Absolutely no one cares about the mathematical purity of a unit like the decibel. It's a hammer that turns log stuff into easy to work with linear stuff. It's x10 because that turns into reasonably easy to use numbers with enough precision to use casually. Otherwise it's just a log ratio. Any suffix is a log ratio to a reference value for the entire system (dBv / dBm etc).
Say in a 50 ohm system I have an level 7 mixer then I need 7dBm into the port. My oscillator kicks out 0dBm (1mW). Then I need a 7dB gain block. I go to the datasheet for the MMIC and pick out a bias network that gives me around 7dB. I grab a eval board for the MMIC, solder the resistors on it, shove this on my VNA and throw 0dBm in one end and see if 7dBm comes out of the other end at the frequency I need.
It's like Laplace. Do you really want to spend hours doing calculus?
Source: actual electrical engineer, not some clueless hobby blogger. I really wish people would stop writing authoritative sounding articles on stuff they clearly have little experience with.
It's not a case of mathematical purity, it's a case of unnecessary ambiguities. Decibels would be a heck of a lot more comprehensible if a) it was always a ratio or an absolute value with respect to an explicitly written unit, and b) it was always in that unit and not sometimes scaled differently because it's power instead of amplitude, and it would be no less useful to you for exactly the case you're talking about. (RF at least generally does say 'dBm' when talking about absolute power, but it would probably be better to way dBmW, and we could also use dbmV where it made sense without having to go 'oh, but that's divide by 20 just because').
Of course, it's far too late to easily change it now, but it doesn't make the current situation optimal.
Have you considered the world outside of your expertise? Or the legibility of your field to people adjacent to it?
dBs work well for RF once people get used to a few quirks. The problem arises when it gets applied to other fields as units, or when people from adjacent fields want to cooperate with people from the RF field.
This is the correct take. These logarithmic scales, in all their various forms, reduce the math involved to something tractable. The reduction is so great that there is no substitute. Griping about the numerous arbitrary reference values and their obscure and ambiguous notations is fine, I suppose, and perhaps fixing this has some value. But ultimately, logarithmic scales are a tool that will be utilized regardless of whatever frustration they create among amateurs. Also, there is certainly the risk that whatever "fix" one might imagine will only produce yet another, parallel "standard."
Since you deal in RF, I thought I'd mention an interesting device frequently found in such systems: the logarithmic amplifier, or log amp. I'm certain you are familiar with these, but others might find this interesting. These devices are often used in Automatic Gain Control (AGC). Real signals in RF span a very wide range of power, and stages of an RF system need the power to fall in a narrow range to function optimally. A log amp computes an output signal that is some logarithmic function of its input. The output typically serves as feedback to control the gain of an amplification or attenuation stage. The latter stage can then maintain tight control over the level of the source signal as it rapidly fluctuates in real world conditions.
So there you have a logarithmic solution in a physical device, as opposed to some model abstraction.
They're not complaining about the basic nature of it being a log ratio to a reference value. Your use case would be pretty much the same if all the contradictory and confusing details of decibel were smoothed out.
Source: I'm pretty sure I understand the actual complaint better than you do, and you're being dismissive of something nobody said.
There are better ways, if you optimize for ease of understanding and not ease of use.
dB-based units make working between parts of a circuit, different devices etc. a breeze because adding 6dB in one part and removing it in another will result in no change in level. This is something the totally uninitated use every day in audio applications (typically that would be dBFS).
What other system do you propose? One where clipping audio has a level of 100% (0dBFS) and the dialog of a movie has 10% (-20dBFS) while the background atmo has 0.003% (-50dBFS) and the noise of the recorder has 0.000001% (-120dBFS)?
You can probably see why we use a log scale here..
dBFS is mostly fine. It's not used to communicate about loudness to anyone beyond experts. Loudness of machines is just specified as dB, and that gives manufacturers soooo much wiggle room, as to be useless.
