Bit of an oddball issue with a half-wave rectified supply

[somerandomengineer]

Hi All,

      Just ran into a bit of a weird one and I figured if I was going to find an answer anywhere it was probably going to be here

A new project I'm working on uses a rock-simple half-wave rectified supply, 6VAC in, through a single diode, through a 56R resistor and whacked across a 47uF cap.

For some reason my image didn't embed here *shrug* schematic attached below.

Weirdness ensued when I actually built it. Design uses a 1206 MLCC cap, and when I built it up I was getting all kinds of nasty ripple on the output, tacked a 100uF electro across the MLCC cap and it was fine, dropped that to a 47uF cap, still fine, removed the electro and strapped a second 47uF MLCC cap across the first, ripple was back.

Initially I thought my supplier must've cocked up and shipped the wrong value caps (it'd be REALLY nice if they'd, you know, LABEL SMD caps... never really understood why they don't), but testing them with my meter they came up as 50uF which is well within spec, removed the MLCC cap entirely and tacked a 47uF electro in and we're smooth.

Am I completely off my tree? Is there something fundamental I'm missing here? or do MLCC caps just behave strangely at mains frequencies?

De-confusion much appreciated.

Thanks,

- SRE

[exe]

Would you mind posting oscilloscope shots?

[capt bullshot]

Look up the capacitance over voltage characteristics of your particular MLCC.
It's quite common for them (especially small physical size and high capacitance) to drop most of their capacitance at rated voltage.
So you just might have an effective 4.7 ... 10uF cap instead of 47uF.

[george.b]

Look up the capacitance over voltage characteristics of your particular MLCC.
It's quite common for them (especially small physical size and high capacitance) to drop most of their capacitance at rated voltage.
So you just might have an effective 4.7 ... 10uF cap instead of 47uF.

This. You can't expect a multi-µF MLCC to have its nominal capacitance with that much DC bias.

[somerandomengineer]

Look up the capacitance over voltage characteristics of your particular MLCC.
It's quite common for them (especially small physical size and high capacitance) to drop most of their capacitance at rated voltage.
So you just might have an effective 4.7 ... 10uF cap instead of 47uF.

Ah, thanks, unfortunately I don't have data on them (probably a red flag there).

But that'd certainly account for it, smaller effective capacitance is consistent with the behaviour I was seeing and my meter is likely testing at a lower voltage.

Lesson learnt, use something other than MLCCs for bulk filter capacitance when operating close to ratings (meanwhile, since it's a personal project I'll just bodge an electro onto the pads and call it good)

[TimFox]

The voltage rating on MLCC capacitors is a safety rating for non-smoking, and not necessarily a recommended operating voltage.  As stated above, a legitimate data sheet will show the expected capacitance vs. dc bias voltage, which can be a surprisingly strong decrease, especially for “Y” and “Z” dielectrics.  These dielectrics have a very high dielectric “constant”, reducing the physical size of the component, but are “ferroelectric”, the analog of “ferromagnetic”, with hysteresis.

[SilverSolder]

The voltage rating on MLCC capacitors is a safety rating for non-smoking, and not necessarily a recommended operating voltage.  As stated above, a legitimate data sheet will show the expected capacitance vs. dc bias voltage, which can be a surprisingly strong decrease, especially for “Y” and “Z” dielectrics.  These dielectrics have a very high dielectric “constant”, reducing the physical size of the component, but are “ferroelectric”, the analog of “ferromagnetic”, with hysteresis.

Sounds perfect for high end audio applications!