To Gain or not to gain? That is the question

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I'm hoping I can get a couple thorough technical explanations, ideally with graphs to help clarify the following:
If the voltage from a source is high enough to achieve the desired volume level from a headphone, why does the amplifier stage need gain?
Gain adds undesirables: noise and distortions.
Why not just buffer the source and add a volume control?
I use Sennheiser HD 800 high impedance headphones yet find that I rarely go beyond half way on most headphone amplifiers.
I set my current one to zero gain and use most of the pot range.
Please help me understand the benefits of gain
 
Moderator
Joined 2011
Only enough gain is needed to be able to drive the power amplifier to full output
with the volume control at its maximum setting. If more gain than this amount is used,
there could be more noise and distortion than necessary, as well as possible clipping
in an earlier stage, depending on the system gain structure. With a commercial product, most
possible user situations must be covered, so often there will be excess gain in the system.
 
Last edited:

PRR

Member
Joined 2003
Paid Member
"Turn it up to 11 !!" If you can't turn it WAAYY TOO LOUD, it's no dang good.

In a past life, capturing wild untamed sounds, I contrived a headphone amp with a wide range of well-controlled gain. If OTOH all your sounds exactly fill a 2V pipe you may need no voltage gain at all.
 
The usual, voltage gain.
Most headphone amps have at least several dB of gain that is "said" to improve things like headroom, dynamic range, and "drive".
I'm not a believer yet it's prevalent in all headphone amps. With the mention that if you have large, high impedance phones, to use the higher gain option....
Why is this so if the volume from line level source is adequate for ear splitting loudness?
PS- Output impedance doesn't seem relevant
 
Keep in mind that there are some headphones with sensitivity 15 dB lower than yours (as well as 30+ dB higher). If you want even a standard 103 dB SPL peak from some 85 dB / V cans, you'll need up to 8 Vrms, plain and simple - that's 12 dB of gain from a 2 Vrms output. Your Sennheisers will do 108 dB SPL with 2 Vrms, so if you are running a CD player straight into the amp, chances are unity gain will do just fine.
(Also, from my experience I would generally describe Australian mastering levels as high to very high and a few years behind the curve in this respect, with ReplayGain values around -11 dB still not being uncommon, although indie circles can be substantially more sensible these days. The Kiwis seem to value their sound quality and dynamics substantially more as far as I can tell from a very limited sample size.)

(And no, I personally never listen that loud either. I would probably get along fine with peak levels in the low 90s.)

Also keep in mind that not everyone is running full output all the time - ReplayGain alone will cost me about 10 dB a lot of the time, and I think shared mode another 3 dB but I'm not sure about that any more - so that's potentially less than 0.5 Vrms @ 0 dBFS. If I needed full output from an amplifier that does 8 Vrms, I'd be looking at a 24 dB gain requirement. Not common. 12 dB for 2 Vrms output, a lot more common.
That 10 dB value was with pop/rock music. Classical generally is turned down much less (maybe about 6-8 dB less), but overall dynamics tend to be greater so average levels are lower still.

Headphone amps need to manage a moderate amount of input and a large amount of output dynamic range (about 100 dB and 130 dB, respectively). It is virtually impossible to make a universal one without at least two different gain settings; you arguably even need a two-stage amplifier with input and output gain settings (like the bigger Lake People / Violectric ones). Since the output amplifier is generally noninverting due to noise considerations and hence has a minimum feasible gain of unity, an inverting input stage may provide sub-unity gain if necessary while its higher noise is generally attenuated by the volume pot.

An interesting case to study might be the RME ADI-2 DAC, which combines automatically switched output stage gain with a dedicated lower-voltage IEM output in addition to the regular headphone output.
 
Last edited:
Useful info @sgrossklass
So the real reason for gain is to permit less sensitive headphones to reach appropriate loudness.
If you read hi-fi reviews of amps, and audio blogs, they make it sound like gain improves the sound somehow.
PS - Interesting DAC the RME. Never heard of that brand.
PSS - Why not just raise the amplifier volume instead of use ReplayGain?
 
If you read hi-fi reviews of amps, and audio blogs, they make it sound like gain improves the sound somehow.
Well, I mean - if you actually need it, it does. ;)
PS - Interesting DAC the RME. Never heard of that brand.
They're a German pro audio manufacturer. Not cheap but quality. RTFM tends to be very much advised though if you want to take full advantage of their gear.
PSS - Why not just raise the amplifier volume instead of use ReplayGain?
I think you don't know what ReplayGain is - it's a volume normalization system for music player software. You first have to scan your music, upon which the algorithm will determine perceived loudness level and calculate an adjustment to hit a defined target loudness which is written into a dedicated tag (actually, two different ones plus two for keeping track of peak levels, one for track and another for album each). Upon playback, the player will then read that value and adjust playback volume accordingly. There are a handful of players supporting scanning, like Foobar2000; reading the tag information is more widely supported.

In practice, the target loudness level in Foobar2000 is conservative enough to rarely give trouble with quiet but dynamic music that is being turned up beyond 0 dBFS peak (I have a handful of '80s releases that are pushing it, so I dialed in an extra -3.2 dB), but in return very loud, heavily compressed modern-day releases are being turned down a fair bit.
The average Australian major label release would be around -10..-11 dB more often than not, I'm still seeing a number of these in US releases as well. Mind you, stuff with guitars tends to generate several dB higher perceived loudness values in general. Still, the Beth's latest Jump Rope Gazers is sitting at about -3.5 dB, pretty much a best case result as it always stays well away from 0 dBFS (RMS average is about -14 dBFS)... could have been mastered in 1989 as well. I'm seeing even lower values in classical music only, often around -1 or -2 dB.

If memory serves, Foobar2000's reference level already is 6 dB higher than what ITU R.128 (as used in broadcasting) originally suggests. They kept that for compatibility with the original RG algorithm even after transitioning to a R.128 based one.

the designer preffer for an amplifier to have some gain anyway as it improves stability, psrr and cmrr...
Good point to bring up - historically it was difficult to make an amplifier stable at low gain without also reducing output slew rate disproportionately. Resulting slew rate distortion, also called TIM distortion, was identified as a problem in the late '70s and addressed fairly quickly. Still, this is why decompensated op-amps exist in addition to their unity gain stable counterparts.

So if you take the same amplifier circuit and compensate it for use at lower and lower gain, you eventually run into the problem that it can't deliver its maximum output level at or beyond 20 kHz with less than a specific amount of distortion any more. (See power bandwidth rating for amplifiers.)

CMRR and PSRR are influenced by gain-bandwidth after compensation as well as actual circuit gain. In practice you're only going to use a unity gain compensated part at moderate gain and decompensate for higher gains, so overall it should be pretty much a wash really.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.