Path: utzoo!attcan!uunet!ogicse!zephyr.ens.tek.com!gvgpsa!gold.gvg.tek.com!grege
From: grege@gold.gvg.tek.com (Greg Ebert)
Newsgroups: sci.electronics
Subject: Re: convert 50Hz to 60Hz?
Message-ID: <1694@gold.gvg.tek.com>
Date: 21 Nov 90 19:06:54 GMT
References: <1990Nov20.185138.22731@rigel.econ.uga.edu>
Organization: Grass Valley Group, Grass Valley, CA  95945
Lines: 58


 victor@rigel.econ.uga.edu (Victor Grubbs (stat)) writes:
>I have a problem:
>
> [...]
>I need a 500 - 1000 watt 60 Hz (110 VAC or 220 VAC) power supply and I can 
>think of a couple of solutions:
>
>5) A step up ocillator that runs on 50 Hz and produces 60 Hz.
>
>      a) Transform 50 Hz power down to 12 VDC and use an inverter to produce
>	 60 Hz 120 VAC -- not bad except it's going to be 50% efficient and 
>	 I can't find anyone that makes it.
>
>I would like to find a number 5 type solution if possible.  I don't mind 
>making a #5, but I would prefer to buy one.
>

I built a 10kW 3 phase inverter with power bipolar transistors. Since you
are only running about 1kW, try a 450v/30 amp device such as the Mitsubishi
QM30DYH. It's a Darlington totem-pole device which actually has 2 switches
inside. You'll need 1 if you have +/- 170 VDC, or 2 if you have +170 VDC
only. Rectifiers/caps are cheaper, so I'd opt for the former. You can also
get an opto-isolated driver for the transistor from Mitsubishi. 

This beast will give you any frequency you want: 50Hz, 60Hz, 400Hz, etc. 
For laughs, run a 3450 RPM induction motor at 200Hz!

You *dont* have to have sinusoidal waves; Induction motors are very happy
with square waves. The key is to provide a signal with the same RMS and
peak values as a sinewave. If you rectify the AC line, you've got the
peak-voltage part solved. For a rectangular pulse, RMS = SQRT(duty_cycle)*Peak
or in this case, a 50% duty cycle gives you what you want. The waveform
will look like this:
       __        __
     _|  |_    _|  |_
	   |__|

The only potential problem is if you are driving a capacitive load, such as a
power supply. High dv/dt gives big current spikes. Solution: add some
inductance.

Once you get that running, you can get tricky and wipe-out the 3rd and 5th
harmonics with a waveform that looks like:
    _   __      __    _   __    ____
  _| | |  |    |  |  | | |  |  |    |
     |_|  |____|  |__| |_|  |__|    |__|

I can dig-up the timing parameters if you're interested. You can also
do a Fourier analysis and solve 'N' simultaneous equations for 'N' zero
(or near-zero) harmonics. SPICE is helpful, too. 

>>> IMPORTANT <<< Make absolutely certain that your control circuit *NEVER*
turns on transistors between the + and - supply, even during power-up/down
and failures. I used a Z80 and it choked on a power-down, and caused 2 devices
to turn on between the supply rails. The transistors *exploded*. There was
enough 1/2LI^2 energy in a 12" wire to destroy another transistor. Yes, it
was a costly mistake.