The Cyber-Spy.Com Usenet Archive Feeds Directly
From The Open And Publicly Available Newsgroup
This Group And Thousands Of Others Are Available
On Most IS NNTP News Servers On Port 119.
Cyber-Spy.Com Is NOT Responsible For Any Topic,
Opinions Or Content Posted To This Or Any Other
Newsgroup. This Web Archive Of The Newsgroup And
Posts Are For Informational Purposes Only.
From: "Bill Sloman"
Subject: Re: Simple adjustable 0 - 400V regulator, with flaw
Date: Sun, 19 Jan 2003 00:00:58 +0100
Organization: Planet Internet
NNTP-Posting-Date: 18 Jan 2003 23:01:41 GMT
X-Newsreader: Microsoft Outlook Express 6.00.2800.1106
"Winfield Hill" wrote in message
> The Intersil HIP5600 is an adjustable three-terminal regulator
> that can be thought of as a 500V version of the LM317. I've
> used it in many of my lab instrument designs and found it to
> be reliable and, like the LM317, an elegant design simplifier.
> For example, here's a simple adjustable 0 to 400V regulator,
> with a fixed-voltage feature.
> . +480V ___ HIP5600
> . ---| |---+---+-----------(O) HV out
> . |___| | |
> . | | 1.5k
> . | | | adjust, 0 - 400V
> . '---- | --+----o->o--------,
> . | o--, |
> . 10uF === 300V | 500k
> . 450V | fixed | pot
> . | 380k |
> . ______|______________|_____|____
> . gnd
> The switch allows one to select a fixed 300V output, or to
> adjust the voltage to any value between 1.2 and 400 volts.
> But this circuit has one potential problem that can cause
> a component failure. As a design quiz, tell us what the
> problem is, and find a solution that doesn't significantly
> affect any of the circuit's functionality. Hint, the '5600
> isn't the vulnerable part.
I came to this thread after the pot had been revealed to be vulnerable to
the stored energy in the 10uF filter capacitor.
My own solution would be stop using the pot as a rheostat, and to replace
the control components of the circuit with a 500V 4-transistor current
mirror. Your 380k resistor would then deliver 800uA into one side of the
current mirror, which would be set up for 1:1 operation in the fixed 300V
The output transistor feeding the 800uA back to the 1.5k would need to be a
500V part capable of dissipating 400mW at 400V, which doesn't sound too
bad - the chip would have to be big enough to able to absorb the 0,45 joules
stored on the 10uF capacitor, almost all of which could be delivered to the
transistor at switchover. The other three transistors would be the matched
pair from a CA3046 at the bottom, and one of the spare transistors in the
CA3046 as the cascode on the other side.
In the variable mode, the current mirror would be biassed to sink a
progressively greater multiple of the 380k current as you twiddled the pot.
Every 60mV of extra voltage across the base-emitter junction of the
current-feedback transistor (lower left) would increase the multiplication
factor by a factor of ten, so you'd have a logarithmic control.
You'd put the pot - something closer to 50k than 500k - between the upper
pair of bases and the base of the lower current sensing (right-hand)
transistor. The wiper of the pot would drive the base of the current
feedback transistor (lower left). You wouldn't use the full range of the
pot, so a 10k padded with 30k fixed resistor above it would be better than a
Your switch would make or break a short between the wiper and the bottom of
This scheme breaks down at 7.6V outputs, when the 380k resistor doesn't
source enough current to sustain the 40k of pot and padding resistor. This
isn't quite your 1.2V lower limit, and the upper limit is 300V rather than
You could get the 300V fixed and the 400V range with an extra pole on the
switch and a second resistor - so that at 300V you'd have 470k//2M2 and for
the 400V down variable range, just the 470k. Messing around with the voltage
offsets would be more fun.
If you were prepared to bleed something like 100uA (48mW) from your 480V
supply to feed a micropower single supply op amp you could build a MOSFET
and op-amp current mirror to do much the same job - the (say 1.2V) reference
voltage for the current source would then be generated by the 380k resistor
as before, either driving 800uA into a fixed 1k5 to ground for fixed 300V
operation, or a variable current into a 500k pot to ground via a 2k4 padding
resistor - the voltage reference would then be taken from the wiper of the
pot, which gives you a low voltage of about 2,4V (with zero volts across the
current sinking MOSFET) and a high which rather depends on the tolerance on
the pot, but would be about 440V at nominal values. As in the previous case,
the output voltage would not be a linear function of wiper position - in
this case hyperbolic rather than logarithmic.
The nominally 2k4 to ground would probably need to trimmable over a fair
range to cope with the fairly horrible tolerances you get on potentiometer
The switch for this circuit would probably have to be double-pole to allow
you to get everything right,
Bill Sloman, Nijmegen
Go Back To The Cyber-Spy.Com
Usenet Web Archive Index Of
The sci.electronics.design Newsgroup