Unwanted spikes on inverting converter output

[gianbo85]

Hello everybody,

I have designed an inverting dc dc converter using a MAX1847 controller (f = 300kHz):
**broken link removed**

My circuit is identical to the application circuit presented in the datasheet. With the following components:

L = 560uH
Cout = 3 x 33uF Tantalium (ESR= 10mohm each)
Cin = 2 x 47uF Ceramic
Rsense = 200mohm
Mos = Si7113 Vishay (100V -13.2A)
Diode = STPS2150 (100V - 2 A)

Plus the compensation network components...

The buck boost works and inverts a +12.5V line into a -30V one with a 1Kohm load.

The problem is that on the output I can measure some ringing. The ringing main spike is syncronized with the rising edge of the switching node (it appears every 3us when the node goes from -30 to +12.5) and is 100-150mV wide.

The ringing has a 175ns period and fades after three periods approx. and then appears again with the next switch of the mos.

I designed the converter to have very low ripple (10mV) and I cannot accept such ringing even if the main peak last for few nanoseconds...

Any idea about what can cause this?
Do you think I can filter it with an additional LC filter on the output?

Thanks

Gianluca

 

[gianbo85]

Ok,

just to help others with a problem similar to the one I am experiencing:

- I haven't found the cause of the unwanted spike: maybe a bad grounding layout, maybe radiated noise from the inductor

- I found a way to filter it out thought... I put a couple of 100nF ceramic capacitors on the trace leading from the dc/dc output to the PSU connector, around 1 cm away from the dc/dc output. In this way the trace inductance plus the added capacitance form an high frequency LC filter that greatly reduce the voltage spikes to around 15mV!

Bye

Gianluca

 

[karesz]

Hi fun,
I dont no, it seems me to be a worst resonance/effect of placements & bypassings...
k.

 

[FvM]

I haven't found the cause of the unwanted spike: maybe a bad grounding layout, maybe radiated noise from the inductor.

I would rather expect the diode reverse recovery behaviour as the main interference source. The commutating current path is usually through output capacitor, output diode, switch transistor and input capacitor. Unsuitabe design of this current path or using capacitors with too much series inductance can cause nasty oscillations.

Condidering your below intention, the circuit design must be done thorougly.

I designed the converter to have very low ripple (10mV) and I cannot accept such ringing even if the main peak last for few nanoseconds...

- Additional input and output LC (dissipative inductors like ferrite beads are good for it) filters are recommended.
- The said commutating currents shouldn't flow through the common ground plane but use a local ground isle connected at one point to the ground plane.
- The area spanned by the commutating currents should be minimized.