Viper26k buck converter has an unstable output.

[Elin1480]

We designed a buck converter based on AN5380 with an input of 600Vdc and an output of 12Vdc. 

The device turns on at around 55Vdc, but the output keeps shutting itself off and turning on again (image1). This keeps happening as we turn up the voltage. However, at around 80Vdc, the output changes to many small bursts and then a large spike up to 12Vdc (image 2). Is this only a problem that appears with lower voltages? 

Attached is the schematic of the circuit (image 3). Are the values we chose unsuitable for this application?

[Peter BENSCH]

Welcome @Elin1480, to the community!

Questions about such buck converters have often been discussed here (e.g. here or there), but the necessary conditions are often not considered enough, e.g:

• load current at the output
• coil used (L2 in your case) and its saturation current
• simulation with eDesignSuite

For example, in contrast to the AN5380 mentioned above, you have only used 47µH for the buck coil instead of 1mH, which can have a major impact. In addition, you only used 44nF as input capacitance at the input, which seems very low to me.

Regards
/Peter

[Elin1480]

Thank you for the reply.

I have attempted to use eDesignSuite to simulate the circuit. However, it seems to only work at a maximum of 70V and not at 600V(image). Is this additional circuitry needed for the ic to operate? (image2) With the simulation, I am able to get an output only if those components are added to the input. 

Additionally, are there values you would recommend for the inductor L2 and input capacitance based on a 600V input, 12V output, and a 200ma load? I have attempted using 470uH and 700nf for these two components as well as putting a load on the output. This resulted in a stable output at 65Vdc, but at anything below or above that, the waveform is similar to image2 of the original post with small bursts. 

[Peter BENSCH]

I was referring more to the simulation via the Power Supply Design Tool > AC/DC. Your 600V application can also be interpreted as AC/DC by assuming 425Vac. Of course, the filter that you have in the image2 is then not necessary, but the buffer capacitor (CIN2A/B in your case), at which the voltage is divided with the balancing resistors, is.

I quickly ran this simulation and with 90...425Vacand 0.2A on 390µH and 2x15µF I get a phase margin of 75.24°, which is just below the recommended 76°. With a slightly lower current of 0.17A, 76.06° is also achieved. Please select a variant with at least 0.8A for the coil.

However, the layout also plays a decisive role because the current spikes are very short due to the large voltage difference. So please follow the usual rules for buck regulators meticulously.