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.
X-Mailer: Mozilla 4.79 [en] (Win98; U)
Subject: Re: Problems with MOSFET drivers
NNTP-Posting-Date: Fri, 15 Nov 2002 17:55:50 GMT
Organization: AT&T Broadband
Date: Fri, 15 Nov 2002 17:55:50 GMT
Mike Poulton wrote:
> I'm in the process of building a big (4kW) solid state tesla coil. The
> basic idea is to use a mosfet H-bridge to drive the coil's primary from
> a 300VDC source. There is a small feedback winding on the coil as well,
> which provides a signal to switch the mosfets. The mosfets are driven
> with a single toroidal transformer with one primary and four
> secondaries. Each seconday goes to a fet, and they are phased to turn
> on one pair on the positive cycle and the other pair on the negative
> cycle. Therefore, the primary needs to be driven with a full bridge.
> To do this, I am using a pair of drivers from Microchip, the TC4421 and
> TC4422. These are supposedly bulletproof. They come in 5-pin TO-220
> cases and are designed for 9A output. I have the pair (one inverting,
> the other noninverting) in parallel pin-for-pin -- except the outputs,
> of course. The output has a .1uF DC blocking capacitor in series with
> the primary of the gate drive transformer. There is a 5.1V zener from
> the inputs to ground, and a 4.7K resistor between the feedback winding
> and the inputs. A .001uF cap goes from the inputs to ground as a low
> pass filter in order to ensure that the coil oscillates in 1/4 wave
> mode, not a higher mode. I have a .1uF bypassing capacitor right across
> the IC power leads. In order to start oscillation, I have to give the
> input a 5V pulse through a 330 ohm resistor in order to trigger a state
> change and get things going.
> Here's the problem: the drivers explode! Loudly and reliably. The top
> of the package disappears and you get to see the die. If the driver
> board is powered but the H-bridge is not, nothing happens. As soon as
> power is applied to the bridge (only 12VDC, not the full 300) and you
> give the input a 5V pulse, it goes nuts and the drivers explode. There
> is no output from the coil during the brief time it operates. I tried
> using a current limited supply and fresh ICs, and the driver board drew
> the full 3A. That current should not damage the ICs, but it did -- they
> show a short from Vdd to ground now. I've gone through six pairs of
> drivers now, at $3.80 per chip. Seems clear that it's oscillating in a
> bad way, but I can't tell how! I cannot comprehend what abnormal
> operation would cause it to destroy itself like this -- they are
> supposed to be super-tough devices. Any ideas?
Yes. The feedback waveform is both delayed and not symmetrical about
the input threshold of the driver ICs. This will lead to an improperly
phased, non square output waveform to the gate drive transformer.
You can check this out by moving the gate drive primary to a second
set of drivers connected to the function generator while leaving the
feedback winding from the tesla coil hooked up to the first driver set
and simultaneously monitoring and noting the differing waveforms between
the outputs of the two sets of drivers. They will not look the same.
Both duty cycle and phase shift will be not what you want.
Perhaps the following modification might lead to the desired results:
The tesla coil feedback winding should have one end ac grounded (this
may be at half of your dc housekeeping supply) with the other end run
through an adjustable R/C phase shift circuit into one input of a
comparator and with the other input connected to the ac grounded end
of the feedback winding. The comparator output goes to the inputs of
the driver ICs.
At this point you should still have both sets of driver ICs connected.
Now with the circuit running under the timing of the function generator,
adjust the polarity and phase shift of the feedback circuit until you
get matching output waveforms from the two sets of driver (you may need
more phase shift than possible from a single R/C circuit). Only when
you've gotten the waveforms to match should you attempt to run under
feedback drive. After you get the feedback drive hooked up and running,
you may find that further tweaking of the phase shift optimizes
performance. Good luck. -- analog
Go Back To The Cyber-Spy.Com
Usenet Web Archive Index Of
The sci.electronics.design Newsgroup