and adjusted together with each set of channels independant gains (for level matching).

I ask this because I will be level matching my CA18's and LPG 26's soon. Using my stock HU should i turn it all the way up, or 3/4 of the way up and then adjust my gains as loud as it goes untill there is audible distortion or untill i think that they are just tooooooo loud. Because i dunno how loud they are going to be before i can audibly hear any type of distortion.

What would be a step by step gain setting routine you would do, including any HU volumes, Processor volume changes for level matching, and amp gain setting and such. And what if everything is too loud to accurately hear any distortion?
I usually set the volume on the head unit as high as it will go without clipping the outputs. Most head units now don't clip the outputs, but you never know so maybe a little lower than full volume. This way you get the highest voltage output from the head unit, which is better because a higher voltage signal is better at resisting induced noise.

Npdang said:
What the "gain knob" does is matches the amplifier's input sensitivity to the maximum output of your source.
Step by step?

1. Find a heavily compressed cd. Pop music with heavy basslines and blaring instruments is good.
2. Set headunit to max unclipped volume.
3. Make sure no eq is set and that your crossovers are set to protect the drivers from damage. Also all level controls should be set to max.
4. Listen to each pair of speakers (tweeters, mids, sub) separately. Turn the gain up until it's either too loud to stand, or you hear signs of distortion, then back off a bit until it sounds ok.
5. Now go back to your headunit/processor/amp and level match all the speakers by reducing the output of drivers that are too loud.
What is good folk. I got to say that this is a great site. Ok now I used this site to set my gains( http://www.resnet.trinity.edu/areynol1/gain.htm )(option 3) and I have to say I have been very pleased with the results. I just make sure that I underrate the power on my amps and turn my volume to 30.
npdang said:
Step by step?

1. Find a heavily compressed cd. Pop music with heavy basslines and blaring instruments is good.
2. Set headunit to max unclipped volume.
3. Make sure no eq is set and that your crossovers are set to protect the drivers from damage. Also all level controls should be set to max.
4. Listen to each pair of speakers (tweeters, mids, sub) separately. Turn the gain up until it's either too loud to stand, or you hear signs of distortion, then back off a bit until it sounds ok.
5. Now go back to your headunit/processor/amp and level match all the speakers by reducing the output of drivers that are too loud.
noob here.. how to listen each pair of speakers separately?? do i have to remove all the connections from the amp?
Yea or just pull the Patch coords from the individual channels you want to kill for the time being.
I just had an addition to suggest for setting gains on sub amps - because I definitely also like to set my gains by ear as well:

With all my amps, I set my gains using test tones - you can pick up CD's with test tones recorded at the maximum -0dB level, and I find this real easy to pick out when distortion starts, with pure tones.
Anyway - with subwoofers-

Bearing in mind that the lower you go in frequency, the less sensitive your hearing is to distortion (did you know with subwoofers, most people can't hear distortion until it reaches 5%-10% levels - even trained ears? :eek: ), I like to use a frequency that's actually well outside my subwoofer's normal playing range... but not too terribly high, because the sub's own inductance will start to roll off higher frequencies...

My sub amp gain setting usually goes like this:
1) turn off the sub amp's low-pass filter (or whatever LP filter is engaged, for the sub's signal)
2) play a 200hz-300hz test tone, adjust the gains listening for distortion
3)turn the low pass filter back on and call it a day

Of course your sub isn't going to ever play 200hz-300hz (outside of this test), but this doesn't matter, since all we're doing is adjusting the size of the doorway that we are allowing the signal to pass through on the way into the amp... and the amp has a frequency response (power output) that's flat across this whole range. :cool:
See less See more
Casedot said:
amplitude is a measure of electrical current, and sound pressure level is a measure of acustical energy. completely different.
amprage (sp?) is the measure of electrical current.
amplitude is the peak of a wave.
SPL is a measure of acoustic intensity.

setting gains by ear seems like a good idea.

there are issues with the DMM method. i'm working on improving the DMM method by allowing for clipping detection without the need for an OScope.
thechris said:
amprage (sp?) is the measure of electrical current.
amplitude is the peak of a wave.
SPL is a measure of acoustic intensity.

setting gains by ear seems like a good idea.

there are issues with the DMM method. i'm working on improving the DMM method by allowing for clipping detection without the need for an OScope.
A typical clipping indicator can do the trick. But the easiest way would probably be to use a relatively wideband notch filter and a single test frequency. Then measurement of power content would be an approximation to harmonic distortion levels. In other words, a makeshift distortion analyzer to detect the onset of clipping.

I just can't see where a DMM, OScope, or any other form of equipment would have any application in setting gains. The gain control is basically there for tuning purposes, to set the relative levels of all your channels if you don't have an outboard processor doing the work. There's really no other reason for adjusting it. You set it initially to be roughly where you want the volume knob to be when you clip a "typical" signal. And that's about it.
The only real use a clipping indicator has it to either verify after the fact that your gains are set properly so as not to clip, or to set a ceiling before the fact to prevent clipping. Once you establish the ceiling, you still have to set your actual gain levels by ear during the tuning process.
I set by ear and spectrum analysis. And I may get flamed for this.... I could give two ****s if I get a little clipping. On large coil drivers (larger mids and subs) the inductance is such that any upper harmonic content of a clipped signal will see a very high impedance load thus not be translated into real power. A little clipping is not bad, as long as it's distortion characteristics don't exceed that of the drivers distortion, which is a bunch!

Those who have observed clipping indicators in action need to understand a couple things. 1. They are delayed to stay on for a while after the actual incident if simple peak clipping is occurring. If they were not timed out this way you would barely if never see them. 2. Sometimes they are set up using a comparator circuit, meaning that it looks for a different signal from the input to the output sans gain and illuminates the device when a difference is sensed. This is the best type. Others are simply a single ended design where the engineer assumed that the amp would clip at a preset voltage level be it input or output, those suck because they don't take into the account of complex output loading or voltage variations on the mains inputs.

I've been doing this long enough that I don't mind seeing a little flicker of the clip light on lows, some mids and all tweets is a different story, if you are clipping your HF amplifier then something is wrong, you are deaf, or you are about to be :) My Sub and low amps in my PA flicker the clip lights every night and have been doing it now for almost 15 years without ever having a failure. Most LF failures I have had were suspension and lead wire failures and the coils looked great after being removed from service. I have never blown one of my JBL subs. But I did tear up 8 of the "research" Peaveys before we got it right, I've been on this set of peaveys now for a while, all have survived. The new sub project cold prove interesting :)

Set them by ear, or a SA, then tweak from there. if you are clipping amps bad then take it as a lesson learned and buy bigger amps :)

Chad
See less See more
SQ_Bronco said:
The only real use a clipping indicator has it to either verify after the fact that your gains are set properly so as not to clip, or to set a ceiling before the fact to prevent clipping. Once you establish the ceiling, you still have to set your actual gain levels by ear during the tuning process.
You don't want to set a ceiling. If you do, then you don't have the extra swing for low-volume tracks. Unless you use the lowest volume track you'll ever use to determine the ceiling, but how would anyone be able to know that?
MarkZ said:
A typical clipping indicator can do the trick. But the easiest way would probably be to use a relatively wideband notch filter and a single test frequency. Then measurement of power content would be an approximation to harmonic distortion levels. In other words, a makeshift distortion analyzer to detect the onset of clipping.

I just can't see where a DMM, OScope, or any other form of equipment would have any application in setting gains. The gain control is basically there for tuning purposes, to set the relative levels of all your channels if you don't have an outboard processor doing the work. There's really no other reason for adjusting it. You set it initially to be roughly where you want the volume knob to be when you clip a "typical" signal. And that's about it.
oddly enough, that's what i've been working on. it seems even a simple twin-t can give sufficient contrast to determine the point of clipping. alternatively you can use unfiltered square waves and the digital volume controls on the HU and just take sucessive measurements. if the amp clips, the ratio between volume settings will change, and quickly end up being 1.

the point of clipping becomes applicable for determining the optimal HU volume setting to use. if the HU outputs a clipped signal, then the volume should be lowered.

also, some people prefer to set the gains on the amps then use the HU for level controls.

I definately agree that the ear is the final judge though. the DMM, OScope, or clipping detector are all just tools used to quantify things.
How does one tell when the outputs of the deck are clipping?

I too have always set my gains be ear and have my amps near 0 gain (Sony XM2000R)
Honestly, it's tough. Rednek rule of thumb, if you have to really open up your deck to get it louder than all get-out you are doing good. If you barely crack it and it loud or can't find a comfortable listening level when the volume control is lower then back off the amp gains. If it just won't go loud enough then open the amp gains up a tad.

When I actually use test gear, (read bored) I'll overlap the headunit gain by about 6 dB or more from where the amp clips. This allows for source material that was not normalized well. You will not be happy with a system where eveything clips at 0dBFS.

Chad
1 - 20 of 41 Posts
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top