TL494 SSTC overheating / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

TL494 SSTC overheating

Move Thread

LAN_403

neuralinhibitor

Sun Jun 28 2009, 06:37PM

Registered Member #2023 Joined: Fri Mar 13 2009, 05:53AM
Location:
Posts: 7

Hello,

I recently made a desktop sized SSTC based on the classic TL494 oscillator design. The TL494 outputs are sent to 2xUCC37322 chips which drive a GDT connected to a simple half bridge. I was getting some FET overheating at really low power levels (running longer than 1 minute unfiltered half-wave rectified caused the FETs to get hot enough to burn my finger) and wondered why so I hooked up my scope and this is what I got.

The top picture is a scope shot of the output from the TL494. The middle shot is the same thing zoomed in. The bottom shot is the signal at the MOSFET gates. There are some weird voltage reversals near the 0-crossing and the switching transition looks like it takes too long.

Am I being too picky about the waveforms (should I look elsewhere for causes of overheating)?

uzzors2k

Sun Jun 28 2009, 09:25PM

Registered Member #95 Joined: Thu Feb 09 2006, 04:57PM
Location: Norway
Posts: 1308

The waveforms don't look terrible, perfectly normal IMO. When using a fixed oscillating frequency I would be more worried about being out of tune. The worst you can do to the fets is run the coil slightly above or below the resonant frequency. Also you haven't mentioned what kind of cooling you use, mosfet model, or power through-put. I need a largish heatsink and cooling fan for each mosfet in my PLL SSTC to permit continuous operation, so make sure yours is adequate. My coil draws 1.5 or 2kW IIRC.

GeordieBoy

Mon Jun 29 2009, 12:59PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

The waveforms looks fine. Although it helps to also post the x and y scales and offsets for the traces!!! For instance, was the top trace showing "signal at TL494 output/UCC input" really going negative?!?!?

We can also deduce the switching frequency from these figures, in case you forget to tell us

What power level were you running at? ...and what heatsinking / cooling arrangments have you made for the MOSFETs?

-Richie,

Dr. Dark Current

Mon Jun 29 2009, 01:37PM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

What is the supply voltage for the gate drivers? I found that going from 12V to 15/18V sometimes makes the FETs run cooler.

Something I can think of
-Cross-conduction, do the FETs heat up with the primary unconnected?
-Too low primary inductance
-Bad tuning

neuralinhibitor

Mon Jun 29 2009, 07:52PM

Registered Member #2023 Joined: Fri Mar 13 2009, 05:53AM
Location:
Posts: 7

The previous scope shots were distorted--I found that channel 1 of my "ebay surprise" oscilloscope is toast (the calibration signal looks like a sawtooth instead of a square wave only on this channel).

I've attached some new ones even though they don't look terribly different (but have axis scales). As you can see, I'm running at about 240kHz

. The negative voltage on the TL494 before was because I was measuring across the outputs of the TL494. In the new shot I'm measuring from ground to one TL494 output.

I already tested for cross conduction in the manner suggested and that isn't the problem.
I'll try some different primary configurations to see if I can find something with better coupling using more turns, maybe adding a bigger toroid to try to get the resonant frequency down below 200kHz.

The heatsink is from a PII CPU and should be fine since I run my half bridge DRSSTC at >1 kW on an identical heatsink and it gets only slightly warm.

The "MOSFET" is actually an IGBT but it's one of the faster hybrid types.

Power draw is about 500W right now and things get uncomfortably toasty after only 1 minute of use

.

Off to do some more tinkering.

GeordieBoy

Mon Jun 29 2009, 09:12PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

If the system is running CW, then the operating frequency may be too high for your unspecified IGBT.

The new gate waveforms also show very shallow slopes. This low dv/dt will cause excessive switching losses if the devices are switching any significant current at this point. Decrease the gate damping resistors to speed up the rate of rise on the gates of the power devices.

-Richie,

neuralinhibitor

Wed Jul 01 2009, 08:14PM

Registered Member #2023 Joined: Fri Mar 13 2009, 05:53AM
Location:
Posts: 7

I don't think the heating is caused by the IGBT being too slow since it is rated for about 15A at 200kHz. The datasheet is linked in my last post if you want to take a look. More likely the problem is, like you suggested, the long, sloping switching transitions.

After adding primary turns and testing larger toploads to lower the resonant frequency with similar heating problems, I am increasingly certain this is true.

The problem is I just can't seem to get a really solid square wave out of the TL494--at anything more than 100kHz I start getting droopy signals and the shallow slopes you noted show up.

Any ideas?

Random notes:
the ucc voltage supply is 12V
the R220 pull-up resistors on the TL494 output get pretty warm
operation is CW (I'm assuming this means not interrupted?)

GeordieBoy

Wed Jul 01 2009, 10:47PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

As far as I know the TL-494 was never intended for push-pull operation at 240kHz. My datasheet here has the maximum ramp oscillator frequency listed as 300kHz, which would make the max output frequency 150kHz after passing through it's internal D-type flip flop. The open collector outputs also require a passive pull-up or pull-down resistor which means that one edge is always going to be significantly slower than the other.

This chip was invented in the days when 50kHz was a typical switched-mode power supply operating frequency. You might have more luck with a more recent driver like the UC3525 or even more up-to-date offerings from TI.

-Richie,

LithiumLord

Thu Jul 02 2009, 12:45PM

Registered Member #1739 Joined: Fri Oct 03 2008, 10:05AM
Location: Moscow, Russia
Posts: 261

Well, in fact I used to run my coil off 494 a long time ago, before moving onto own deadtime generator. As the driver, I used four UCCs rigged straight to the gates of four IRFP260s forming an H-bridge (the GDT was used as signal xformer). This construct worked off approx. 170vdc draining 30A from the input of the voltage doubler (I used a stepdown decoupling xfmr and voltage doubler on it's output). No major heating, although I used HCMC494 which features a lower deadtime.

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob