Transformerless PSU using a zener/transil to drop voltage

[[email protected]]

Hi all,

For a hobby home automation system I need to make a very small, non
isolated, direct-mains power supply unit regulated via a 7812 IC or such.
Thus it has to be transformer-less, and since I need to draw something
like 100mA, even a capacitor (of the kind that do not explode on mains!)
would be too big. I have two ideas so far to lower the voltage to a range
acceptable by the 7812:

1) a rectifier in serie with a zener (even better a transil), the latter
used to drop the voltage so that only the peaks of the AC top waveform
come through, and lessed by ~300 volts. Care must be taken because if the
(230VAC case) RMS voltage can be anywhere from 210V to 250V, then it means
that the DC coming out of this circuit will have the ~same offset range
(eventually a fuse will be blown or, better, I'll use a polyswitch).
Another thing that worries me is power dissipation: if I draw 100mA, then
of course those 100mA will have to be multiplied by the zener/transil drop
voltage, which (again, 230VAC case) will be, say, 300V: 30W! But anyway
this will only happen for a fraction of the 50Hz waveform, when the diode
conducts (probably much, charging a low voltage capacitor), so the average
power dissipation of this diode hopefully should still be acceptable.

2) if the above is not feasible, I was thinking about working on the low
part of the AC waveform rather than on the peaks: maybe a high voltage
MOSFET which conducts only when the serie voltage is under e.g. 20V? For
the rest of the AC waveform the MOSFET would be turned off. I should use
a 20V zener (in a high impedance way, something zeners really don't like)
and a P-MOSFET, intuitively, or a depletion-mode N-MOSFET.

Any suggestions/comments please?

Thanks!
Andrea

 

[Fred Bartoli]

[email protected] a écrit :

Hi all,

For a hobby home automation system I need to make a very small, non
isolated, direct-mains power supply unit regulated via a 7812 IC or such.
Thus it has to be transformer-less, and since I need to draw something
like 100mA, even a capacitor (of the kind that do not explode on mains!)
would be too big. I have two ideas so far to lower the voltage to a range
acceptable by the 7812:

1) a rectifier in serie with a zener (even better a transil), the latter
used to drop the voltage so that only the peaks of the AC top waveform
come through, and lessed by ~300 volts. Care must be taken because if the
(230VAC case) RMS voltage can be anywhere from 210V to 250V, then it means
that the DC coming out of this circuit will have the ~same offset range
(eventually a fuse will be blown or, better, I'll use a polyswitch).
Another thing that worries me is power dissipation: if I draw 100mA, then
of course those 100mA will have to be multiplied by the zener/transil drop
voltage, which (again, 230VAC case) will be, say, 300V: 30W! But anyway
this will only happen for a fraction of the 50Hz waveform, when the diode
conducts (probably much, charging a low voltage capacitor), so the average
power dissipation of this diode hopefully should still be acceptable.

No, it still will be 30W (average current in = average current out).

2) if the above is not feasible, I was thinking about working on the low
part of the AC waveform rather than on the peaks: maybe a high voltage
MOSFET which conducts only when the serie voltage is under e.g. 20V? For
the rest of the AC waveform the MOSFET would be turned off. I should use
a 20V zener (in a high impedance way, something zeners really don't like)
and a P-MOSFET, intuitively, or a depletion-mode N-MOSFET.

Any suggestions/comments please?

I think it's Linear that has a circuit doing exactly this but I can't
find the part# now.

You can also adapt this: http://www.edn.com/contents/images/21700di.pdf

 

[Homer J Simpson]

For a hobby home automation system I need to make a very small, non
isolated, direct-mains power supply unit

Why non isolated, direct-mains?


--
..

--
..
..
..
..
..
..
..
..

--

 

[Wimpie]

Hi all,

For a hobby home automation system I need to make a very small, non
isolated, direct-mains power supply unit regulated via a 7812 IC or such.
Thus it has to be transformer-less, and since I need to draw something
like 100mA, even a capacitor (of the kind that do not explode on mains!)
would be too big. I have two ideas so far to lower the voltage to a range
acceptable by the 7812:

1) a rectifier in serie with a zener (even better a transil), the latter
used to drop the voltage so that only the peaks of the AC top waveform
come through, and lessed by ~300 volts. Care must be taken because if the
(230VAC case) RMS voltage can be anywhere from 210V to 250V, then it means
that the DC coming out of this circuit will have the ~same offset range
(eventually a fuse will be blown or, better, I'll use a polyswitch).
Another thing that worries me is power dissipation: if I draw 100mA, then
of course those 100mA will have to be multiplied by the zener/transil drop
voltage, which (again, 230VAC case) will be, say, 300V: 30W! But anyway
this will only happen for a fraction of the 50Hz waveform, when the diode
conducts (probably much, charging a low voltage capacitor), so the average
power dissipation of this diode hopefully should still be acceptable.

2) if the above is not feasible, I was thinking about working on the low
part of the AC waveform rather than on the peaks: maybe a high voltage
MOSFET which conducts only when the serie voltage is under e.g. 20V? For
the rest of the AC waveform the MOSFET would be turned off. I should use
a 20V zener (in a high impedance way, something zeners really don't like)
and a P-MOSFET, intuitively, or a depletion-mode N-MOSFET.

Any suggestions/comments please?

Thanks!
Andrea

Hello Andrea,

When it is a hobby project, and even a professional one, I would drop
both options. 100mA at 12 volt is 1.2W and that is not low power in
my opinion.

Many designs that does not use a transformer, use a capacitor directly
connected to mains. The capacitor serves as a dissipation less current
limiter. The capacitor feeds a 2 diode rectifier with storage
capacitor and zener (regulator) diode.

The rectifier must be loaded, otherwise the voltage rises to several
100 V, therefore the regulator diode is added. Note that when your
circuit doesn't consume the designed current, all power is dissipated
by the regulator diode (in your case about 17V to feed the 7812).

A resistor is in series with the capacitor to absorb transients.
Mostly a varistor across the mains is added to absorb large
transients. The design of such a capacitor/rectifier circuit is not
easy if you want that you circuit will survive over voltage category 3
or 4 transients.

When it comes to space saving, the capacitor will be in the 2..4 uF
range and therefore will have a large size. I would consider a
transformer for that power level. Even professional energy meters do
use a small mains transformer rather then the transformerless approach

Hope this will help you a bit.

Best regards.
Wim

 

[Bruce Varley]

Why non isolated, direct-mains?

There's another option if the situation involves another device nearby
that's mains powered. Put a current transformer (which can be made very
compact) in the feed to the other device and use secondary windings on the
transformer to provide your power. Since these will be current sources not
voltage sources, you can feed zener diodes without series resistors, just
rectifier diodes or a bridge. Just one thing, if you only need one half of
the cycle, make sure that you include a backward diode across the secondary
to absorb the other half cycle.

I've used this technique for signal conditioners inside industrial power
panels, where there was plenty of current around and I didn't want to use a
'real' PSU.

 

[MooseFET]

2) if the above is not feasible, I was thinking about working on the low
part of the AC waveform rather than on the peaks: maybe a high voltage
MOSFET which conducts only when the serie voltage is under e.g. 20V? For
the rest of the AC waveform the MOSFET would be turned off. I should use
a 20V zener (in a high impedance way, something zeners really don't like)
and a P-MOSFET, intuitively, or a depletion-mode N-MOSFET.

Any suggestions/comments please?

Use a comparitor to switch the MOSFET off if the output voltage
exceeds 12V or the drop on the MOSFET exceeds, lets say, 6V.

The mains spends only a smallish time inthe 12V to 18V span so the
MOSFET needs to be able to conduct way more than 100mA when on.

 

[Alex]

Hi all,

For a hobby home automation system I need to make a very small, non
isolated, direct-mains power supply unit regulated via a 7812 IC or such.
Thus it has to be transformer-less, and since I need to draw something
like 100mA, even a capacitor (of the kind that do not explode on mains!)
would be too big. I have two ideas so far to lower the voltage to a range
acceptable by the 7812:

could also use a lightdimmer !! the TRIAC or SCR could be a solution
as it do not dissipate much power . Maybe it can be a problem to
filter the voltage from the TRIAC/lightdimmer as the consumption is
100 mA but I guess it will be the same if a FET is on only until the
voltage are over say 20 volt
i

 

[B Fuhrmann]

2) if the above is not feasible, I was thinking about working on the low

"MooseFET" wrote in message ...

Use a comparitor to switch the MOSFET off if the output voltage
exceeds 12V or the drop on the MOSFET exceeds, lets say, 6V.

The mains spends only a smallish time inthe 12V to 18V span so the
MOSFET needs to be able to conduct way more than 100mA when on.

I think that the Origonal Poster should consider using a transformer since
the need for the larger capacitor is a hint that it isn't really a small
supply.

In the idea of switching the power:
About 15 years ago, I used an app note from Linear Technology to create a
very high efficiency transformerless power supply (it was going into a well
insulated case that plugged into the utility meter socket on industrial
buildings).

The first stage of the supply was a simple and dirty switching supply.
Basically (there were more parts involved) a capacitor with a transistor to
the voltage input that turns on whenever the capacitor is < x volts. As
soon as the capacitor is charged, the transistor shuts off.
As the load draws power, the capacitor drops voltage slightly and the
transistor drops more energy in from the supply. The hysterisis in the
control of the transistor becomes ripple in the supply.
In the system they were pushing in the app note, the output of the cheap and
dirty switching supply went into a low dropout linear regulator that they
had recently come out with.

 

[CWatters]

Hi all,

For a hobby home automation system I need to make a very small, non
isolated, direct-mains power supply unit regulated via a 7812 IC or such.
Thus it has to be transformer-less, and since I need to draw something
like 100mA, even a capacitor (of the kind that do not explode on mains!)
would be too big. I have two ideas so far to lower the voltage to a range
acceptable by the 7812:

I'd advise you not to try this. Too dangerous unless you now what you are
doing. Wall warts are as big as they are for a reason.

If you are aware of the risks consider buying an AC-DC converter module
perhaps..

http://www.cdiweb.com/ewave/435

 

[Paul Hovnanian P.E.]

[email protected] a écrit :

No, it still will be 30W (average current in = average current out).

Exactly. So if the OP can afford the real estate to heat sink 30W, a
transformer shouldn't be an issue.

A switched mode supply shouldn't be too demanding at this power level
and will be lighter and smaller than one with a transformer. But if the
OP is challenged by the exploding capacitor problem, I'd suggest a quick
trip to the local electronics shop to buy a ready made wall wart.

100 mA @ 12V should be a cinch and will eliminate the safety risk. The
stuff on the 12V side will be safe and easier to tinker with.