To experiment(out of curiosity) the effect of a low pass filter for an SMPS switching noise I wanted to try a filter where an LC cascaded by a linear regulator as shown in below diagram, I obtained the AC analysis in LTspice.

enter image description here

I don't know anything about the output impedance of the SMPS so I chose is as 1 Ohm. Then I swept the input between 0.1Hz to 1 Meg Hz.

But why in the Bode plot Vout starts from -180dB but not 0dB?


Following some comments I superimposed the noise on a 24V DC as shown below:

enter image description here

The AC coupling caps C4 and C3 are there to view the noise component at the input and output.

Here is when 1V 100 kHz square wave noise(with 1p rising and falling edges) is superimposed on the 24V DC:

enter image description here

Here is when 1V 100 kHz sine wave noise is superimposed on the 24V DC:

enter image description here

Below is when 1V 0.1 Hz sine wave noise is superimposed on the 24V DC:

enter image description here

And finally here is the AC response when superimposed on the 24V DC:

enter image description here

It seems in the AC response at 0.1 Hz the y axis shows around -70dB. This corresponds to a voltage ratio of 3162 and for 1V the output would be expected as 320uV which almost matches to the 0.1 Hz transient response noise amplitude.

Are these results expected in practice? If so it seems to me the linear regulator works as a very very strong low pass filter.

  • \$\begingroup\$ 0dB is unity gain; -180dB is a gain of \$\small 10^{-9}\$. Why is -180dB surprising? The idea of a regulator is to give pure direct voltage, or as close as possible. This one is pretty good at it. \$\endgroup\$ – Chu Jan 18 '19 at 14:06
  • \$\begingroup\$ I interpret the dB vertical axis of the plot mathematically as 20*log10(Vout/Vin) verus the frequency axis. If not what is it? \$\endgroup\$ – user1245 Jan 18 '19 at 14:21
  • \$\begingroup\$ At 0 or 0.1Hz the regulator should not attenuate anything but it does. Very confusing to me. \$\endgroup\$ – user1245 Jan 18 '19 at 14:27
  • \$\begingroup\$ Yes that's correct for your first comment;volt/volt = gain . How have you checked it at 0 Hz? that frequency isn't on the Bode plot, also, the supply voltage must be greater than the intended output regulated voltage otherwise the regulator can't work. \$\endgroup\$ – Chu Jan 18 '19 at 14:37
  • \$\begingroup\$ @Chu I will try to elaborate on my confusion very soon hope you can help me with this. Thanks \$\endgroup\$ – user1245 Jan 18 '19 at 15:30

I believe that the reference level for LTspice ac plots is that 0 dB = 1 V. So, it looks like your ac signal is very, very small. Keep in mind that this is a simulation, not real life, so there are round-off errors that can accumulate and look like very small noise signals.

EDIT: Your simulation provides an input voltage of 1V for the regulator. An LM340T-12 is not going to function properly with that small of an input voltage, you need at least 14V to make this device work. I think your simulation results are probably bogus.

| improve this answer | |
  • \$\begingroup\$ At 0Hz OR 0.1Hz the regulator will pass 1V DC input as 1V DC output which is 0 dB, but in the plot at 0.1 Hz the output is -180dB. Can that be because I need to bias this regulator? \$\endgroup\$ – user1245 Jan 18 '19 at 13:56
  • \$\begingroup\$ What makes you think that there is an ac component of 1V at 0.1 Hz? You are saying that over the course of ten seconds the output voltage changes from 0 to 2V and back again. When you perform an ac simulation there is no dc component in the results. \$\endgroup\$ – Elliot Alderson Jan 18 '19 at 15:15
  • \$\begingroup\$ What I want to see is the "regulator filter"'s effect to the SMPS switching noise superimposed on the DC output of the SMPS. Either it is really filtering 180dB at 0.1Hz or my sim is wrong. \$\endgroup\$ – user1245 Jan 18 '19 at 15:46
  • 1
    \$\begingroup\$ Then you want a transient simulation, not an ac frequency sweep. \$\endgroup\$ – Elliot Alderson Jan 18 '19 at 15:50
  • 2
    \$\begingroup\$ The input source must be dc and ac: the dc fixes the operating point, for instance 15 V and then you add a 1-V ac modulation. Please make sure the operating point is correct, e.g. \$V_{out}\$ delivers the correct value before considering the results. I know LTspice can reflect the OP back to the schematic. It must always be done before considering any ac results. \$\endgroup\$ – Verbal Kint Jan 18 '19 at 19:20

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.