[Patrick Turner]

Phil Allison wrote:

"Patrick Turner"


When I measure the "240Vac" here is usually is stable enough to get a
nearly constant reading on a DMM,

** Must be a basic 3.5 digit one ( 2000 count) with only 1 volt resolution
when reading 240 volts AC.

Any DMM with a larger count allows changes of 0.1 volts to be seen -
then the last digit is never steady.

Indeed, I'll get 240.XX Vac maybe even 24X.XX if the voltage is just
either side of 240.0Vac.

That's less than 1% Vac change.

But it slowly varies between 235Vac on cold winter nights of heavy
loadings to 255Vac when load is light.

** What drivel.

Not so, this is without changing local loads here in my shed.

Rarely does the mains ever bounce rapidly between 235Vac and 255Vac.

One has to be careful when setting bias in fixed bias amps when mains
Vac is low. When it rises to 255V sometimes, the B+ rails goes high and
the Ia of the output tubes rises alarmingly in fixed bias triode amps
and also in multigrid output tubes if the screen voltage is not
regulated, and rises also with mains rise.

It will instantly drop by 7 or 8 volts if you switch a ( 2.4 kW) electric
heater on AND jumps up by 6 volts when the ( 2kW) jug turns itself off
when it has boiled.

Not necessarily so.

But the combined efforts of hundreds of people in my suburb all boiling
jugs and turning heaters on and off contributes to making the mains
jitter up and down in levels we see, and not by the amount you state
unless the mains wiring is over a long distance.

A heater of 2.4kW draws 10 amps ac, and if the wire has only 1 ohm
resistance from street wires to the jug, a 10Vac drop would occur.
But my house supply wires are rated for much more than 40 amps, and the
shed power supply is an extension off the 40amp stove & laundry circuit
and has a sub board in the shed with 6 metres of wire rated at 20A to
the outlets. I doubt resistance is much more than 0.1 ohms.


When I look at the rectified Vdc, it shows the expected variations of
+/- 30mV.

** Complete ********.

No.

Any unregulated DC supply FOLLOWS all variations in the AC voltage by the
same percentage.

Agreed.

You misunderstand me. I have been trying to say this 3 times now.

But the typical B+ rail variations I see is in the form of low F noise
below 5Hz and down to Dc and ranging
+/- 30mV peak. Occasionally, the peaks are larger when someone turns
something on or off that draws severe current, or when 10 people
nearby all happen to turn heaters on/off within a very short time.

Some places will be much worse than mine.


Regulated DC supplies REGULATE against supply voltage and load current
variations, not just load current.

Indeed they should, unless the input Vdc falls below the level required
across the pass device to maintain regulation or if the current output
is limited, and the output tries to exceed the limit. Thus regulators
have to work with a voltage x current range and power is dissipated in
the pass element, or shunt element if its a shunt reg. Both cost time
and effort to make and you pay for the power.

Not greenhouse friendly either. But then many people use far more energy
in their warm cosy houses in winter while I type to you huddled over a
750W heater, or use 9kW air conditioning to prevent being sweaty and
smelly in summer.

But I digress...

LC filters have resonances and can make rail noise at LF *worse*, but
don't do as well as RC filters at real low F. Humungous values of C is
the simplest solution to B+ rail changes at F above 0.3Hz. But those
pesky noise voltages below 0.3Hz will persist. But such noise is easily
filtered by regulating the the B+ for the critical stages only, so the
power consumption and cost of building the reg for say 20mA is very
easy.

Perhaps there are ways and means of making a two stage CLCLC filter so
that the Fo of each LC section is different, and the first one acts to
make a series impedance which critically loads the Q of the second LC
section so the final response is without peaks around either Fo of each
LC section. Perhaps adding a series L between the center C and 0V of the
CLCLC so that its tuned to 100Hz will make a trap for the 100Hz, and
keep the size and weight of all chokes to small while C is high because
its cheaper than L.

There used to be an only passive filter calculator at some pomme
university but the guy who wrote the program which couldn't be
downloaded died and I have not seen any other program. I'm not much good
at calculating passive CLR filters with 6 or more elements.

Once established, the CLC type of filter wastes very little power.

Patrick Turner.

..... Phil