ard at p850ug1.demon.co.uk
Tue Jun 8 13:10:49 CDT 2010
> > Transformers are extremely simple beasts - they either work or they
> > don't. When they fail you really know it - they blow fuses and/or start
> > smoking with goop leaking out and making a general smelly mess or a
> > winding opens and you get no output at all. A transformer does NOT get
> > weak.
> I completely understand and I never believed it "got weak". I believe
> it has been defective from the beginning. Probably not a failure but
> a manufacturing defect or was just plain too small for the job.
Do you even know that it was the right transformer? From what you said
elsewhere, you got the specs from the SWTPC scheamitcs, not from markings
on the transofrmer itself. Somebody might have used the wrong part (or
tried to save money, or...)
> > Switchers are great for protecting against power supply failures,
> > however some reset circuits assume linear supplies and don't work
> > reliably with switching supplies unless you modify them so the voltage
> > rails go up as the system expects.
> Understood. This system has a 555 timer that resets it sometime a few
> SECONDS after 5V goes valid. I think we'll be OK.
Not always. Some ICs -- the 4116 is the best known example, can do
strange things (including latching up, then drawing excessive supply
current and destroying themselves) if the supply lines come up in the
wrong order. This has nothing to do with the system reset line. IIRC the
4116 needs the -5V bias supply applied before the +12V supply.
> > If you decide against switches get TRANSORPS for the critical voltage
> > lines to clamp the voltage (and blow fuses) if the linear supply voltage
> > rises due to failure.
> A good plan. Thank you.
Or make a crowbar circuit. It's just a zener diode, a couple of resisotrs
and an SCR. Actually, I am suprised nobody sells crowbar 'ICs' -- 2
terminal components that contain that circuiit. Not the same a sa
Transorb, becuase once a crowbar has triggered, it stays effectively
shorted until the current through it drops to a low value (it is an SCR
More information about the cctalk