Switch mode power supply failure mechanism?

[N_Cook]

I have a use for one of these all new unused in 1990s, not stored in a
shed or garage over the years. A few dozen ,made in 1992 nice compact
70W Electronic Halogen Transformers (so called including labels as
primary and secondary on the external label). Outputs the unrectified HF
to drive the 12V lamps. They used to work 15 years or so ago. Now I've
tried 3 and all the same , should be 12V HF ac and if attached to high
speed diode and a cap and no load then only about 10 to 12V dc. No way
will they power even a 20 W lamp (supplies are rated 20W to 70W) tried
40W in case load dependent. Scoping, with no load and a complex
oscillation signal. Then as little as 27R load very poor mark space
ratio of output and fraction of a volt as DC. These units are potted
with that white hard-rubber-like stuff but I have one that I excavated
out a lot of the fill and that is the same fault. No green corossion
stains or oil films seen associated with any components still in the
fill (only removed enough to check what components were in there.
Philips EHC 070 S/40 , no controller IC , 2 STL57 driver transistors and
the small componentry as SMD on a small daughter board. Stable 350V on
the rectified mains caps, polyester, no electrolytics at all in these
units. Transformer is good up to 1Gohm on insulation test. Fuse and
thermal fuse in the exposed one are good andother stuff cold test ok

 

[Sjouke Burry]

I have a use for one of these all new unused in 1990s, not stored in a
shed or garage over the years. A few dozen ,made in 1992 nice compact
70W Electronic Halogen Transformers (so called including labels as
primary and secondary on the external label). Outputs the unrectified HF
to drive the 12V lamps. They used to work 15 years or so ago. Now I've
tried 3 and all the same , should be 12V HF ac and if attached to high
speed diode and a cap and no load then only about 10 to 12V dc. No way
will they power even a 20 W lamp (supplies are rated 20W to 70W) tried
40W in case load dependent. Scoping, with no load and a complex
oscillation signal. Then as little as 27R load very poor mark space
ratio of output and fraction of a volt as DC. These units are potted
with that white hard-rubber-like stuff but I have one that I excavated
out a lot of the fill and that is the same fault. No green corossion
stains or oil films seen associated with any components still in the
fill (only removed enough to check what components were in there.
Philips EHC 070 S/40 , no controller IC , 2 STL57 driver transistors and
the small componentry as SMD on a small daughter board. Stable 350V on
the rectified mains caps, polyester, no electrolytics at all in these
units. Transformer is good up to 1Gohm on insulation test. Fuse and
thermal fuse in the exposed one are good andother stuff cold test ok

Old stored electronics can suffer from degraded caps.
To re-format them, connect the device to a variac or labbench power supply,
leave them for some hours at a low voltage, and step wise go to
nominal voltage(input side).
That will sometimes re-form the isolation layer in the caps.

If you already connected them to full voltage, they might be
permanently damaged.

 

[N_Cook]

Old stored electronics can suffer from degraded caps.
To re-format them, connect the device to a variac or labbench power supply,
leave them for some hours at a low voltage, and step wise go to
nominal voltage(input side).
That will sometimes re-form the isolation layer in the caps.

If you already connected them to full voltage, they might be
permanently damaged.

There are no electrolytics in there, do non-wet caps suffer from that
sort of failure? Only polyester or ceramic caps in there AFAICS. Got me
thinking, This is quite early SM so perhaps MLCC failure due to silver
migration on one or more of the chip caps on the small daughter board,
I'll try testing for ohmage/varying ohmage

 

[[email protected]]

Do any of the units work so you can make a point by point comparison??

 

[N_Cook]

Do any of the units work so you can make a point by point comparison??

I'll try powering up some more, but as they are fully potted it would be
just go/no go. I had an exploration of the SM caps on the excavated one
and nothing stood out. There is one cap on the main pcb that is dark red
, maybe silvered mica, I will excavate the remaining "pottage" and
desolder it and test for silver migration. Then isolation transformed
and scoping the live board as now I've started I've got to get to the
bottom of what could be a smps generically fault process. I don't
usually have a batch of all the same anything to look over

 

[N_Cook]

I powered up 6 more. 3 bad a short duration glimmer, 3 usuable, but not
quite right.
IIRC these could have the lamp load connected at power up. These 3 will
not do that but connect load after powerup and will power 20W test lamp
ok. So presumably something has gone wrong with load sensing/demand
somewhere, what to look for in the open exposed one?