Source vs Amp Volume

[BradH]

Not sure if this belongs here, but for me at least it pertains to multi-way speakers.

It seems that maybe some speakers come alive when you give them a bit of juice. At the risk of venturing into a very subjective realm, do you think some speakers might sound better if you reduce the input level (ie. laptop output) and increase the amp output vs having laptop out at full and controlling volume through the amp?

 

[SpeakerBob]

I'm sure it varies between manufacturers but I've always assumed that a laptop output on full chat would be distorting so have always set mine to around 70 percent to give some headroom.

 

[AllenB]

It may reduce distortion and increase noise.

The differences you suggest could be caused by a number of things. Driver levels due to voice coil heating can change. The drivers can distort differently at higher levels. The amps can introduce various distortions between them.

Another way of doing this is to make sure neither the speakers nor amps produce any undue distortions, but introduce some consistent distortion somehow.

 

[SpeakerBob]

Maybe have a read up on gain structure. Basically you need the laptop output as high as it can go without distorting itself, but also not so high that it distorts the amp input.

Rob.

 

[Liquid Arthur]

I'm sure it varies between manufacturers but I've always assumed that a laptop output on full chat would be distorting so have always set mine to around 70 percent to give some headroom.

What he said. When I'm using my laptop as a source I set the volume to about two thirds, adjust the volume on the amp, and enjoy.

 

[puppet]

If your laptop has output metering you should set the levels 6-12db below the loudest program material you'll be encountering to prevent input clipping while maximizing S/N ratio. A passive system will then use the amp's output to adjust SPL's.


Carefully designed active systems are similar but in that case the amp(s) should be the first to clip and the SPL's are are adjusted at the output section of the active processor. Signal input levels are treated the same to take full advantage of the active processors S/N ratio. Input levels shouldn't be clipping either way.

 

[JMFahey]

Not sure if this belongs here, but for me at least it pertains to multi-way speakers.

It seems that maybe some speakers come alive when you give them a bit of juice.

And that´s the point.
Speakers played louder sound "better" because our ear response is non linear and frequency sensiyivity improves at higher level.

It´s strictly a function of actual SPL reaching our ears and not related to "the number" shown by our volume pot knob.

It is not a magical property the volume potentiometer has.

At the risk of venturing into a very subjective realm, do you think some speakers might sound better if you reduce the input level (ie. laptop output) and increase the amp output vs having laptop out at full and controlling volume through the amp?

If you end up listening at the same Acoustic volume (SPL) as I presume, the answer is no.

 

[celef]

Is there no other explanation to that some speakers need higher spls then others to come alive then it is all up to our hearing?

I can imagine that the suspension of loudspeaker drivers has some to do with it, just like some cars and their suspension settings, some are so insensitive or stiff that there must be big holes in the road surface or need for high speed to get any form of response

 

[Douglas Blake]

It seems that maybe some speakers come alive when you give them a bit of juice. At the risk of venturing into a very subjective realm, do you think some speakers might sound better if you reduce the input level (ie. laptop output) and increase the amp output vs having laptop out at full and controlling volume through the amp?

Why do some speakers appear to "come alive" at higher volumes? That's probably 50-50 between the way voice coils work and the way our ears work. My current speakers are rather low sensitivity (88db) and it does seem to take a bit of push to get them sounding their best.

The amplifier's volume control is not magic. It's a simple voltage divider between the input signal and the first stage of the amplifier. It doesn't matter if you have the amp volume down and the laptop volume up... 10 watts out is 10 watts out.

Because I have neighbours (who doesn't?) I took a strategy of treating the amplifier's volume control as a pseudo gain control, setting it to the maximum I think I can get away with. From there I adjust the actual listening volume, up to the limit set on the amp, using the software in my HTPC. That way, even in my most enthusiastic moments I don't exceed the bounds of reason ... or end up with the cops at my door.

 

[Ugg10]

I am happy to be educated but I have always thought that the best way is to keep any digital volume control at max (or close to it) just in case the software is reducing the volume by dropping bits thereby reducing the fidelity of the digital signal (sorry my terminology may be a bit off).

Therefore I have the source app (Spotify) at max, the chromecast audio or pc (windows) volume at max and then control the volume of the speakers via analogue preamp between the dac and the power amp (passive in my case so just a ladder attenuator i.e. single resistor in line, between the dac and the power amp). I also try and keep the digital signal digital until the last minute if possible as well.

 

[puppet]

Keeping the digital at max is dangerous especially for a tweeter. Aside from the unnecessary distortion, program peaks can/will pooch the driver.

 

[Ugg10]

Keeping the digital at max is dangerous especially for a tweeter. Aside from the unnecessary distortion, program peaks can/will pooch the driver.

I am not sure I understand, how do you get distortion in a digital signal, isn’t it just 1’s and 0’s that define a digitised waveform. If the volume is set to max you get 16 bits in and 16bits out, with some digital volume controls this is reduced to say 12 bits at lower volumes and resampled reducing the fidelity.

So, source data is 16bit (for arguments sake) digital this is sent over internet to my pc or streamer, the 16bits then get converted to an analogue signal in the dac which outputs a nominal 2v signal if all 16bits are set To a 1 i.e. full volume (which rarely happens in recordings). This is then attenuated by a preamp down to say 0.25v that the power amp then multiplies by approx 30 times to drive the speaker with approx 7.5v with only a portion of that going to the tweeter through the crossover.

I’d be grateful if you could explain where the digital distortion comes in so I can understand what I have been doing wrong all these years and have been lucky not to blow a tweeter on either passive or active speakers that I have used.

Thanks in advance.

 

[4real]

I’d say it all depends..

For an integrated laptop output I’d keep the output at max usually. If it distorts there is a design problem.

When using an external DAC, e.g. via USB, it also depends. Some come with an integrated volume control, others rely on the OS volume control, meaning they will throw away bits. With DAC volume control it obviously depends on the DAC implementation.

Also OS’es might differ in volume control quality. Do they use proper dithering? Is it resampling? If so, is it any good? That is the reason most people tend to use some kind of player that circumvents the OS and has full control of the hardware (stuff like Jriver or foobar).

Also, not all songs will play just as well on any speaker at any volume. Generally they will not all be mixed at the same volume, therefore you’ll almost always be off in regards to the equal loudness contour. It would be interesting to have a reference volume and impulse response of the reference system of the song metadata so one could actually reconstruct the proper response for any volume given the equal loudness contour.

 

[puppet]

I am not sure I understand,

Thanks in advance.

It's a matter of input levels. Digital set at max does not allow for program peaks. Your DSP will clip internally when these peaks are encountered. Clipping = distortion/noise. Amps (only) should clip first. Otherwise, your amplifying the distortion further.


This is gain structure 101

 

[4real]

It's a matter of input levels. Digital set at max does not allow for program peaks. Your DSP will clip internally when these peaks are encountered. Clipping = distortion/noise.

No it won’t (usually). First off calculations are usually done in floating point domain with quite a few extra bits to retain resolution. When you then go back to integer usually the DSP software will take into account the maximum gain a filter will have and lower the level by the same amount. If not you’ll have to account for it yourself. Better do that inside of the DSP than outside. You’ll retain more resolution this way.

 

[Ugg10]

What dsp would that be if I have the following -

Spotify > Chromecast audio > optical out > dac > passive preamp > power amp > speakers?

I guess if I put a minidsp nanodigi in between the CCA and the Dac(s) then I would have a dsp in the chain, but without ......

 

[4real]

Well, in that case the DAC is the “DSP” (chances are huge it is a 1-bitter (or variant), so will do digital filtering and over sampling). The same principal applies: give it the best possible signal, meaning volume at 100%. In such a case that when playing a Lossless file, the optical bits are actually the same as the lossless audio bits. Back in the old days one could play an AC3 signal to the receiver to validate lossless playback..

 

[BradH]

I'm more confused than when I started. lol.

I go digital from my laptop via HDMI to the receiver.

So 16 bit is 65,536 values. What are these values used for? I've heard 'volume' But then for each unit of time how does it know which frequencies are 'on' and the amplitude of each frequency? I Skimmed a couple youtube videos but they didn't go into that kind of detail.

Someone said the 16 bits is IEEE floating point. What is the benefit of floating point in this scenario over unsigned binary?

My original thinking was just that some drivers, maybe low efficiency ones responded better to the waveform when there's more power from the amp going to them. So same volume level but one uses more power. Does this even make sense?

Please forgive me, I'm sure it's painfully clear how little I know about this. I'm not completely ignorant, I'm just not trained in EE. If I have time to crack open a book (or the internet), and properly teach myself something, 9 times out of 10 it's going to be about software, since that's what pays my bills. I lean on the community here to help me learn these things for pleasure.

 

[academia50]

The reserve power margin will always be welcome (if we understand what we do with the volume control).

There are three situations in the pairing of an amplifier with a pair of cabinets.

Excess power with low sensitivity cabinets.
(dangerous)

Low power with low sensitivity cabinets (high distortion and lack of dynamic range)

Slightly excess power with high sensitivity cabinets.

What do you guess which one I prefer?

 

[SpeakerBob]

My original thinking was just that some drivers, maybe low efficiency ones responded better to the waveform when there's more power from the amp going to them. So same volume level but one uses more power. Does this even make sense?

No it doesn't. You cannot 'give' a driver more power with the same volume level unless you are heavily into power compression.