Ed Maste wrote:
>
> In article <32A85C12.10AE@whidbey.com>,
> Paul Mathews <optoeng@whidbey.com> wrote:
>
> >I don't hesitate to design without transformer isolation.  Many of my
> >products use series capacitor power supplies, and any mention of this
> >always brings out dozens of responses of the "You're crazy if you don't
> >use a transformer!" variety.  I take comfort in the simple fact that the
> >vast majority of electrical products in the world don't use a
> >transformer.  However, it is too true that knowledge and care are
> >essential.  Electricity can kill.  I use isolation transformers, battery
> >operated scopes, differential probes, etc.  Take care.
>
> Could you start me off on the practical design of a series capacitor supply?
> I've got a 2 channel sequencer using a 16C84 which uses a resistor-zener
> supply from the 120V, 120V through a resistor for zero crossing and two
> triacs driven directly from PIC pins.  However, the current required by
> PIC and triac drive means that the resistor has to be low enough that it
> gets somewhat warm.  I can see the idea of using a capacitor there but
> don't know where to start in terms of cap. value or type etc.
>
> Thanks.


For 120VAC nominal, 1.0uF film capacitor rated 400VDC or 250VAC will
provide 35mA or so at a low DC voltage.  You can scale the capacitance
and voltage rating up or down for other situations, but you will
probably find that many considerations lead to taking other approaches
if you need more than about 50mA.  Some people insist that you should
use 2 capacitors of twice the capacitance, one connected to each line
(mains) input wire, to further isolate the circuit from the line,
especially with unpolarized plugs.  My own view is that this is
unnecessary, provided that you insulate everything properly.  If you
need to drive your triac(s) directly, the 2 capacitor approach
complicates things (see below).

It's wise to put some resistance in series with the AC capacitor,
preferably a carbon comp type of 100 Ohms or so, to limit the effects of
transient turn-on and high frequency noise currents.  If you use SMT
thickfilm resistors, at least 3 '1206' resistors in parallel should be
substituted.  Also, I recommend transient absorbing avalanche breakdown
diodes (tradenames like 'Tranzorb') instead of MOVs to protect
everything.  MOVs degrade over time.  One popular tranzorb P/N is
P1.5KE120CA.

The load seen by the line with series capacitance must conduct current
in both directions, and there are lots of diode and regulator
arrangements that work.  Perhaps the simplest is a bridge rectifier with
a shunt regulator such as a zener diode.  Of course, you need a filter
capacitor across the zener, and you may need further regulation,
depending on your requirements.  The zener conducts any current not used
by the remainder of the circuitry.  For a 5.1V zener, it will have a
worst case dissipation of less than a quarter Watt.  If you want to use
a 5V regulator, choose a zener voltage high enough to keep the regulator
happy, > 7.5V for 78L05.

Regarding how to drive the triacs: You will discover that your circuit
'ground' is at a slightly different potential than one of the line
inputs (the rectifier forward voltages) and at a radically different
potential compared to the other (the voltage across the capacitor).  One
approach to the problem of putting the triac cathode at the appropriate
potential for direct gate drive is to use additional diode(s) in series
with the triac cathode so that load currents raise its potential to
approximately the same as circuit common.  Another approach is to use
optocoupler(s) to drive the triac(s), or, for light loads, use triac
output optocouplers.  Have a look at Motorola MOC3021 or similar.

Of course, the 'common' or 'signal ground' potential of your circuit
will ride on the line waveform, and makes for some interesting
oscilloscope probing problems.  Please be careful to use isolation
transformers, etc. when you're working with a live circuit.   In fact,
avoid using the line power input as long as possible and whenever
possible during development.  You may find that much of your development
work can be done with a dc supply, particularly if you use optocoupler
drive.  Alternatively, consider using 24VAC during development: scale
the capacitance up to 5uF or so (film cap rated ~100VDC).

Film capacitors come in lots of shapes and sizes.  Good small ones are
made by Illinois Capacitor and Paktron, among others.
--

Paul Mathews, consulting engineer
AEngineering Co.
optoeng@whidbey.com
non-contact sensing and optoelectronics specialists