gdjacobs wrote:

It's the same for tuned LC band stop filters, but that wasn't what they were using. More like a second order RLC filter, but split into two sections.

I was looking through my cache of old ATX schematics (trying to find typical values for the Pi filter inductors) and it suddenly occurred to me what this RLC configuration probably is.
Many PSUs have load resistors somewhere near the outputs. They are basically small "dummy loads".
They provide a minimum load on the given rail to ensure it regulates properly if the external load is too small.
Some PSUs without them will not stay running if they don't have an external load.
I believe their use is fading away because they reduce a PSU's overall efficiency.
.
Some schematics show them on every rail and others only on some rails.
Some show them before the series inductor and others after.
.
I also came across a few that have the load resistor but there is no filter inductor at all.
They count on just the toroidal transformer and a cap being enough filtering.
.

PCBONEZ wrote:

... They count on just the toroidal transformer and a cap being enough filtering ...

And thus they achieve "spectacular" results in the ripple "suppression" chapter. 🤣

- Gutless Wonder Exhibit 1 - No PI coils, just one single crappy cap per main rail:

. . .
http://www.clubedohardware.com.br/artigos/tes … x-de-230-w/2876

^ 148.5mV of peak to peak ripple on +12V rail (ATX spec puts the limit at 120mV). 😵

But we can go even cheaper (copper is expensive after all) ...

- Gutless Wonder Exhibit 2 - As above but with just one single toroid inductor for all three main rails (no separate inductor for +3.3V):

. . .
http://www.clubedohardware.com.br/artigos/tes … t-500-p42s/2790

^ 155.4mV of peak to peak ripple on +12V rail. 😵 😵

^ And those two ticking bombs were brand new when tested. After some time the crappy caps will go south, and ripple off the scale. 😵 😵 😵

Some cheap PSUs manage in principle to barely keep ripple inside ATX spec without PI coil and just a couple of crappy caps in parallel per main rail, like this one, but as time passes by and crappy caps go south they'll quickly go off spec again.

As a rule, for classic group-regulated PC SMPSs (like vintage and cheap modern) I'd stay away from anything that doesn't come with two decently sized & reasonable quality caps flanking one PI coil per each main and +5VSB rails (-5 and -12 rails can usually do OK with just one or two smaller caps, though a PI coil wouldn't harm).

http://www.hardwareinsights.com/wp/how-a-swit … r-supply-works/

Considering the fact that they problem have no mains filtering either, I'd hate to see what noise is flowing back to the power socket.

"Input filtering??? That'd cost a whole bunch of pennies!!!" 🤣

Well... they are trying to improve airflow!

One of the other old farts over at BCN came across some that had dummy weights glued inside - to make them seem heavier.
.

I've heard of dummy passive PFC coils made of iron, lead, or cement!

Yep fake pPFC coils have been found inside some gutless wonders.

And I thought I had seen it all...

Like the "Russian doll" capacitors we all have to watch out for.

Another example. This pic is a classic. Been around the web a while. IIRC it was an eBay purchase.

I love how the real cap inside the fake is a knock-off anyway. Rulycon, indeed!

Well, the Mod that moved this here decided the title should be "Capacitor Replacement" so I suppose we should try to get back to that.
Maybe tomorrow. It's close to the end of my day today.

I put this in another thread but it should be in this topic too.

PCBONEZ wrote:

One big deal thing to check on ESR meters is that they use a 100kHz test signal. Might be square wave, saw tooth, sine wave - th […]
Show full quote

One big deal thing to check on ESR meters is that they use a 100kHz test signal.
Might be square wave, saw tooth, sine wave - that doesn't matter too much but the frequency does.
.
The datasheets for low ESR caps give the ESR at 100kHz.
If the test signal is not 100kHz you have to do math for each cap to know what it is at 100kHz before you know if it's in spec.
(There is a conversion chart in most datasheets for this purpose.)
Huge PITA.

Most combined LCR-ESR meters don't have a 100kHz test signal.
Many cheap ESR (only) meters don't either.
.

Link below is a thread regarding needed equipment. (So we don't need to duplicate all of it here.)
Recapping gear!

Found a good article describing non-ideal capacitor behavior. I've included the graphic of the equivalent component model to help those who wish to understand.

https://eewiki.net/display/Motley/Non-Ideal+P … s+of+Capacitors