hexfet mosfet get hot too fast

If you have a scope, take a look at the signal which is being applied to the mosfet gate. If you are not getting a high enough peak voltage or if it takes too long to transition from peak to ground (or vise versa), you would get the symptoms you describe. You should be getting to 10 volts for this mosfet. If you need a logic level device (5 volt gate drive) use the IRLZ34N.

Gary summed it up nicely. In addition, take a really close look at what happens immediately after the gate was turned on. Crank up the scope's time base to 50nsec/div or less. Does the voltage collapse right away and come back or, worse, bounce a few times? That would be caused by the Cgd capacitance and can lead to a lot of grief, mainly to losses and heat. If this is the case or if you see too shallow slope you need a harder driving source.

I haven't used the TL494 but IIRC it's rise times aren't much to write home about. I am used to 10nsec or less and I believe the TL494 is quite far from there. OTOH 20kHz ain't that much either so maybe that would be ok.

Regards, Joerg

I altered a TL494 gate drive circuit last week.

The IC ran from a 15V supply, and had a level-shifted gatedrive. The TL494 collectors were tied to +15V, the emitters went into a 1k resistor then into the base of a C-E NPN, along with a 1k B-E resistor.

in theory, it works fine - the datasheet says 200mA.

In practice, I got a narrow 9V spike, the internal transistors pulled out of saturation dropped down to 1V or so, tapering towards Vbe.

Using a 10k, rather than 1k, output resistor gave me a lovely 15V square wave. POS :)

Cheers Terry

How are you driving the mosfet ? You can't just substitute one in place of a bipolar arrangement.

Graham

Hello Terry,

Well, it's not a hot rod but you can buy these for a little over 15c in modest quantities. There ain't much else out there this cheap in the world of PWM chips. Which is why quite a few of my switcher designs don't use any PWM chip :-)

The good news about the TL494 is that, unlike many other PWM chips, it is not single sourced. You get a modest chip at low cost, meaning you can do better but probably not at 15c a pop unless you do a discrete design from scratch. After all, you also don't win a Formula One race in a Vauxhall Astra yet it's still a great and practical car.

Regards, Joerg

Sorry, guess it's the Holden Astra in your case :-)

Regards, Joerg

IIRC the TL494 has a single-ended output. Not very usefull for driving mosfets. Even an NE555 with some current and overvoltage circuitry will do better.

--
Reply to nico@nctdevpuntnl (punt=.)
Bedrijven en winkels vindt U op www.adresboekje.nl

The combination of the TL494 and an N-channel MOSFET seems guaranteed to end up with a whole whack of additional parts to level-shift and drive the gate.

I wonder what kind of inductor the OP is using. If it is too small or saturating the result won't be much better than a linear regulator.

Best regards, Spehro Pefhany

--
"it\'s the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com

Hello Nico,

It has two outputs and can be configured push-pull. It contains double pulsing inhibit and even comes with dead-time control. So yes, you can drive FETs with it, as long as you remember that it is an old design and not a young race horse.

Regards, Joerg

That would explain an awful lot !

Graham

Use another chip seems sensible.

I had an efficiency problem once with something like that. On reflection I reckon maybe the inductor may have been wound on a lossy ferrite. Trouble is, Farnell didn't supply much info about magnetics back then ( or maybe now too even ).

Graham

The relevant comment I suspect being *can* . We have no idea what the OP has done.

Graham

Hello Graham,

That is why posting a schematic would be so very helpful for discussions like this.

Regards, Joerg

Hello Graham,

reckon

You can measure ferrite pretty well. The equivalent series resistance and Q tell about the "usual" losses. That's why I was in a hurry to repair the HP4191A, to avoid having to kludge all that.

After that it's on to the DC test. Add a 2nd winding and send DC through it, then look where the inductance begins to drop. I usually do that differently, applying a ramp and looking where the slope begins the shark fin. Typically wearing eye protection in case I make a mistake and the FET turns into a spacecraft.

Regards, Joerg

I do both simultaneously, on a probably indestructible rig.

The IRFZ34 in use seems to bottom out at 80A peak or so, minus DUT resistance.. I haven't run it short circuited yet. The flyback/power resistor dissipates more power (when biased to 15V), on account of running the transistor at a duty cycle in the 5 or 10% range.

Tested a whole bunch of stuff lately, seems those yellow and white-one-side toroids used in 200-300W computer PSU's are roughly 200nH/T^2, and have eddy current loss on the order of 0.2 ohm (expressed as one turn shorted with said resistance).

Tim

--
Deep Fryer: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

As in a push-pull converter, not as in driving a mosfet properly. It will turn on the transistors in an alternating way (first one on in de even cycles, second one on in the odd cycles). Between that there is just dead time and time for the energy to transfer from the inductor (whatever) into the load.

--
Reply to nico@nctdevpuntnl (punt=.)
Bedrijven en winkels vindt U op www.adresboekje.nl

Hello Tim,

My ramp testing happens on a similar rig. Indestructable? That's what I thought, too. Until the day when the load resistor popped its thermofuse and the electrolytic complained about exceeding its rated voltage with a loud bang. White fluff was sailing down from the ceiling and I had to get trowel and spackling to fill the cavity that its can had made upon impact.

Regards, Joerg

Hello Nico,

It's been a long time but IIRC you can use them both on one drive channel. One to charge and one to discharge the gate. I don't remember how I controlled it but it was unorthodox. Pulling dead-time plus frequency via the CT node or something like that.

I ended up not using it because the TL494 consumed to much quiescent current and, believe it or not, it cost around 20c at that time and this was too much.

Regards, Joerg

its nice in that it has a pair of error amps, but the reference and oscillator are pretty sloppy. and yep, its cheap. not a bad choice in this case, I was just surprised that the output transistors couldnt drive 15V into 1k. All the units I tested were like that, so its probably not a faulty IC (although that is possible). Other than that, the level shifted 40V gatedriver works pretty well.

Cheers Terry