Ever notice how many projects use the same sort of oscillator ->
gate driver -> half-bridge setup? If not, flybacks, induction
heaters and SSTCs are what I'm thinking of. Since they can all use the
same
basic driver why not make one universal driver to use for all these
projects? Fueled by this thought I made a TL494 based oscillator which
has separate pulse width and frequency control. This then runs a simple
gate
driver stage and finally a half-bridge. To ensure that I could keep
mosfet deaths
to a minimum I put some safety features in place. First a primary
current detection transformer is used to monitor the current through
the load, and shut down the inverter if some threshold is reached.
Second a thermistor is placed directly on one of the mosfets to monitor
their current temperature, again once a threshold is reached the
inverter is shut down. This eliminated two major causes of catastrophic
failure, though that's not to say things don't still go wrong. ;-) This
driver was used in my offline
flyback driver,
and is basis of the
breadboarded driver used in the series
resonant induction heater setup.
I now use my multi-inverter for both projects whenever I need a quick
demonstration.
The
TL494 part is pretty
straight-forward, pretty much just taken from a datasheet. The internal
comparators are in parallel and set up as voltage followers. The
voltage
at pin 1/16 (non-inverting inputs) is varied between 0,6 and 3,0V using
a potentiometer and some resistors. This alters the duty cycle from
maximum to a predetermined "safe" minimum based on the sharpness of the
switching waveforms. The discrete
gate driver is a full-bridge, allowing the use of 1:1:1 GDTs. It shorts
the core during dead-time, which effectively shorts the gates of the
mosfets preventing spurious drive signals. The CT part of the circuit
is simple analog circuitry. If voltage through a transformer is
stepped
up,
current is stepped down, so if 14A are flowing through the current
transformer's primary, and it has a ratio of 1:35, the current is 400
mA. Knowing Ohm's law, this
implies that the voltage over the burden resistor must be U= R*I, so
6,8Ω *
0,4A = 2,72 V. The rectifying diode steals about 0,13
volts (measured) from this,
resulting in 2,59V at the comparator. With the voltage divider on the
other input set to 3V this means the over current latch would still be
roughly 0,4V (2A) from triggering. I've made an open-office
spreadsheet to simplify
customizing the design. Don't bring the reference value over 3,5V, or
the shutdown state won't latch! Blame the LM324 for this... The reason
16A is selected despite the 14A
rating of IRFP450s is due to the anticipated current waveform the
inverter will be switching. The average value of the current should
remain well under 14A in most conditions. An idea for future iterations
would be an external current limit potentiometer. The thermistor
section is
just a couple of voltage dividers and comparator. The resistor values
were found by heating the thermistor while on a heatsink, until it was
deemed hot enough. The resulting thermistor resistance was roughly 3,3k.
The
sponge is for isolating the filter- and halfbridge capacitors.
I've
designed a PCB for this project, and the files can be
downloaded here.
The PCB in the image is of revision 1, which had some errors in the
shutdown circuitry caused by the LM324 not swinging above 3,5V. The
errors are fixed now, hence the different appearance of the PCB
layout in the zip.
Not
specified in the schematic are heatsink requirements. When switching a
14A load the IRFP450s will dissipate a total of 85W, and on top of that
the rectifier will contribute
with another 5W or so. In comparison the average PC draws
120W, most of which is turned into heat. This power needs to be removed
somehow, and we know how much effort goes into cooling a
PC. Though my setup isn't ideal, it's not designed for
continuous use so the puny fan/heatsink should suffice. Another
unmentioned aspect is construction of the GDT (gate drive
transformer) and CT (current transformer). Since it's used in so many
projects I've made a small article about them, which you can read here.
I usually scavenge CTs from ATX supplies and adjust the number of turns
to my requirements. When purchasing one however the same points apply
as when selecting a GDT core.
Multipurpose Inverter Mega
I've since built an upgraded inverter for driving even larger loads.
Utilizing a full-bridge with advanced discrete gate drive, it's my most
powerful inverter yet. Currently it is only used to power the Big Mofo
HV transformer.
The lightbulb serves to limit inrush current to the filter capacitors,
which could otherwise destroy the rectifier if it has an insufficient
pulse current rating. Most importantly it prevents the breaker from
tripping when turning this beast on, since 680µF or
more of filter capacitance is
quite the short when uncharged! Otherwise the circuit should be similar
to the MKI shown above. For powering loads requiring more than 1kW, use
even more filter capacitance. The over-current shutdown is set at 17A
peak,
rather than 12A. IRFP450s can withstand 18A peak, or 14A RMS, but
require heavy cooling to do so continuously.
Disclaimer:
I do not take responsibility for any injury, death, hurt ego, or other
forms of personal damage which may result from recreating these
experiments. Projects are merely presented as a source of inspiration,
and should only be conducted by responsible individuals, or under the
supervision of responsible individuals. It is your own life, so proceed
at your own risk! All projects are for noncommercial use only.