# SMPS voltage ripple

I'm using the MC34063 for my 35 VDC -> 4VDC SMPS. Here is the circuit (modelled in LTSpice):

Edit: However, using a MC34063 calculator and setting Ipeak = 0.5A,

• L1 = 37 uH
• C3 = 12.5 uF

Which, plugging those values into LTSpice, results in a 40V transient.

Here is the output waveform with a C3 MLCC and no load (I know, probably bad idea but I'll explain a little more later):

And here is the output waveform with a C3 electrolytic capacitor and no load:

So, the Vpp values with the MLCC is 376 mV and the electrolytic is 120 mV. Is this because the electrolytic capacitor is serving as a load (since it has a higher resistance than an MLCC)? Or is there something else at play?

Regarding why these experiments are done with no load, when adding a 10 ohm high wattage resistor, the MC34063 burns out. This does not make sense in the steady state since the MC34063 can handle up to 1.5A. In the transient state, LTSpice shows this:

Which is about 1.3A (for <1 ms) but still less than the maximum current. Any other ideas why the chip is being fried?

• What are your calculations for C3 and L1? C3 seems a bit low. – scld May 13 '16 at 2:04
• C3 = 12.5 uF, L1 = 37 uH but those resulted in a super high voltage transient in LTSpice, which destroyed a few ICs as well :| – rith87 May 13 '16 at 2:15

This circuit uses a diode as the low-side switch. This means it is incapable of handling negative inductor current, and to it operates in a pulse skipping mode -- notice tour switching frequency is around 20 Hz. This also causes large ripple.

Don't just use a 10 ohm R to test -- try 100 ohm. Also don't start at 36 V -- the 10 V.

Without further info it is hard to know why it is failing, but perhaps you built this on a plugboard and the decoupling capacitor on VCC, or the diode d1 are too far from the IC ? This would cause inductive spikes which could damage the IC.

• So, actually, the SMPS is supposed to operate at 100 kHz. The reason why it shows 20 Hz is because there is no load, so the inductor is unable to discharge fast enough. I guess you're right, the next thing to try is to see what happens with a 100 ohm resistor.. – rith87 May 13 '16 at 2:18
• FYI, this circuit was built on a 2 layer PCB but I'm not too sure if the layout is up to scratch. – rith87 May 13 '16 at 2:20
• A 1N5822 would offer a bit more headroom. It is rated at 50 vdc. The kick-back from L1 with a 36 volt pulse could over voltage a 1N5819. These kind of circuits are not meant for plug-boards or random layouts as the ground ESR is/can be very high. – Sparky256 May 13 '16 at 4:51

The reason for the difference in voltage ripple is caused by the DC bias on the MLCC. The (Murata) capacitor I was using had the following DC bias profile:

Normalizing for the capacitor chemistries, this meant that I was actually comparing a 100 uF capacitor and a 40 uF one. Given the same current draw, it was expected that the voltage ripple on the MLCC was a lot larger.