4
\$\begingroup\$

Why is the voltage amplification factor of an amplifier or a filter generally expressed by its gain in decibels, and why is the bandwidth of this amplifier defined at −3 dB?

\$\endgroup\$
1

4 Answers 4

15
\$\begingroup\$

Why is the voltage amplification factor of an amplifier or a filter generally expressed by its gain in decibels

The decibel is a very convenient way of expressing gain especially in filters. For instance a simple low-pass filter will have unity gain in the pass band and, above the pass-band, the gain will diminish at 6 dB per octave. So, if the filter has a cut-off (-3 dB point) at 1 kHz, we can say that it will be nominally 6 dB down at 2 kHz, 12 dB down at 4 kHz etc..

This is exactly the same as saying that the output voltage (above the cut-off) falls at the same rate that the frequency increases hence, if frequency doubles then voltage halves or, if frequency rises ten times then voltage drops then times. This leads us to also say that a 1st order low-pass filter has an output "roll-off" of 20 dB per decade. It's exactly the same as saying 6 dB per octave: -

enter image description here

Image above from Normalised first-order low-pass filters supplied by the open university. Hence, the use of decibels makes the roll-off above the cut-off frequency into a linear line at 20 dB/decade. You can see the equivalent voltage gain (as a ratio) on the right y-axis.

Of course you could make an argument based around using a linear frequency x-axis and a linear amplitude y-axis and, this would naturally avoid using decibels. However, think about the limitations of using a linear frequency axis; the graph above covers a frequency range of a million to one and a linear graphs would be so much more limited in its range that it would be useless in comparison.

and why is the bandwidth of this amplifier defined at −3 dB?

When the output of a low-pass filter (for example) drops to 3 db below is nominal pass-band value, the output amplitude power amplitude has halved. This should be obvious to anyone understanding the decibel but, there's another observation. For a simple RC low-pass filter, when the magnitude of the output is 3 dB down, the value of resistance equals the magnitude of the capacitive reactance.

Of course, -3 dB is a close approximation to the half-power-point. The true half-power-point is \$10\log_{10}(0.5)\$ = -3.0103 dB but, we call it -3 dB for reasons of convenience.

If we looked at the voltage transfer function of a simple RC filter we would get this: -

$$H(j\omega) = \dfrac{1}{1+j\omega RC}$$

And we know that the cut-off frequency (\$\omega\$) is \$\frac{1}{RC}\$ so, the above formula becomes: -

$$H(j\omega) = \dfrac{1}{1+j}$$

And that implies the voltage gain has dropped by \$\sqrt2\$ at the cut-off point. This also equals -3.0103 dB using this formula: \$20\log_{10}(\frac{1}{\sqrt2})\$.

\$\endgroup\$
5
\$\begingroup\$

The bandwidth of an amplifier is defined at the −3 dB point because at this frequency the output signal power drops to half (i.e., 50%) of its maximum value. Since power is proportional to the square of voltage, halving the power corresponds to a voltage equal to approximately 70.7% of its peak value, which in logarithmic terms equals −3 dB.

The voltage gain of an amplifier or filter is commonly expressed in decibels (dB) because this logarithmic unit conveniently represents a wide range of gain values (often very large) on a more manageable scale. Additionally, using decibels simplifies calculations in multi-stage systems, as cascaded gains are added rather than multiplied.

\$\endgroup\$
2
\$\begingroup\$

It's quicker to subtract a measured logarithm than to divide by a measured voltage or current. In the days of analog meters, most AC volts ranges were scaled in dBm (decibel measure of signal power in milliwatts into a 600 ohm load). Probe the input and output of an amplifier, and note the difference on those logarithmic scales, and you knew the power gain of the amplifier. Digital voltmeters aren't as direct, in some ways. With a notebook and calculator, and some time, you can develop the same info.

Amplification roll-off at high or low frequency is usually characterized by the asymptote line on a log(power) vs. log(frequency) plot. Those asymptotes aren't straight lines unless you have a logarithm scale, thus: dB is the natural label appropriate to the power, just as octave is appropriate for the frequency.

\$\endgroup\$
0
\$\begingroup\$

Not all filters are specified at -3dB, I ran across a very old spec once that used -6dB, I think it was from the 1950s or earlier. Everybody using the same conventions helps everyone compare between different vendors, and -3dB is close to half power.

Also, if you are looking at amplifiers 1dB and 3dB compression points are common specifications, and have nothing to do with the filter bandwidth. Here, 1 dB makes more sense for the specification and the devices and 3dB is easier to translate to linear math.

\$\endgroup\$
2
  • \$\begingroup\$ Everybody using the same conventions helps everyone compare between different vendors...except for when the convention is stupid, like consumer and "prosumer" speaker manufacturers using -10dB to specify (inflate) their (not so) "usable" bandwidth, to deceive people who don't know any better. Similar with their power ratings, and the hopelessly non-representative tests that they devise to technically be repeatable, but they're really designed just to give a big number. \$\endgroup\$ Commented Oct 26 at 17:23
  • \$\begingroup\$ Serious engineering brands that don't have consumer divisions at all, are actually honest...and have head-scratchingly low ratings compared to their actual performance, for those that are used to the consumer inflation. Bottom line: if you're actually serious about designing a system, instead of emoting one, you need to understand and evaluate the tests too, not just the "results". \$\endgroup\$ Commented Oct 26 at 17:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.