John et al 
Consulted the reference you gave (page 144 in the 6th edition) and can see
where the figure comes from.
The value of 4.2% quoted is, as you said, the RMS value of the ripple / mean
output voltage. The RMS value is not really appropriate for this purpose
and in fact is rather meaningless; ie. providing a DC supply to a PA. The
circuits in the book assumes a series DC choke  something the heavy
electrical people would use routinely but unnecessary for our purposes.
The actual instantaneous value of the full wave rectified waveform using the
6 diode equivalent of a bridge rectifier, (no connection to the star point)
is as follows :
Minimum value is 1.5 Vpk at the crest of one phase, current returning to
the other two phases at their 30 degree points, at amplitude 0.5.
Maximum voltage is at the 60 degree point of two phases, where the
respective amplitudes are SQRT(3) / 2 Vpk and SQRT(3) / 2 giving 1.732
Vpk. The third phase has a value of 0 here.
Hence max possible ripple is 1.732  1.5 = 0.232 Vpk, pkpk having a
fullwave rectified shape at 300Hz.
Mean output voltage is obtained by integrating over this 1.5 to 1.732
waveform, giving 1.65 Vpk, and a Peak / Mean ratio of the 1.05 quoted in
the Data Book. Can't be bothered calculating the RMS ripple value as it is
of little value, (and is complicated to do) but it looks as if a figure of
4.2% would be in the right sort of region.
If using a capacitive PSU input, the unloaded voltage would just be Vpk, and
ripple would be determined by capacitance and load current, but now at 300Hz
prf meaning lower values caps.
Using the standard mean UK supply of 234V rms nominal (405V phase voltage,
peak = 573V) the rectified supply would be 573 volts max, so my earlier
statement that 600V FETs could be used was of course rubbish, as the margin
is much too low. (Unless using a half wave rectified supply giving the same
340V as available from a single phase + neutral supply, with 150Hz ripple
frequency  not a good idea)
800  1000v FETs are only just beginning to appear and are very expensive
compared with 500V devices, so the idea of using full wave rectified 3 phase
supply for a mosfet PA is definitely not on.
However, US and Canadian stations with their 110V mains (supplied as
effectively 6 phases  3 phase with plus / minus each plus centre tap /
ground, seven wires in total, if I understand it correctly based on looking
at pictures of power cables in US towns) would result in a 270 V rail so,
much more realistic and safe using low cost 500V devices.
Andy G4JNT
Original Message
From: John Rabson [mailto:[email protected]]
Sent: 20010625 19:05
To: [email protected]
Subject: LF: 3 phase ripple (was Re: Re: RE: LX1PD)
Sorry, I meant rms.
For a 3 phase full wave bridge
RMS ripple/mean OP = 4.2%
(Ref Data for Radio Engs 5th ed p 135)
73 de
John Rabson G3PAI

The Information contained in this EMail and any subsequent correspondence
is private and is intended solely for the intended recipient(s).
For those other than the recipient any disclosure, copying, distribution,
or any action taken or omitted to be taken in reliance on such information is
prohibited and may be unlawful.
