Manfred
You provided an enlightening and useful comment; thank you. I was unaware of
this problem with electrolytics in "switchers", but most of my experience has
been with elctrolytics used to filter 120 Hz ripple. In the few switching
power supplies I have made, I used low ESR electrolytic caps that the
manufacturers said were designed for use in switchers -- for frequencies up to
several hundred KHz (hereinafter "MF electrolytics"). They have worked fine,
but I've never used them in anything for CCS, and I tend to overdesign power
supplies.
I have some questions about this subject, and would appreciate your advice:
1. I've read conflicting recommendations about adding a cap that presents a
low impedance to RF, such as a .01 uF disc ceramic, in parallel with the MF
filter electrolytics (in "switchers"). This would be to shunt out the RF
components of power from the switching circuitry. I've heard some say "yes"
to this cap, and others say "no" because it can form a resonant circuit at some
RF and overheat or cause instability in the switching circuit, and others say
it is a good idea but only if you use a cap that will handle the probably high
RF currents. What do you think about this practice -- paralleling the
electrolytics with disc ceramics for the RF? Would this advice vary depending
on whether the ripple frequency was 120Hz or 20 KHz?
2. What other sort of requirements and standards, besides low ESR and being
designed for several hundred KHz, should a filter cap for, say, a 20 KHz
switcher meet? Again, I like to overdesign some for long-term reliability.
Are there any to avoid?
3. In the switching power supplies I built, I have two banks of filter
capacitors: one (that I nickname the "reservoir") being the electrolytic caps
fed by rectified AC from bridge rectifier; this supplies DC to the switching
circuitry, and two, the filter caps for the rectified DC from the switching
circuitry. In the "reservoir" capacitor bank, I use both "LF" electrolytics
(the ones usually used for rectified 60 Hz) and MF caps in parallel. The idea
is for the LF caps to smooth the 120 Hz ripple, and the higher frequency, low
ESR caps to serve as a low impedance path for the switching supply circuitry.
I've had no problems with this, but do you have any comments? Might these caps
form some resonant circuit like a .01 uF cap?
TNX ES 73,
Gene May
WB8WKU
> Date: Thu, 21 Feb 2013 13:58:33 +0000
> From: manfred@ludens.cl
> CC: amps@contesting.com
> Subject: Re: [Amps] Peter Dahl transformers
>
> Peter,
>
> > The downside to the switcher must be reliability. There are far more
> > parts, so the MTBF must be less.
>
> Yes, this part is true, specially when comparing a switcher to a simple,
> non-regulated transformer-type supply. Instead when comparing a switcher
> to a regulated linear supply, the MTBF is often in favor of the
> switcher, because the pass transistors in the linear supply typically
> run very hot and are a common failure point.
>
> With switchers, the most important factor to MTBF are the electrolytic
> capacitors, and specially the small ones, not the big filter caps. A
> great many designers of switching supplies just don't pay any attention
> to the pretty low ripple current rating of small electrolytics, and run
> them far above their rating! The result is that these caps dry out, and
> the supply fails. Probably anyone who fixes elecotronic equipment has
> already run into a power supply using an UC3842 controller, in which the
> small filter cap in the chip's supply has failed, making the beast
> hiccup instead of starting up correctly. I guess this is the most common
> single failure in all of electronics over the last 20 years! But this is
> not an intrinsic problem of switching supplies. It's a problem of
> circuit designers who don't know enough, or aren't careful enough, or
> just plain simply are intentionally designing equipment with a rather
> limited and quite specific life time!
>
> When a switching supply is designed in such a way that the electrolytic
> capacitors are used well within their ripple current and temperature
> ranges to last long enough (20-30 years, at least), and the
> semiconductors are all used well within their SOA, typically the result
> is a very high MTBF. It's not hard to get 300,000 hours MTBF! Some
> quality manufacturer already are around 5 million hours MTBF, but such
> numbers are pretty academic to most users. On the other hand, I have
> suffered my share of badly designed switching power supplies, including
> cellphone chargers, compact fluorescent lamp ballasts, LED drivers,
> laptop computer power supplies, and the like, that hardly reach 500
> hours MTBF!
>
> It's in the hands of the designer.
>
> I don't have MTBF figures on the switching supplies I designed myself.
> When I was starting doing this, at age 18, several of them failed rather
> quickly, and through them I learned to do it right. I'm not aware of any
> failure of any of the power supplies I built as a professional, and some
> of these are now in use for more than 20 years, 24/7, in an
> industrial/scientific environment. The only goof was one multivoltage
> switcher that had to work under extreme conditions, and failed to start
> at a temperature of -25 degrees Celsius. It was supposed to work at that
> temperature, and of course didn't show any weakness while testing it in
> the freezer... I had to swap an IC for an equivalent by another
> manufacturer, to fix this. I don't consider this a failure, but a "field
> testing result"... :-) A failure would be something that worked, and
> someday, under the same conditions, stops working.
>
> In short, it's indeed more difficult to achieve a required MTBF in a
> switcher than in a very simple non-regulated linear supply, but it can
> be done, and the cost is often still very much more convenient than that
> of a big transformer!
>
>
> Manfred
>
> ========================
> Visit my hobby homepage!
> http://ludens.cl
> ========================
> _______________________________________________
> Amps mailing list
> Amps@contesting.com
> http://lists.contesting.com/mailman/listinfo/amps
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|