Re: BEWARE of dead gate drive IC that still works

["Tesla list" <tesla-at-pupman-dot-com>]

Original poster: "Malcolm Watts by way of Terry Fritz <twftesla-at-qwest-dot-net>" <m.j.watts-at-massey.ac.nz>

Hi Justin,

On 18 Feb 2002, at 1:37, Tesla list wrote:

> Original poster: "Justin Hays by way of Terry Fritz <twftesla-at-qwest-dot-net>"
<pyrotrons2000-at-yahoo-dot-com>
> 
> Hi All. 
> 
> (this post is SSTC/SMPS related, and is longwinded)
> 
> Maybe not just Maxim. Possibly other manufacturers also, but I've
> seen this a LOT with Maxim specifically.
> 
> I had a circuit set up as follows:
> 
> PWM controller (NTE1729) that puts out a 100kHz to 800khz square wave
> that is 12V. This was voltage divided with 2 resistors to get 5V, and
> fed straight into the input pin of a MAX4420 gate drive IC. The
> MAX4420 and MAX4429 IC's are simple but very powerful. They put out a
> high-current high voltage output that will drive big power MOSFET's
> directly, with logical inputs. (a typical gate drive IC, but read
> futher)
> 
> Maxim's awesome IC:
> 
> The datasheet says it will slam 18 volts into 2500pF (MOSFET gate) in
> 20nS! And claims the output impedance is 1.5 ohms! I got datasheet
> performance easily even on a breadboard, it is a wonderful device but
> unfortunately, like the other chips, the output is ground referenced.
> So no good for half or H-bridge SSTC's unless you do something about
> that (opto, pulse xfmr, PVI, etc.) 
> 
> The Supertex HT0440 IC has ISOLATED OUTPUTS to 400 volts!! I/we need
> to play with that IMHO. I think it works to 1Mhz.
> 
> To my main point:
> 
> Say like, for instance, just speculating, this didn't really happen
> to me now ; ), that your gate isolation transformer kicked back into
> the chip and fried it. Oops. I swear some 12V zener diodes literally
> jumped out of my parts drawers and soldered themselves onto my board
> after that happened ; )
> 
> Failure mode:
> 
> The failure mode is typical I've heard, but strange nonetheless. It
> seems the output stage of my chip failed......the sucker still works,
> but it won't put out any current! In other words, when I take the
> MOSFET load off of the output of the chip, the chip's output looks
> like a beautiful 20nS risetime, 50nS falltime square wave (along with
> a little ringing). But as soon as the MOSFET is back in, the output
> waveform looks like a triangle wave riding on 12VDC ; ) Loads it down
> like a ______.
> 
> This would lead the user to believe that the chip doesn't have enough
> output current, when actually, the chip is fried.
> 
> After destructively testing several more of the IC's (which included
> MAX4420's and MAX4429's), I determined that the chip fails when the
> steady-state supply to the primary of the isolation transformer is
> disturbed. This is when:
> 
> 1). The chip's supply voltage is applied.
> 2). The chip's supply voltage is removed.
> 3). Connection to the gate drive transformer is disturbed. (bump it)
> 
> If the chip fails upon power-up, it will look like the chip doesn't
> have enough juice to drive your isolation transfomer. Thats what got
> me the first time, so beware.

My reading of that is that the chip wasn't designed to operate into a 
low shunt inductance. What was the inductance of the transformer 
primary? In my experience, gate driving chips are only designed to 
operate directly into capacitive loads. All the SMPS supplies I've 
dealt with that had driver transformers had discrete devices dring 
the primary. 

Regards,
Malcolm
 
> Obviously, applying Vcc and removing it is a must. So something has
> to be added to protect the chip. I'm thinking a couple of
> back-to-back zeners. Other methods slow down the rise and fall times
> of the output waveform (which is ultra-critical 'cause it controls my
> H-bridge running at 1.5kW!) 
> 
> I'm looking for a maximum 100nS for rise and fall times. To keep the
> MOSFETs biased out of their linear regions as much as possible. To
> increase efficiency and make the heatsinks smaller ; ) (even though,
> in SMPS terms, typical SSTC H-bridges operate as ZVS (zero voltage
> switched) power supplies, where the current/voltage at switching time
> is zero.) But.......that's only when it's perfectly tuned........so
> unless you've got a PLL going on you're hot!
> 
> Didn't mean to ramble so much. 
> 
> Anyway, thoughts?
> 
> Justin Hays
> KC5PNP
> Email: justin-at-hvguy-dot-com
> Website: www.hvguy-dot-com