7
\$\begingroup\$

It is common to recommend the placement of capacitors between the input and output sides of an isolated DC-DC converter, such as in this example layout:

enter image description here

I understand that the purpose of this is to reduce EMI. However, since capacitors allow the passage of AC, doesn't this defeat the purpose of isolation on the AC side at the very least? Isolated DC-DC converters are often rated for both DC and AC isolation; in fact, some advertise their AC isolation with greater emphasis than their DC isolation.

Are these example circuits predicated on the assumption that the user doesn't care about AC isolation, or is there something fundamentally wrong with my reasoning?

\$\endgroup\$
10
  • 4
    \$\begingroup\$ Legally, you are still isolated enough for almost all purposed except for medical. You will feel the 50/60 Hz tingling in your fingers when you touch it and if your DMM has high input impedance, you will measure AC voltage on the secondary. Still, it's very cheap and effective for EMI. \$\endgroup\$
    – winny
    Commented Mar 10, 2023 at 16:47
  • \$\begingroup\$ Because you speak of a DC-DC converter everything that there is AC (no matter if on input or output side) is considered as unwanted. And in most of the cases is just noise, ripples etc. Those CY1-4 are the only chance to provide a return path for noise of any source. What type of DC-converter did you copy this image from? \$\endgroup\$
    – datenheim
    Commented Mar 10, 2023 at 16:54
  • 4
    \$\begingroup\$ AC isolation relates to 50 or 60 Hz: at such frequency 1 nF translates to a resistance of ~ 3 MOhm - not that much power that wil couple over here. For noise (MHz to GHz) that is few Ohms to mOhms - quite another story. In a pure DC world, those capacitors do not exist, or just to nothing worse :) \$\endgroup\$
    – datenheim
    Commented Mar 10, 2023 at 17:22
  • 2
    \$\begingroup\$ @JeremyBoden Y1 is a safety rating - they have certain specifications and "never" fail short-circuit \$\endgroup\$ Commented Mar 11, 2023 at 1:27
  • 1
    \$\begingroup\$ Do note that there's no such thing as being "totally" isolated in AC. As the common (oscilloscope) experiment youtube.com/watch?v=ya3BpcJLoLg that reads AC voltage from yourself in a room with some AC power around (cables in walls) would tell you. There's always parasitic capacitance, even without a capacitor as a specific component. The questions, like with the poison, is one of dose/how much. \$\endgroup\$ Commented Mar 11, 2023 at 6:28

2 Answers 2

8
\$\begingroup\$

However, since capacitors allow the passage of AC, doesn't this defeat the purpose of isolation on the AC side at the very least?

It does undermine the AC isolation. Whether or not it defeats the AC isolation depends on:

  • Max leakage current allowed for the isolation. The values often comes from a standard such as IEC 60601, IEC 60950, IEC 62368.
  • The value of the capacitor.
  • Max mains AC line voltage. 110 VAC, 220 VAC, and such.
  • Mains AC line frequency. Usually 50 Hz or 60 Hz.

Example.

  • IEC 60601 max AC patient leakage current for a BF applied part under normal conditions is 100 µA AC. (Table 3 in the standard.)
  • 470 pF EMI capacitor
  • International voltages (240 VAC plus 10%, 60 Hz)

We get 47 µA leakage current.

Larger capacitors may be better for EMC, but they may introduce too much leakage current, thus defeat the isolation. This is a common tradeoff when designing for electric safety, EMC, ESD.

Speaking of electric safety, notice that the capacitors across the isolation have designators CY, instead of more common C. The capacitors across isolation need to be Class-Y rated.

I understand that the purpose of this is to reduce EMI.

The purpose of these capacitors is to provide a return path for high frequency EMI currents. High frequency EMI in case of a small DC-DC switcher is somewhere on the order of 10 MHz to 200 MHz, depending on the parasitics. A 1 nF capacitor at 10 MHz has 1.6 Ω impedance. That's the low impedance path.

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

since capacitors allow the passage of AC, doesn't this defeat the purpose of isolation on the AC side at the very least

It degrades performance in some applications so, if you have an application that needs a galvanically isolated supply for a high-side (and high-voltage) MOSFET or IGBT then, you choose a converter that doesn't use external capacitors and has very low barrier capacitance (<10 pF).

But, for some folk, maybe 10 pF would be too much and, you have to use different means.

is there something fundamentally wrong with my reasoning?

No.

\$\endgroup\$

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.