[sci.electronics] DC-DC inverter

tytso@athena.mit.edu (Theodore Y. Ts'o) (04/26/89)

I am trying to design an alternative power supply for a photometer (the
GR1501-A, if anyone carse).  The problem is that it requires 3 45 Volt
batteries which are almost impossible to obtain, and when they are
available, they are extremely expensive.

So what I'm trying to design is a DC-DC inverter that will take power
from a 6V latern battery (or something similar) and covert it to a +90
and -45 volt output.  Since it's going to be used to power a light
meter, the output current is going to have to be pretty will filtered.
Ideally, it should be regulated, but I have no idea how to regulate
voltages this high.  Can you get IC's that will do something like
this?  

The other constraint is that it would be best if it could be designed to
use mostly common parts, but that's not that important.

I'd appreciate any suggestions on how to design such a beast or
pointers to books or articles on the subject.  Thanks!
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Theodore Ts'o				bloom-beacon!mit-athena!tytso
3 Ames St., Cambridge, MA 02139		tytso@athena.mit.edu
   Everybody's playing the game, but nobody's rules are the same!

commgrp@silver.bacs.indiana.edu (04/26/89)

>I am trying to design an alternative power supply for a photometer 
>(the GR1501-A...) it requires 3 45-Volt batteries which are almost 
>impossible to obtain, and when they are available, they are extremely 
>expensive.

>So what I'm trying to design is a DC-DC inverter that will take power
>from a 6V latern battery (or something similar) and covert it to a +90
>and -45 volt output...  Can you get IC's that will do something like
>this?   ...
>--
>Theodore Ts'o				bloom-beacon!mit-athena!tytso


Interim solution if space allows:  Stack groups of five 9v 
"transistor" batteries by their snap connectors, or remove their cases 
and solder the batteries together.  It's still expensive if you use 
alkalines, but has much greater capacity than those old-style "B" 
batteries.  9v alkaline batteries contain six little cylindrical cells 
smaller than AAA size, reported to have 500 mA-hour capacity.  The 
manufacturers claim 3-year shelf life.

--

Frank Reid     W9MKV @ K9IU   reidgold.bacs.indiana.edu
{inuxc,rutgers,uunet!uiucdcs,pur-ee}!iuvax!silver!commgrp

john@stiatl.UUCP (John DeArmond) (04/26/89)

In article <10837@bloom-beacon.MIT.EDU> tytso@athena.mit.edu (Theodore Y. Ts'o) writes:
>I am trying to design an alternative power supply for a photometer (the
>GR1501-A, if anyone carse).  The problem is that it requires 3 45 Volt
>batteries which are almost impossible to obtain, and when they are
>available, they are extremely expensive.
>
>So what I'm trying to design is a DC-DC inverter that will take power
>from a 6V latern battery (or something similar) and covert it to a +90
>and -45 volt output.  

I've been building this kind of inverter for years in order to satisfy 
my peculiar perversion with keeping old tube-type portable radios working.
My circuit is very easy to understand and build and is fairly efficient.

Rather than try to enter a schematic in ASCII, I'll describe the circuit
in enough detail to reproduce.

My circuit uses a toroidal or cup-type transformer whose center-tapped primary 
is driven by a square wave oscillator push-pull style.  I apply power 
to the center-tap and take each end to ground through a power FET transistor
of appropriate rating.  In series with Vcc applied to the center-tap is
another power FET series pass element used for regulation.

The oscillator is built around a hex schmitt trigger in a feedback 
configuration.  This connection uses 3 gates hooked end to end with
the output fed back to the input via resistors and caps.  See the 
National linear databook for a discussion of this circuit.  I trim the 
components to generate a square wave at about 5 khz.  Feed the output 
into another of the triggers in the package to clean the wave up.  Connect
the output of this gate to the gate of one of the FETs.  Also connect
the output of this gate to another trigger to generate an inverted signal.
Connect this to the other FET.

Take the last gate and connect a feedback resistor from input to output
to bias it as a linear inverting amp.  About 10k or so should do.
Connect the output to the series pass FET.  This is the regulation
circuit.  You will later develop a voltage from the output of the
inverter that controls the pass element.  You will have to play around 
with scaling resistors in order to establish an operating point for
the pass element.  This is not a very precise regulator but it works fine
for most tube applications.

Winding the transformer.

I build up the previous ciruit before winding the transformer so I can use
it as a test circuit.  You can get design manuals from core vendors such
as Feroxcube but for one-off construction, you can usually do as well
characterizing the core experimentaly which is nice if you find some 
uncharacterized cores.  

What I do is make a guess as to the proper number of primary turns.  
Experience says that 30 turns per side is a good starting point.  I then
connect the core to the driving circuit and measure the idling current
draw. Set the applied voltage to at least the minimum operating voltage 
pluse some margin.  For 12 volts, I usually use 10 volts.  I increase the 
number of turns until the idling current ceases decreasing.  The core is 
saturating at this point.  

Record the number of turns and calculate the necessary turns ratio 
for each output voltage.  I wind a separate coil for each 
voltage rather than trying to tap one coil.  Greatly reduces the incidence
of ground loops, a real problem with high impedance tube circuits.  Put
a convential full wave bridge rectifier and sufficient filtering as needed.

I usually select the highest current winding for regulation.  Establish
a voltage divider from this output to ground such that a voltage of the
proper level is developed for feeding back into the regulator circuit
described above.  The easiest way to find this voltage is to connect a
variable power supply to the input of the regulator and determine what
voltage is necessary to drive the pass transistor to the operating
level.

Since the transformer core is saturating, the other outputs should 
regulate very well.  In fact, for most radio applications where I run
the inverter from a gel-cell, I rarely need a regulator.  I simply apply
battery voltage to the CT of the primary and use the output without any
active regulation.

This circuit is very non-critical and easy to configure.  A schematic 
should not be necessary.  Just get some parts from radio shack and go to it.

John



-- 
John De Armond, WD4OQC                     | Manual? ... What manual ?!? 
Sales Technologies, Inc.    Atlanta, GA    | This is Unix, My son, You 
...!gatech!stiatl!john    **I am the NRA** | just GOTTA Know!!! 

tomb@hplsla.HP.COM (Tom Bruhns) (04/26/89)

> tytso@athena.mit.edu (Theodore Y. Ts'o) / 11:52 am  Apr 25, 1989 /
>I am trying to design an alternative power supply for a photometer (the
>GR1501-A, if anyone carse).  The problem is that it requires 3 45 Volt
>batteries which are almost impossible to obtain, and when they are
>available, they are extremely expensive.

I'll bet!!

>
>So what I'm trying to design is a DC-DC inverter that will take power
>from a 6V latern battery (or something similar) and covert it to a +90
>and -45 volt output.  Since it's going to be used to power a light
>meter, the output current is going to have to be pretty will filtered.
>Ideally, it should be regulated, but I have no idea how to regulate
>voltages this high.  Can you get IC's that will do something like
>this?  

I won't make any value judgements on the merit of doing this.  But
sure, it should be not only possible but fairly easy.  First, if the
thing is at all a power hog, you probably ought to use a rechargable
battery.  Either nickel-cadmium or sealed lead-acid ("gel cell")
could be appropriate, depending on operating time/charge and desired
size and weight.

The DC-DC converter should be fairly easy with a switching regulator
IC, a small transformer, possibly an additional transistor to handle
the current, and a handful of passive parts.  The IC would monitor
one of the output voltages through a voltage divider, and regulate
its pulsewidth (or freq) to maintain a constant voltage.  Proper
design of the supply should allow use of the other output with no
further regulation.  I'm a little out of touch with state-of-the-
art sw-regulator IC's, but would check on what Linear Technology,
National, Silicon General, and maybe TI have to offer.  They usually
publish fine apnotes that will get you very close to what you want.
My recollections are that the biggest problems have to do with practical
transformer design.  One nice thing is that transformers at 50 kilohertz
even for the voltages you want have rather few turns.  A final note:
for the rectifiers on the output, be sure to use high-speed devices!
Common rectifiers designed for low frequency work will blow up.

Hope these thoughts help -- folk have written whole books on this topic,
but you should be able to get something acceptable without having to
take a course on it...
>
>The other constraint is that it would be best if it could be designed to
>use mostly common parts, but that's not that important.

I'm sure I could do this with all parts except the transformer obtained
from a place like DigiKey or Active; the transformer core could be a
problem...maybe some kind soul in your area could help you out.
>
>I'd appreciate any suggestions on how to design such a beast or
>pointers to books or articles on the subject.  Thanks!
>=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
>Theodore Ts'o				bloom-beacon!mit-athena!tytso
>3 Ames St., Cambridge, MA 02139		tytso@athena.mit.edu
>   Everybody's playing the game, but nobody's rules are the same!
>----------