This is a simulated AC analysis of a small noise generating circuit.

LTSpice AC sweep output

and comes from this circuit

LTSpice schematic

I have one of those fundamental but embarrassing beginners questions. What does the solid curve refer to? My circuit is powered at 30V, and the LTSpice AC sweep analysis was set to an AC current amplitude of 80 nA. One of my circuit nodes is called output, and there is an output signal there of magnitude 1 V p-p. The curve is flat at approximately -7 dB, but that is a relative measure. Relative to what please?

  • \$\begingroup\$ schematic please? \$\endgroup\$ – berto Mar 23 '16 at 0:39
  • \$\begingroup\$ @berto Does it depend on the schematic? I thought that it depended on some sort of input to the circuit, whatever it may be... \$\endgroup\$ – Paul Uszak Mar 23 '16 at 1:03
  • \$\begingroup\$ You can change the scale to linear or logarithmic in V or A units if you want, just right-click on the y-axis (IIRC). Then you'll know what reference level the program is assuming. \$\endgroup\$ – The Photon Mar 23 '16 at 1:43
  • \$\begingroup\$ is the C2 value 0.1 F right? \$\endgroup\$ – berto Mar 23 '16 at 2:22
  • \$\begingroup\$ @berto Well you can see the simulation curve for response. To be honest, as far as simulation goes it's value doesn't make much difference as all I want is up to 20 kHz. I tried it with a 1 uF and it just shifted the left hand side of the curve a few 10s of Hz. You have an opinion? \$\endgroup\$ – Paul Uszak Mar 23 '16 at 9:48

Craig K gave the definition of the dB, so I won't need to go into it. I'd like to point out though, that the reference LTSpice specifically uses is indeed 1 SI unit of whatever you're measuring (as The Photon was assuming in the comments) so for voltages 0dB is 1V, currents it is 1A and so on. So the reference is arbitrary, and in my opinion a bit useless.

You can work around that though, by plotting specifically the ratio of the output to the input, in which case the ratio is dimensionless and unity gain gives 0dB.

For a bit more detail, see my answer here

| improve this answer | |
  • \$\begingroup\$ Yes, I've confirmed that it's relative to 1 volt, not the input or power supply. \$\endgroup\$ – Paul Uszak Mar 28 '16 at 3:03

First lets honor the gentleman who gives the capital letter "B" to dB. Alexander Graham Bell.

Sometimes it is useful to record power values in a logarithmic scale rather than a linear scale. Example is 1 million to 1 ratio may be more useful to record in a logarithmic scale as 60dB instead of 1 000 000:1

Ratios of equivalent power units are expressed in dB. So just like saying something is 1/100 you could say -20dB. Or if something is 100/1 you could say +20dB.

Where some confusion arises is when the denominator is relative to some fixed value of power. Example is RF power is usually relative to 1mW of power. So 0dBm is 1mW, 10dBm is 10mW, 20 dBm is 100mW and 30 dBm is 1.0W of power. Makes sense? Honestly it is useful to engineers but very confusing to laymen.

Another source of confusion with the dB notation is when dealing with Voltage. Remember I stated that ratios of equivalent power units are expressed in dB. Power is proportional to the Voltage squared. So if you have a -20dB attenuator it is a 10:1 divider of voltage while it is also a 100:1 divider of power. Again I'll say it is useful to engineers but very confusing to laymen.

| improve this answer | |
  • 2
    \$\begingroup\$ RE your 3rd paragraph, dB refer to power ratios (not ratios of any two quantities with the same units). Sometimes we sloppily use them to refer to voltage ratios, in which case the value doubles (100:1 voltage ratio is called 40 dB), because we're thinking of a case where the two voltages are applied to the same impedance so a 100:1 voltage ratio implies a 10,000:1 power ratio. \$\endgroup\$ – The Photon Mar 23 '16 at 1:38
  • \$\begingroup\$ @ThePhoton Thanks. I made an edit. Let me know what you think of what I've written. \$\endgroup\$ – Craig K Mar 23 '16 at 1:52
  • \$\begingroup\$ You way you want to honor AGB with the upper case B in "dB" in the first paragraph, then write it lower case in the second! \$\endgroup\$ – Olin Lathrop Mar 24 '16 at 11:21
  • 1
    \$\begingroup\$ @OlinLathrop Thanks for spotting my goof. I edited it. \$\endgroup\$ – Craig K Mar 24 '16 at 15:02

It is relative to the amplitude of the signal (voltage or current) supply that makes the AC sweep.

A convenient way to make an AC analysis is to define clearly which supply makes the sweep.

You can double click on the voltage/current source, go to "Advanced", and define it the "Small signal AC analysis: AC amplitude" property. You can double check by placing a probe in this supply to make sure the amplitude is always the same for all frequencies.

| improve this answer | |
  • \$\begingroup\$ That's exactly what I did. So how do I get a -7 dB voltage gain relative to an 80 nA AC current sweep? The piccy clearly says V(output). \$\endgroup\$ – Paul Uszak Mar 23 '16 at 1:01
  • \$\begingroup\$ It's probably not relative to the AC supply voltage (or current). It's probably relative to 1 V or 1 A. Consider you could have a circuit with two ac sources with different amplitudes, and LTSpice will still happily plot the outputs in dB. If you want to get the output relative to the input, you need to make the input 1 V or 1 A in amplitude. Even if that's an unrealistic level for your circuit, it won't matter because in SPICE an AC analysis is an idealized analysis that assumes the circuit behaves perfectly linearly. \$\endgroup\$ – The Photon Mar 23 '16 at 1:41
  • 1
    \$\begingroup\$ @PaulUszak if you put a current probe in series with I1 source it should be 0 dB for all frequencies, that is the double check. Also try what ThePhoton suggests in you question, change to magnitude instead of dB. If ThePhoton is right above, and the reference is 1 V, then -7 dB corresponds to 10^(-7/20)=0.447 V (because X in dB is 20*log10(-7) ) \$\endgroup\$ – berto Mar 23 '16 at 2:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.