How to avoid voltage rises in a buck converter ?

[ze_dib]

Buck converter problem

Hi,

I tried to use LM2676 from national to do a buck converter with those specification :

Vin from 8 to 26 V
Vout = 5 V
Iout =1 to 2 A

Seems to work fine at first sight. But I found a Voltage rise when I reduce the load. Can anyone suggest me a Way simple and cheap to avoid those rises.

Regards,

ZE_DIB

 

[FANT]

Re: Buck converter problem

How much is the rise ?
Could be done by the tracks resistance ( heavy lines on the data sheet ) and also if the feedback pin is rounded around the inductor.
Also have a look with the scope if the output signal is clar or affected by noise at low load.

Mandi

 

[ze_dib]

Re: Buck converter problem

The rise is about 2.5 V, which is to much for my application.

There is not much noise at low load.

Thank for advice.

 

[E-design]

Re: Buck converter problem

A few things in general:

As with any switcher, layout is very important, keep the enclosed area from pin1 thru choke, thru output cap back to pin4 (GND) as small as possible. It will help to reduce unwanted radiation that may couple into the feedback pin as well as other nasty side effects.

Make sure your GND connections are of a "STAR" type. This means that the GND ends of the diode(anode) and output caps should be connected to pin4 with the shortest possible connection using thick tracks. They should not share a common GND track back to pin4 as in the databook drawings. Avoid schematic type connections as they are made for clarity and not for best performance.

Avoid running the feedback pin track close and in parallel with the output choke to prevent coupling into this line.

Also use a scope with sufficient bandwidth (50-100MHz) as some of these interference pulses will be very narrow at low load, and you may need to crank up the intensity to see them. Otherwise a DSO with clitch capture mode will help a lot.

E