On 06/02/2019 13:04, Ahem A Rivet's Shot wrote:
> On Wed, 6 Feb 2019 10:54:45 -0000
> "NY"  wrote:
>
>> "The Natural Philosopher"  wrote in message
>> news:q3e8ur$ms2$1@dont-email.me...
>>> On 06/02/2019 09:13, NY wrote:
>>>> I'm always amazed at how small switched-mode power supplies are these
>>>> days: the ones that Amazon supply for the Kindle fit the mains
>>>> transformer, the diodes and the PSU control circuitry into the size of
>>>> a normal-pin plug.
>>>
>>> Dont be.
>>>
>>> Understand that a transformer is, for the purposes of power transfer,
>>> able to store so much energy in its magnetic feild per cycle of the
>>> frequency it is operated at.
>>>
>>> A 5W transformer at 50Hz needs to be 1000 times the volume (or ten
>>> times thh (liner) size) of one operated at 50KHz...and as power FETs
>>> get better and better at HF switching so the frequency goes up.
>>
>>
>> I presume the switching is done at mains voltage (to avoid the need for a
>> step-down transformer at 50 Hz) and then  the 50 kHz switched, regulated
>> voltage is then stepped down to 5 V where it is smoothed.
>
>  The regulation is usually applied by measuring the 5V output to
> derive the feedback to the switcher which does indeed run at rectified mains
> voltage. High voltage switching transistors (FETs) are far smaller and
> cheaper than low frequency transformers and efficiencies can run well into
> the 90s mainly due to storing far less energy in the PSU which means
> there's less to lose.
>

Mm. The main sources of loss in such a beast are

(1) the diode losses in the HT rectifiers. A good power diode will lose
about 0.6-1V at say 400V (European voltages) so with a bridge that is
one in 200 - say 0.5% there.

(2) ON resistance in the power FETS. this is again pretty low. Similar
to the diodes. So lets add anpother 0.5%.

(3) Switching losses in the power FETS. This is a very variable thing
depndoing not jost on the device used - the actual one, just not what is
on the can - and how fast it us chopping. The actual losses are high
here, because at the time when its not got either high current but low
(<1V) volts across it nor yet no current and high voltage...it is in a
state of extremely high potential dissipation. This is where design
compromnises happen. The faster you switch it the more this time in
between on and off is, compared with the  cycle  period, and the hotter
it gets. BUT the faster you chop it the less ferrite you need in the
transformer. This is probably one of teh gretest losses in the deveice =
and accounts for much of the 5%-10% loss that you state is typical,
which feels about right to me.

(4) Ferrite losses. Driving a small core into near saturation coupled
with its hysteresis will make the core hot. Cost savings will push the
designer towards the smallest core with acceptable efficiency. Higher
switching speeds mean the core is less stressed, but that stresses te
FETS more.

(5) copper losses in the transformer. The ohmic losses from the rather
long bits of wire which will exhibit skin effect higger resistance than
you might expect, also is a feature. Less turns is better, and fatter
wire is better. This again drives you twoards higher switching speeds.

(6) output diode losses. If - say - you have 1V lost in a 5V output
thats a LOT of actual losses. Fortunately te cahracteristics of a
shottky diode are good in the seconadry circuit - very fast switching
times and low forward voltage. With a balanced secondary you can end up
dropping not much more than 0.4V in - say 5V,. which is around 8%.

(7) some power needed by the actual electronics inside. This is probably
bugger all - a few mW.

the rectifier losses and electronic losses  are more or less fixed
percentage wise. The other losses are likely to increase with load.


--
Karl Marx said religion is the opium of the people.
But Marxism is the crack cocaine.

--- SoupGate-Win32 v1.05