MC34063 Step-down Converter Heating Problem

Can you hold your finger on it for 20 seconds?

Which package are you using?

John S

Well yes, I can hold my finger on it for 20 seconds? Can I take that as a "thumb" rule? :)

Forgot to mention, the package is 8 pin PDIP.

Thanks a lot for the prompt reply.

Anand

Well yes, I can hold my finger on it for 20 seconds? Can I take that as a "thumb" rule? :)

Forgot to mention, the package is 8 pin PDIP.

Thanks a lot for the prompt reply.

Anand

Well yes, I can hold my finger on it for 20 seconds? Can I take that as a "thumb" rule? :)

Forgot to mention, the package is 8 pin PDIP.

Thanks a lot for the prompt reply.

Anand

Well yes, I can hold my finger on it for 20 seconds? Can I take that as a "thumb" rule? :)

Forgot to mention, the package is 8 pin PDIP.

Thanks a lot for the prompt reply.

Anand

I use the rule-of-thumb that if you can hold your finger on the device, then its temperature is not much more than about 140F. Of course, your finger cools the chip a little. Looks like you are safe.

John S

The most obvious way is to use a thermometer of some sort. Do you have thermocouples, RTDs, or thermistors?

Aside from actually measuring the temperature, you could measure input power, output power, and ancillary component powers and then subtract the output power and ancillary component powers from the input power to get the chip's dissipation. You can then calculate the chip's approximate temperature from its thermal resistance.

It is likely to heat under almost any loaded condition. It has no choice but to lose some power. The question is how much will it heat at your maximum load condition and is that too much?

You could try reducing the switching frequency along with increasing the inductor.

Not for your current operating condition according to the "finger test."

Cheers, John S

You asked an open-ended question. Here's an open-ended answer that I often give fledgeling power supply designers.

What's the cost of failure? I've seen people save $20 by copying a random schematic off the web, spending a week building it with not-quite-the-same parts, not testing the result competently, then plugging it into a $1000 device. Often, it works. One of the significant failure modes is that the part melts and shorts input to output. How lucky do you feel today? Got a spare $1000 device?

IF you can afford to fail, put on your boots and wade on into the power supply swamp.

Power supply design is an art. Yes, you can simulate the schematic quite accurately. The art is in simulating the ACTUAL circuit including parasitic elements that are not visible in the pile of parts.

Even the best-intentioned reference designs from parts vendors have errors. All it takes is a sleepy typesetter and an incompetent proofreader. Most of the stuff you find on the web was "designed" by people without a clue, but some luck.

The aspects of the design that you question and FIX are not the things that will cause failure. It's the things you didn't think about or test for. Since you disclosed little, we can't help with that.

For example, nasty things can happen if the inductor saturates.

Anything you intend to plug into a car electrical system requires SIGNIFICANTLY more care.

What does the system do when a peak load transient exceeds the current limit? I've seen USB hard drives go into a limit-cycle oscillation when they don't get enough peak current and thrash themselves to death.

Power supply design is "system design" and we often have no clue what's inside the load end of the system.

FWIW, I've seen that chip used in car cigarette lighter adapters. They always chose to use a heat sink glued to the chip. But, just cause you can't feel the package get hot doesn't mean that the chip temperature isn't making wild transient swings. Heat sinking a cool chip won't help that.

Are we having fun yet?

Dear Mike,

Thanks a lot for your reply. I inserted the URL so that it serves as a ready reference for readers to see the diagram of my circuit. As such, the circuit I have constructed is from the Motorola/ON/TI datasheet and the ON Semi design worksheet.

There a few things I can pick-up from your reply:

• This chip seems to be used quite a lot (like some other people tell me). I had feared that this chip is not so popular for standard products.

• This chip should probably be used with a heat sink.

There is a thing with the inductor though. I have to use higher values than that are worked out in the calculations. (I use inductors that are wire wound on a powder iron torroidal core which is what quite a few people seem to be using.)

As for the testing, I am not testing for peak/transient loads currently. I would first like to ascertain that the heating issue is addressed for steady state load (Vin: 10V, Vout: 5.3V, Iout: 250mA).

This power supply design will not plug into a $1000 design but then whatever it plugs into is worth more than that to me!

Thanks, Anand

If the temperature is acceptable without it, you don't need a heat sink. BUT You have to make sure you're testing under the conditions that cause the most heating.

What's the symptom of needing higher values? Toroidal cores come in a WIDE range of materials. Inductance goes down as flux/current goes up. If the rate of change of current goes up as a function of current, current can quickly get out of hand. If the transistor comes out of saturation, it can melt before you can shut it off. This is particularly troublesome at high voltages, like the 30V in your original spec. Google "safe operating area". In general, the more the inductance/turn, the more likely it'll saturate. You really need to look carefully at the curves and stay out of the saturation region for the exact core you're using. Picking a random core and measuring only the inductance is risky.

I don't think you'll have any problems at that operating point. It's the 30V and transient/short-circuit load conditions that may give you problems. Assumes your inductor still looks like an inductor under all operating conditions...not saturated.

Another thing newbies do is not pay attention to where the currents go. The wiring has parasitic inductance and resistance. So, connecting to different places on the same wire can have different effects on the stability/response of the system. It's very easy to get source transients or load transients coupled into the loop that controls the switch. Typically, not a serious problem on a simple buck converter, but can be disasterous on a push-pull forward converter that depends on an EXACT symmetrical drive to keep the core out of saturation. Even though your switching frequency may be only 20-100KHz., you have to use design practices for a much higher frequency.