There is a lot of confusion about units here, and it stems from the fact that the way of measuring levels is different from the old analogue world and the new digital one. But they both use the same (more or less) unit, the dB (decibel). If you understand this, it will help.
The decibel is a RELATIVE unit - by itself it means nothing. It needs to be referenced to something. There are common things it can be referenced to
- some sound pressure level (this is common when talking about actual sound levels)
- an analogue voltage (this is used in an analogue signal chain like an old style mixer)
- in a digital system, where signals are modeled as numbers, we reference to the BIGGEST number that our system can represent
These are some (not all) of the ways that dB's are used.
Now, dB's are basically ratios. When we talk about a positive dB number, that means we are BIGGER than the reference. When we talk about NEGATIVE dB's we are smaller.
In a digital signal chain we reference to the biggest possible number and so the numbers are always negative.
In ana analogue mixer we reference to some "nominal level" (you don't really need to worry about that, but it is typically some standardised number of volts - 0.775Vrms is common) and the actual signal could be bigger or smaller, so both positive and negative numbers are possible.
When we talk about a certain "gain stage" in a signal chain, that stage has the effect of amplifying or attenuating the signal by some amount - so that gain stage gets a number representing it's gain. Now this is a pure dB number.
And because dB's are log units - we ADD the numbers (don't worry about it - it is a math trick that just works).
So if we take an analogue signal of -60dBu (that's a real analogue signal that might come out of a microphone) and put it throigh a micamp with 70dB of gain, the output of the mic amp is +10dBu. OK so far?
Now, in a DAW, the biggest signal is 0dBFS (FS means "Full Scale", that is our biggest possible number). We want our typical average signal in our mix to be less than that. A typical number is -20dBFS.
This gives us "headroom" (think about getting your car under a low bar) of 20dB before we encounter digital clipping.
Now, if we set up a mix, let's say there are 3 signals, and we want the SUM of them to be -20dBFS - they all need to be somewhat less than that individually (how much exactly we adjust by ear when we mix).
The usual way to do this is to get EVERY signal, no matter how many to roughly that -20dBFS level, and then attanuate (i.e. apply negative gain) to each via the mix fader, so that the overall mix is also about -20FS. (This doesn't have to be exavt, but it is a guide.
What often happens is that as we build up a mix, adding more and more sources, the mix gets louder and louder until it starts to clip. Then we need to pull back all those faders by an equal amount (maybe 6dB each) so that they all stay more or less the same relative to each other, but the whole mix gets quieter.
In general your MONITORING chain (that is, headphone, power amp, speakers) should be setup to give you a sensible volume for your typical mix level (here -20FS). Changing monitoring volume should have no effect on your mix level! (but not vice versa).
I hope that helps. The key here is to more or less understand the fundamentals, so that you are able to take sensible actions when needed.