mc34063 step-down switching regulator weared Oscillation

[mj21]

hi.

few days a go i decided to make a mini switch mode power supply. as i had bunch of mc34063 lying around i decided to put them to good use. now i used the diagram in datasheet to make a 5volt step down regulator. as i understand if i put a pot between pin 5 and the ground i can adjust the voltage. so i did that and everything works fine, i mean i can adjust the voltage. I am feeding 20 volts.  i used a logic level mosfet to increase the output current. now using oscilloscope to see the output ripple its around 20mv ptp no load and about 250mv ptp under load at below 1amp and about 80mv ptp above 1amp.

when there is no load there seem to be an oscillation at the out put. looking at the input with the scope i can see the same oscillation but weak but at the same frequency. no matter how many capacitor or inductor i use this oscillation does not go away. i even used battery to power the device in case the oscillation was caused by other power supply but did not change anything.  can some one please let me know to fix this and where i have go wrong.



many thanks

[Bored@Work]

Due to lack of load the MC34063 is in a kind of discontinuous mode. In practical terms, while the MC34063 is a bad switcher in continuous mode it becomes a very bad switcher in discontinuous mode, doing all kinds of funky stuff.

[c4757p]

Yep. Unlike more sophisticated PWM converters, where the pulses come at a fixed frequency and varying duty cycle (strength), the MC34063 works generating a fixed pulse train, then selectively omitting pulses from it when the output voltage is high enough. This means that if one single pulse of the fixed duration is too much, you still get the whole thing whether you like it or not. You can try decreasing C_T to shorten the pulses (this will affect your maximum output current though), adding a small dummy load, using a different (PWM) chip, or just dealing with it.

MC34063's applications are usually limited to non-variable cases, where the load will look pretty much exactly like the load you designed it for at all times.

[mj21]

thank you very much guys for your help. seeing dave using it in his dc to dc converter i thought it be a good idea to use this chip, but it seems it not all that good. putting a load on the output just to make the dam thing stable seems to defeat the purpose of using a dc to dc converter as i am only using it to reduce power dissipation.

any suggestion on what chip is good for a dc to dc converter capable of 3 to 5 amp?

thank.

[c4757p]

It's fine for nearly constant loads - I use it all the time - or when you don't care about the hiccuping. It's not the best chip in the world, but it's cheap as hell and easy to use.

any suggestion on what chip is good for a dc to dc converter capable of 3 to 5 amp?

LM2576 does 3A and works quite well. I'm not sure about 5A but I'll have a look around.

Edit: LM2678.

[mj21]

will the mc34063 be OK to power pic or atmel chips with its oscillation? the chips don't take that much power so the oscillation might still exist. i have made a pic proto board and i used the mc34063 as a 5volt reg the board works fine but i do get glitches every now and then, could this be due to the oscillation?

[c4757p]

It should be fine. What you need to keep in mind is that the load it's used with needs to be close to the load it's designed for. If you calculate L_min to be 47 uH and then whack in a 1000 uH inductor because "it says 'minimum'", it's not going to work very well. And if you need 0.1mA and design it for 500mA "just in case", you're going to have a bad time there too. Those glitches are because of the small load - the hysteresis thresholds don't seem to be very stable, so the pulse gaps drift around and make it look like a faulty "glitch".

The microcontroller isn't really going to care about some 200mV glitches, anyway. If it's a problem, you can always put the MCU behind an RC filter. (Just don't be tempted to put the voltage sense behind the filter as well to negate the R loss, unless you like very unstable buck converters.)

Edit - I thought you meant power supply glitches, but re-reading it I think you meant MCU glitches. That I can't explain. I doubt some 200mVpp ripple is going to cause that - I would look elsewhere.

[mj21]

thank you very much for all your help guys.