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: Mike Poulton
Subject: Re: Problems with MOSFET drivers
Date: 17 Nov 2002 20:44:31 GMT
Organization: MTP Technologies
References: <email@example.com> <firstname.lastname@example.org>
On 16 Nov 2002, email@example.com (Adam Seychell) said:
> You say that the drivers blow only when you connect the feed back
> You have not said what's exactly coming out from this winding. The
> drivers should be driven with fast square waves. It would be helpful
> for us if you explained more about your transformer configuration and
> what wave forms you observed during bench tests with function
> generator and no bridge voltage.
With no bridge voltage, there's no feedback signal. If I give the
primary a single pulse (or slow train of short pulses), the feedback
winding shows a decaying sine wave at the resonant frequency of the LC
circuit formed by the secondary and the terminal capacitance. When the
coil is operating at that resonant frequency, driven by a function
generator, with 24V to the bridge, the feedback winding gives about a
75V p-p sine wave.
I've tried two methods to condition this signal before giving it to the
FET drivers. First, I used a 10:1 voltage divider to provide a 7.5V p-p
wave. It didn't do anything. I discovered that a single pulse to the
primary did not transfer enough energy to the secondary resonant circuit
to produce more than 20V or so of ringing across the feedback winding.
Consequently, the driver was seeing 2V, which is not a logic 1. So I
tried a 10k resistor between the feedback winding and the driver input,
with a 5.1V zener from input to ground. This gave the driver a nearly
square wave from -1 to +5.1V.
Last night, I tried this again and got it to oscillate. I used 48V
across the bridge and a 1.5A constant current supply with 12V compliance
for the drivers. It worked quite well for about a minute -- long enough
to probe most of the waveforms. The input to the drivers was, indeed, a
5.1V nearly-square wave. The gate drive didn't look so good. There was
tremendous ringing at each switching transition, about 12V p-p at a
frequency too high for that scope to accurately show. Looking at the
primary voltage waveform showed partial switching due to that ringing.
The current was low enough that the mosfets didn't mind, but I'm sure it
would have caused extreme dissipation if I were running higher power.
Some gate resistors are definitely in order here.
While I was playing with some NE2 bulbs near the secondary, it suddenly
stopped working. The driver power supply showed 1.5A and just over a
volt. After I turned everything off, I measured a resistance of just
over an ohm from Vdd to ground on both of the drivers. No smoke this
time, due to the reduced current, but this shows that it is a power-to-
ground short which causes the explosive failures.
> I'm interested in how you made a 600
> KHz 1:1.5 gate drive transformer for 4 large die FETs ?
> If it were me building it I'd use four twisted pairs around a u ~ 3000
> ferrite and take one wire from each pair and parallel them to make the
> primary. The other wire form each of the pairs goes directly to the
> MOSFETs. Its much easier to make a low leakage 1:1 transformer than
> anything else.
The core is CWS Bytemark material 77 (u=2000). The windings are 28awg
kynar insulated wire-wrap wire. Four wires were twisted together and
wound as the seondaries. Four more wires were similarly twisted and
wound interleaving with the first to form the primary, skipping across a
few turns occasionally so that both windings cover the whole toroid.
All four primary wires are in parallel. Your idea is better.
> You probably need something more clever like a phase locked VCO to
> control the bridge. That way you can guarantee 50% duty cycle and
> frequency can never accidentally get out of range to cause everything
> going up in smoke but rather just quietly oscillates out of tune.
Yes. I had an idea last night. I think the way to do this will be to
use a VCO to control the primary switching. The VCO will be controlled
by a circuit that senses primary current (averaged over 0.1 seconds or
so). It will vary the drive frequency over a range of about 10% to
maximize current. This will correspond to the resonant frequency, where
input impedance is minimized.
Not only do I speak for my company, I AM my company!
Live free or die!
Go Back To The Cyber-Spy.Com
Usenet Web Archive Index Of
The sci.electronics.design Newsgroup