The main point though is that we can call a lot of things a 'gain' and the understanding comes from the value if need be. In cases where we don't yet know the 'gain' or 'loss' if you will, we have no way of knowing YET anyway:
Vout=Vin*A

Is A a gain factor or a loss factor? We don't know yet and we don't have to know yet, symbolically.

In the above expression, 'A' is a gain factor. Period.

We don't know if it represents an actual increase or decrease in the magnitude of Vout, but that's fine.

Anyone that would describe A as a loss factor is incorrect and is injecting confusion into any discussion.

 

Just to add to this disjointed discussion....

In a past life, doing shipboard sonar and vibration analysis, we had two different types of instrumentation amplifiers to adjust levels into 1" 14 track tape recorders (fixed 1 volt input range) or ½" 7 track tape recorders (which have stepped input level controls).

One type of instrumentation amps had input GAIN adjustments. The other (oddball) amps had input ATTENUATORS. Both accomplished the same goal of adjusting signal levels to the range needed by the tape recorder, and turning the knob clockwise increased the signal level, but in one case the "level" was going from 0, 5, 10, 15....., and in the other it was going from -60, -55, - 50..... Yes, those numbers are decibels, as God intended when dealing with gain and attenuation.

With gain/attenuation in dB, handling the signal levels is simple. Instead of multiplying all numbers, you just add/subtract them:

(Sonar gain) + (amp gain) + (tape recorder gain) + (analyzer input gain) + (analyzer output gain) + (plotter gain) ---> plot in dB.

That's kind of an extreme case with that many options to keep track of, but with everything in terms of decibels, keeping track of it all was pretty simple.

 

Except -3dB is a higher amplitude then -5dB so it's gain, not loss. It is an amplification, not an attenuation. (People often leave off the baseline indication when using dB in a non relative way.)

Apples and oranges.

The -3dB is the ratio of output to input. It is NOT a signal level. It doesn't have "an amplitude".

What is the amplitude of a non-inverting opamp with a gain of 10? Is that amplitude greater or lesser than a 100 mV input signal? These are meaningless questions. It's like saying that a person is unhealthy if their weight is less than their height. Meaningless.

Describing the input signal as -3 dB is meaningless. But no matter what is meant, it is a signal level and not a transfer function.

If the input is -5 dBm and the system has an "attenuation of -3 dB" (and assuming they were using that phrase correctly), then the output signal would be -2 dBm.

But the output signal will have twice the power (into the same load) as the input signal regardless of what the input signal is (assuming it is within the amplifier's limits, of course). If it were +10 dBm, then the output would be +13 dBm. If it were -30 dBm, then the output would be -27 dBm. If the input signal were 0 dBm, then the output would be +3 dBm.

 

Just to add to this disjointed discussion....

In a past life, doing shipboard sonar and vibration analysis, we had two different types of instrumentation amplifiers to adjust levels into 1" 14 track tape recorders (fixed 1 volt input range) or ½" 7 track tape recorders (which have stepped input level controls).

One type of instrumentation amps had input GAIN adjustments. The other (oddball) amps had input ATTENUATORS. Both accomplished the same goal of adjusting signal levels to the range needed by the tape recorder, and turning the knob clockwise increased the signal level, but in one case the "level" was going from 0, 5, 10, 15....., and in the other it was going from -60, -55, - 50..... Yes, those numbers are decibels, as God intended when dealing with gain and attenuation.

With gain/attenuation in dB, handling the signal levels is simple. Instead of multiplying all numbers, you just add/subtract them:

(Sonar gain) + (amp gain) + (tape recorder gain) + (analyzer input gain) + (analyzer output gain) + (plotter gain) ---> plot in dB.

That's kind of an extreme case with that many options to keep track of, but with everything in terms of decibels, keeping track of it all was pretty simple.

The decibel was actually created (IIRC) specifically to facilitate this kind of analysis of a chain of processing blocks, specifically for telephone and telegraph systems. The number of elements in a typical RF system from input transducer on the transmitter to output transducer on the receiver tends to be significantly higher.

 

Describing the input signal as -3 dB is meaningless.

I do have to disagree somewhat with this (or at least expand on it). A decibel is a ratio. In terms of gain, it's the ratio of output/input.

But a level of say, 120dB can also be a scalar value if the reference level is given. In my case, 120 AdB re 1 micro-g is a typical machinery vibration level of 1 g. But the reference level must be given.

120AdB by itself is meaningless – the US Navy uses a reference level of 1 micro-g. The European standard is 10^-5 g, meaning the vibration level is 10× higher.

 

I do have to disagree somewhat with this (or at least expand on it). A decibel is a ratio. In terms of gain, it's the ratio of output/input.

But a level of say, 120dB can also be a scalar value if the reference level is given. In my case, 120 AdB re 1 micro-g is a typical machinery vibration level of 1 g. But the reference level must be given.

120AdB by itself is meaningless – the US Navy uses a reference level of 1 micro-g. The European standard is 10^-5 g, meaning the vibration level is 10× higher.

But when it is used as a scalar, then the unit is not dB, but some modification of it, such as your AdB. And, as you say, even it only has meaning if the reference is known. So 120 dB isn't a scalar value, 120 AdB is.

The dBm is understood to be power relative to 1 mW. When used for actual power, there's no ambiguity. But when used to describe voltage or current signals, it is necessary to know the load. The proper way would be to say something like "10 dBm at 50 Ω", but as long as the load is stated, can be inferred, or is not relevant to the discussion at the end of the day, this is usually left off. I think it is still the case that, lacking any other information, the assumption is that the load is 600 Ω.

Also, note that using dB (modified in some way, such as AdB) is not accepted by standards bodies (I can't say by 'any', but none that I'm aware of). In fact, dB is not recognized officially by SI, though it is recognized by IEC and ISO for power and root-power quantities (such as voltage and current). Though, even there, suffixes such as dBA and dBV are not recognized, probably because people that use them tend to not use the scaling factor of 20 instead of 10 because they don't grasp that dB is still a measure of power, not amplitude.

Of course, this doesn't prevent them from being bastardized and commonly used in various fields -- nothing can stop that.

 

Apples and oranges.

The -3dB is the ratio of output to input. It is NOT a signal level. It doesn't have "an amplitude".

What is the amplitude of a non-inverting opamp with a gain of 10? Is that amplitude greater or lesser than a 100 mV input signal? These are meaningless questions. It's like saying that a person is unhealthy if their weight is less than their height. Meaningless.

Describing the input signal as -3 dB is meaningless. But no matter what is meant, it is a signal level and not a transfer function.

If the input is -5 dBm and the system has an "attenuation of -3 dB" (and assuming they were using that phrase correctly), then the output signal would be -2 dBm.

But the output signal will have twice the power (into the same load) as the input signal regardless of what the input signal is (assuming it is within the amplifier's limits, of course). If it were +10 dBm, then the output would be +13 dBm. If it were -30 dBm, then the output would be -27 dBm. If the input signal were 0 dBm, then the output would be +3 dBm.

Hi again,

I am not sure what you are saying here.

If we had two levels in the same system, one -5dB and one -3dB, couldn't we say that the -3dB signal had a higher amplitude?

 

Hi again,

I am not sure what you are saying here.

If we had two levels in the same system, one -5dB and one -3dB, couldn't we say that the -3dB signal had a higher amplitude?

He is talking about two different things. The -5 dB is somehow the measure of the input signal (it needs to be some other unit, not dB -- perhaps dBm or dBW) while the -3 dB is the attenuation of the system. The first is giving the amplitude of a signal, the second is characterizing what a system does to that signal. Apples and oranges.

It is exactly like saying that a non-inverting opamp circuit with a gain of 100 is somehow a higher amplitude than an 2 V input signal. They are fundamentally different things that cannot be compared.

 

I agree with most of that. In some contexts though, we would just call it all a 'gain' even if it is a loss. That's because it's actually easier to think about it being a gain in so many cases.

In texts about converters for example, the resistive divider that divides the output by some factor like K, we would not say it's an attenuation we would say it's a gain. In this case it would be G=1/K, but we never point that out. If K=10 then we would just say G=0.1 and that's all. That is easy to understand too and works well with any following gains that may be encountered because we just ALWAYS multiply, we never have to divide.
.............
.............

The main point though is that we can call a lot of things a 'gain' and the understanding comes from the value if need be. In cases where we don't yet know the 'gain' or 'loss' if you will, we have no way of knowing YET anyway:
Vout=Vin*A

Is A a gain factor or a loss factor? We don't know yet and we don't have to know yet, symbolically.

* What about the term "antenna gain"? It does not match the above "definition" for "A"
* What about "space loss" or "insertion loss" ? Is it really better and more clear to speak about "space gain" and "insertion gain"?

 

The decibel was actually created (IIRC) specifically to facilitate this kind of analysis of a chain of processing blocks, specifically for telephone and telegraph systems. The number of elements in a typical RF system from input transducer on the transmitter to output transducer on the receiver tends to be significantly higher.

Yes, it was. And it was created with the idea of a reference level. That's why it is often (imprecisely) used without explicit reference to that level yet depending on it.

For example, gain over an isotopic radiator (dBi), or over 20 μPa (dB SPL), or over 1mW (dBm).

In the case of dBm, it is very common for numbers to be tossed around without the "m", and in the other examples, regardless of the explicit notation, this is, in fact, an amplitude. The 1mW reference for dBm makes the signal level:

\[ \mathrm{dBm} = 10 \log_{10}\!\left(\frac{P}{1\,\mathrm{mW}}\right) \]

So, in this domain:

• 0 dBm = 1 mW
• +10 dBm = 10 mW
• +20 dBm = 100 mW
• −10 dBm = 0.1 mW = 100 µW
• −30 dBm = 1 µW

dBm is impedance independent but obviously if we want voltages then it matters so an impedance is also implied based on the particular application (e.g.: 50Ω for RF, 600Ω for telephone circuits, &c.)

This is what my original response to @BobTPH was referring to:

And yet, if someone said “an attenuation of -3bB” I would never assume that meant a gain, would you?

In a very large number of cases where dB is used it is without explicit reference to the qualifying reference level—but it is still there. People use dB, in general, with very loose or absent understanding of what is actually means. So for someone to mean dBm of dBi or dB SPL—and say only "dB" is common.

Where you are unquestionably correct is where it is used purely as a ratio (as in our strict case here concerning gain)

\[ G_{\mathrm{dB}} = 20 \log_{10}\!\left(\frac{V_{\text{in}}}{V_{\text{out}}}\right) \]

A ratio of input vs. output where the input is the reference level. But I was referring specifically to vernacular use since ambiguity was the topic.

 

Yes, it was. And it was created with the idea of a reference level. That's why it is often (imprecisely) used without explicit reference to that level yet depending on it.

For example, gain over an isotopic radiator (dBi), or over 20 μPa (dB SPL), or over 1mW (dBm).

In the case of dBm, it is very common for numbers to be tossed around without the "m", and in the other examples, regardless of the explicit notation, this is, in fact, an amplitude. The 1mW reference for dBm makes the signal level:

\[ \mathrm{dBm} = 10 \log_{10}\!\left(\frac{P}{1\,\mathrm{mW}}\right) \]

So, in this domain:

• 0 dBm = 1 mW
• +10 dBm = 10 mW
• +20 dBm = 100 mW
• −10 dBm = 0.1 mW = 100 µW
• −30 dBm = 1 µW

dBm is impedance independent but obviously if we want voltages then it matters so an impedance is also implied based on the particular application (e.g.: 50Ω for RF, 600Ω for telephone circuits, &c.)

This is what my original response to @BobTPH was referring to:



In a very large number of cases where dB is used it is without explicit reference to the qualifying reference level—but it is still there. People use dB, in general, with very loose or absent understanding of what is actually means. So for someone to mean dBm of dBi or dB SPL—and say only "dB" is common.

Where you are unquestionably correct is where it is used purely as a ratio (as in our strict case here concerning gain)

\[ G_{\mathrm{dB}} = 20 \log_{10}\!\left(\frac{V_{\text{in}}}{V_{\text{out}}}\right) \]

A ratio of input vs. output where the input is the reference level. But I was referring specifically to vernacular use since ambiguity was the topic.

I'm still at a loss for your claim, so perhaps you could explain it in some detail. For convenience, here's the exchanges I'm referring to:

And yet, if someone said “an attenuation of -3bB” I would never assume that meant a gain, would you?

Only if the input signal was -5dB or the like.

The size of the input signal should have nothing to do with the amount of attenuation, since that is a characteristic of the system, not the signal.

I also wouldn't know how to interpret what an input signal of -5 dB even is. There's no reference. I would expect it to be expressed relative to some reference power, such as milliwatts (dBm) or watts (dBW).

Except -3dB is a higher amplitude then -5dB so it's gain, not loss. It is an amplification, not an attenuation. (People often leave off the baseline indication when using dB in a non relative way.)

You say that you would only assume that an attenuation of -3 dB meant a gain if the input signal was -5 dB. What would you assume if the input signal was +20 dB, or -30 dB.

How does the amplitude of the input signal have any impact on whether the processing block provides gain or attenuation?

And how does it make sense to directly compare a gain (or attentuation) to a signal level?

And what does it mean to talk about the amplitude of an attenuation.

If the input signal is -5 dBx (for whatever reference level x implies), then after it goes through a system with an attenuation of -3 dB, the resulting signal is -2 dBx.

The system has gain because the output signal, at -2 dBx, is larger than the input signal, at -5 dBx.

But the same would be true for ANY input signal.

If the input signal is +5 dBx, then after it goes through a system with an attenuation of -3 dB, the resulting signal is +8 dBx.

The system has gain because the output signal, at +8 dBx, is larger than the input signal, at +5 dBx.

The output of this system will always be 3 dB more than the input, so the system has gain, regardless of the amplitude of the input signal.

 

* What about the term "antenna gain"? It does not match the above "definition" for "A"
* What about "space loss" or "insertion loss" ? Is it really better and more clear to speak about "space gain" and "insertion gain"?

Hi,

Well this is not so much about speaking as it is writing. When we talk we often use different ideas and ways to express a given situation, but when we write it down it always makes sense to use the 'A' gain idea even if less than 1 (or negative).

Antenna gain is an interesting one though. We just imagine that we multiply by some number, but it will always be some number which we can say is 'A' although it would probably only be used comparatively. We could look at various scenarios to see how this flies I guess. This might be a true exception. We should still be able to list them though I think:
Ant1: A=3dB
Ant2: A=6dB
Ant3: A=8dB
I am forcing the use of 'A' here rather than 'G' just to be consistent.

 

Hi,

Well this is not so much about speaking as it is writing. When we talk we often use different ideas and ways to express a given situation, but when we write it down it always makes sense to use the 'A' gain idea even if less than 1 (or negative).

Antenna gain is an interesting one though.
..........
...........
Ant1: A=3dB
Ant2: A=6dB
Ant3: A=8dB
I am forcing the use of 'A' here rather than 'G' just to be consistent.

Just a small correction:
Antenna gain must always be given in dBi (dB with reference to an isotropic antenna)

 

Just a small correction:
Antenna gain must always be given in dBi (dB with reference to an isotropic antenna)

Hi,

That's interesting, because back when I was into antenna design and purchasing almost everything was in dBd which they wrote as simply dB.
That must have changed over the years as technology got better or something.
I think there is a simple conversion from dBi to dBd or the other way around.

 

I'm still at a loss for your claim, so perhaps you could explain it in some detail. For convenience, here's the exchanges I'm referring to:









You say that you would only assume that an attenuation of -3 dB meant a gain if the input signal was -5 dB. What would you assume if the input signal was +20 dB, or -30 dB.

How does the amplitude of the input signal have any impact on whether the processing block provides gain or attenuation?

And how does it make sense to directly compare a gain (or attentuation) to a signal level?

And what does it mean to talk about the amplitude of an attenuation.

If the input signal is -5 dBx (for whatever reference level x implies), then after it goes through a system with an attenuation of -3 dB, the resulting signal is -2 dBx.

The system has gain because the output signal, at -2 dBx, is larger than the input signal, at -5 dBx.

But the same would be true for ANY input signal.

If the input signal is +5 dBx, then after it goes through a system with an attenuation of -3 dB, the resulting signal is +8 dBx.

The system has gain because the output signal, at +8 dBx, is larger than the input signal, at +5 dBx.

The output of this system will always be 3 dB more than the input, so the system has gain, regardless of the amplitude of the input signal.

I will return to this when I get a chance. Unfortunately we are talking past each other but I need to work out how to clarify my point. None of what you are saying is wrong but it misses my intended point.

 

I will return to this when I get a chance. Unfortunately we are talking past each other but I need to work out how to clarify my point. None of what you are saying is wrong but it misses my intended point.

Look forward to seeing your clarification so that we can clear up the confusion.

 

I think there is a simple conversion from dBi to dBd or the other way around.

Yes, we have: dBi = dBd + 2,15

 

Yes, we have: dBi = dBd + 2,15

Hi again,

Yes thanks.

What else I find interesting these days is a lot of TV antennas are rated in "miles". I think that's kind of funny in a way.
They don't even give any kind of dB rating.

I liked it back when they used dB alone which meant dBd because a dipole was fairly simple to build, so if I bought an antenna that has a gain that was 2x what a dipole would be, I would know right away if the purchase price seemed worth paying or just build one myself.

I did end up building my own during the times when CB was very popular. It was a fairly large dipole I mounted on the roof of the house (about 20 feet or so up). It was made mostly out of 1/2 inch copper tubing with a tough fiberglass center mount. It was a full size dipole.
I thought that was big, but I was amazed though when one of my brother in laws showed me pictures of his short-wave antenna. It had a span of 40 feet horizontally and was mounted up pretty high too. It was literally a monster antenna.