0
\$\begingroup\$

I have been experimenting with this circuit:

schematic

simulate this circuit – Schematic created using CircuitLab

Vin has a lower frequency ripple of 2V and a high frequency (switching?) noise of 6.5V. At V1, lower frequency ripple is about 200mVpp and the high frequency noise is 4.2Vpp. On Vout, the high frequency noise is the same, 4.2Vpp. But if I remove C4 then the high frequency noise at Vout decreases to 1.3Vpp.

I don't understand why? How it that possible?


Low frequency is ~100Hz

high frequency is 25Khz

this is when I remove C4:

and this is when I add it:

UPDATE: after replacing the power supply of the buck converter with a battery:

enter image description here

\$\endgroup\$
1
  • 3
    \$\begingroup\$ Resonance is the clear answer. Plot a bode diagram and all will be revealed. \$\endgroup\$
    – Andy aka
    Dec 8, 2019 at 13:03

1 Answer 1

1
\$\begingroup\$

Welcome to the sad realities of doing real electronics work.

There's two possibilities (one of which I missed at first). Frankly, I'm not sure which is the problem.

The first problem is that your filter impedance is roughly \$\sqrt{\mathrm{\frac{220\mu H}{100\mu F}}} \simeq 1.5\Omega\$ but you're terminating the filter with 100 ohms and 6.8k ohms. So even with ideal components, the filter response will be very underdamped. This is probably not the problem if you have ideal components (edit: it's not the problem -- the resonances are there, and they're strong, but they're between 1kHz and 2kHz), but it would show up if you analyzed the filter response or simulated it.

The second problem is that with those component values, you may be having issues with parasitic effects: when you design with any cap, you think you're getting this a nice pure capacitor, like you write down on the schematic. But you're really getting this infinitely complicated thing that may act like a cap in series with a resistor, or in parallel with a resistor, or in series with an inductor, or all of the above, or -- as you push into higher frequencies or lower impedances -- ever more complicated stuff.

Aluminum electrolytic capacitors are made by taking a couple of strips of aluminum foil, layering them with paper soaked in salty goo* and winding them into a a cylinder. When you use the cap outside of its intended frequency range, the fact that there's a coil of conductors involved, it starts to look inductive.

Similarly, inductors act like inductors in parallel with capacitors (because the wires are spaced closely together, and have winding-to-winding capacitance).

That's a pretty big cap to be using in the presence of a 100kHz signal; I suspect that if the filter as designed doesn't have a resonance at 100kHz, then the parasitic elements of your various parts (coils + caps) is causing a resonance.

Aluminum capacitor innards, Elcap, Licensed under CC0 1.0, retrieved from Wikimedia

\$\endgroup\$
7
  • \$\begingroup\$ I'm using a breadboard, so where are parasitic capacitanes everywhere. I'm not sure how to test this otherwise. I'm a beginner in electronics, and I'm trying out various circuits to learn from them. I cannot design and order a custom PCB for my test circuits, that would be too slow and too expensive. So I'm learning by putting these together on breadboards - maybe I should not care about that noise at all? But it was very suprising and I did not understand how and why it appeared. \$\endgroup\$
    – nagylzs
    Dec 9, 2019 at 11:44
  • 1
    \$\begingroup\$ Do a web search on "dead bug construction". That's probably the best method of construction you can use with through-hole parts without a manufactured PCB. I suspect that even with a properly laid-out PCB you'd still have problems with the parasitics of the parts themselves, and certainly with the impedance mismatch. You should simulate that circuit with a square wave or step input, and with an AC sweep -- I expect that the former will ring like a bell, and the latter will show multiple and pronounced resonances. \$\endgroup\$
    – TimWescott
    Dec 9, 2019 at 16:10
  • \$\begingroup\$ I have added a new image that shows the signal when the power supply of the buck converter is replaced with a battery. (The buck converter was powered with a 5V SMPS). The buck converter itself is the same. So most of the noise was coming from the power supply. There is still noise, about 100mV, probably because of resonance. I need to learn more about LC circuits and how to simulate them. Interesting part is that the buck converter (mc34063) changes its base frequency when the load is changes. I don't see how I could terminate any LC filter correctly when the frequency is not constant. \$\endgroup\$
    – nagylzs
    Dec 10, 2019 at 6:36
  • \$\begingroup\$ You can simulate that circuit that you drew. You'll have to change the name of the source to "Vin" -- the simulator doesn't appear to handle spaces in the name well. Do an AC sweep, from 10Hz to 1MHz, with 500 points per decade or so (because the resonances are narrow). You'll see that they're not in spots to cause this problem, though -- so you have issues with parasitics. You need to choose low ESR caps, and think about your inductors more. \$\endgroup\$
    – TimWescott
    Dec 10, 2019 at 15:58
  • 1
    \$\begingroup\$ You can make quite satisfactory switch mode PSs on a 2 layer board, which are not very to make yourself. This is especially true with a power supply due to fact the traces are much larger and easier to etch. This makes it cheap, fast and fairly easy to make prototypes or one offs. About the largest board I can make is 6"x6", which does not stop me from tinker toying larger systems. \$\endgroup\$
    – GB - AE7OO
    Mar 13, 2020 at 14:31

Your Answer

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

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