MC34063A DC/DC Switch Mode Inverting Noise

[joegtp]

I've bread-boarded a MC34063A in an inverting configuration and used the component values calculated from http://www.nomad.ee/micros/mc34063a/ using 12V input, -12V output, 100mA output, 10mV, 80kHz frequency. Is it normal to see lots of noise on the output of the converter (see the attached image)? I'm seeing roughly 400mV noise at a frequency of 50kHz. I measured the TC pin (pin 3) and it look to be oscillating at roughly 80kHz. Not sure where the 50kHz is coming from.

[kxenos]

What kind of coil and what type of diode did you use?

[joegtp]

I used a 220uF Ferrite Core Radial type inductor (Digikey RLB9012-221KL-ND) and a Schottky diode (Digikey VS-1N5819IR-ND).

I also looked at the input voltage from my bench power supply and it looks relatively clean.

[amspire]

I've bread-boarded a MC34063A in an inverting configuration and used the component values calculated from http://www.nomad.ee/micros/mc34063a/ using 12V input, -12V output, 100mA output, 10mV, 80kHz frequency. Is it normal to see lots of noise on the output of the converter (see the attached image)? I'm seeing roughly 400mV noise at a frequency of 50kHz. I measured the TC pin (pin 3) and it look to be oscillating at roughly 80kHz. Not sure where the 50kHz is coming from.

It is common to see noise when protoyping a switching circuit. In many cases, the noise you are seeing is not real noise. It may be noise entering the scope via the earth lead, or back through mains earth. How are you earthing the scope probe ground? If you attach the scope probe tip to the circuit ground, do you see the same noise?

Or it might be real noise. The way you organize the ground plane is important. The parts you choose:- The diode turn on characteristics. The ability for your filter capacitance to handle high frequency transients - are you using an electrolytic or a multilayer ceramic capacitor on the output?

If you have just tested the circuit on a breadboard, then the layout will almost certainly cause noise.

[amspire]

By the way, I can't see any 50kHz in your waveforms - looks more like 26kHz. That could be a switching frequency of a bit over 80kHz, but it is only actually switching one pulse in three. Is this with a very light load?

[joegtp]

By the way, I can't see any 50kHz in your waveforms - looks more like 26kHz. That could be a switching frequency of a bit over 80kHz, but it is only actually switching one pulse in three. Is this with a very light load?

Oops, yes you are right 26kHz. Yes this is with a light load. I'm not as concerned with the noise frequency but more concerned with the amplitude.

[joegtp]

I've bread-boarded a MC34063A in an inverting configuration and used the component values calculated from http://www.nomad.ee/micros/mc34063a/ using 12V input, -12V output, 100mA output, 10mV, 80kHz frequency. Is it normal to see lots of noise on the output of the converter (see the attached image)? I'm seeing roughly 400mV noise at a frequency of 50kHz. I measured the TC pin (pin 3) and it look to be oscillating at roughly 80kHz. Not sure where the 50kHz is coming from.

It is common to see noise when protoyping a switching circuit. In many cases, the noise you are seeing is not real noise. It may be noise entering the scope via the earth lead, or back through mains earth. How are you earthing the scope probe ground? If you attach the scope probe tip to the circuit ground, do you see the same noise?

Or it might be real noise. The way you organize the ground plane is important. The parts you choose:- The diode turn on characteristics. The ability for your filter capacitance to handle high frequency transients - are you using an electrolytic or a multilayer ceramic capacitor on the output?

If you have just tested the circuit on a breadboard, then the layout will almost certainly cause noise.

Ground and input voltage both look clean. I'm also grounding the probe as close to the output as I can.

I'm using a electrolytic on the output. I also tried adding a couple different values of ceramics in parallel with the electrolytic to see it would smooth things out.

So far I've just breadboarded it up. Wanted to make sure I had a clean output before I went to PCB.

[c4757p]

I suspect it will perform much better with a properly designed PCB. If it's still a problem, though, you can do a pretty good job of filtering out some of that noise with a beefy ferrite bead or choke.

[kxenos]

I had a look in the datasheet and it states ripple of 400mV or more. Placing an LC filter in the output makes ripple 40mV. Have a look there.

[amspire]

That noise is probably caused by a small amount of circuit inductance in series with the electrolytic or it could be the inductance of the electrolytic itself.

I would add a 1uF to 10uF ceramic cap in parallel with the electrolytic on the board.  With a proper PCB, it should be less noise. How low do you need the noise? As c4757p suggested, you may need some extra filtering after the electrolytic cap.

[joegtp]

I had a look in the datasheet and it states ripple of 400mV or more. Placing an LC filter in the output makes ripple 40mV. Have a look there.

Didn't see the 400mV ripple figure you mentioned in the STMicro datasheet also no mention of adding an LC filter. Ended up looking at the onsemi and ti datasheets and they both mention the high ripple noise and adding the LC filter.

Now that I got the LC filter in I'm getting much better numbers. I got the noise down to around 120mVpp (didn't quite have the values in my bin to make a better filter).

[joegtp]

Overall it looks like I might be better off going with a better DC/DC converter with a lower noise. 120mVpp isn't bad but I would love to get under 20mVpp.

Thanks for all your help.

[c4757p]

"Better" DC/DC or no, it's going to take a lot of work to get such low ripple with a switcher.

[joegtp]

"Better" DC/DC or no, it's going to take a lot of work to get such low ripple with a switcher.

Found an LT part (LT3462AES6) that claims 1mVpp. Still need to do more research and testing.

[c4757p]

OK, let me rephrase - "lot of work or money". Did you see the price?

[Harvs]

Overall it looks like I might be better off going with a better DC/DC converter with a lower noise. 120mVpp isn't bad but I would love to get under 20mVpp.

Thanks for all your help.

Hang on a tick.  Your ripple isn't directly related to the part, if that's what you're trying to claim here.  The MC34063 is just another controller with a few extras like an integrated power transistor.

It comes down to your choice of parts and your PCB layout that will determine the end noise levels.  You can buy the most expensive controller off the shelf and still get awful performance.

There's a really nice and practical book all about these sorts of things.  And it's actually quite an entertaining read because the author tells lots of real world industry stories along the way.  I highly recommend it.
http://www.amazon.com/Troubleshooting-Switching-Power-Converters-Hands/dp/0750684216

[codeboy2k]

Watch Dave's video #110 , he did a boost converter with that exact same chip, got similar ripple (i think it was 200mV or so) and he calculated for 100Khz, but its only 50khz at light load.  It didn't actually make 100khz until it was loaded up.  And in that video he also adds the output LC filter and reduces the ripple to 20mV.

http://www.eevblog.com/2010/09/10/eevblog-110-lets-design-a-dc-to-dc-switchmode-converter/

[Precisiontools]

There's a really nice and practical book all about these sorts of things.  And it's actually quite an entertaining read because the author tells lots of real world industry stories along the way.  I highly recommend it.
http://www.amazon.com/Troubleshooting-Switching-Power-Converters-Hands/dp/0750684216

That was a magnificent recommendation! Just what I was after and is one of the few technical books that I could read cover to cover rather than just using the index to reference certain topics so the rest wouldn't send me to sleep! Some of the anecdotes had me laughing out aloud!

This is staying in my technical/reference library for sure.

Thanks again.