Synchronous Buck Constant Current Source with LT3741

Hello Everyone,

I curently working on adjustable constant current source with LT3741. Design specifications are:

Input voltage: 10 V - 15 V (lead battery)
Output current: 0 A - 3,3 A
Maximum output voltage: 5 V
Switching frequency: 200 kHz

Schematic:
**broken link removed**

Parts were chosen with datasheet equations and advices:
http://cds.linear.com/docs/en/datasheet/37411fd.pdf

Without a load i got nice 5.06 V ouput over whole input range.

Voltage waveform of switching node at high and low input volatage:
**broken link removed**

Problem is when i connect load i get waveform like this with audible noise from inductor.
**broken link removed**

With 1.8 ohm load I tested from lowest input current (Vctrl1 set to minimum). Power supply is working fine till about 1A (also regulating well over input range). Over 1 A there are again oscilations like on the picture and supply dont ragulate well. Are these subharmonic oscilations? LT3741 using average current mode control and gm OPAMPs.
I used equations from datasheet to calculate R14 and C11.

I think problems can be caused by:
a) bad layout (huge noise)
b) bad compensations parts (i know i need to set the downslope of the current signal comming into PWM comparator to be lower than slope of the PWM ramp, but i dont know if i did it properly)

I tried different values for R14 and C11. Nothing changed.
Here is layout:
top
http://www.dropbox.com/s/a4hfaaefjeu9z6o/top.png
bottom
http://www.dropbox.com/s/qxu0jzsfmqe9o17/bottom.png

Anyone have experience with average current mode compensation or with LT3741? Any advices? Thank you

The irregular waveforms appear to be 'hiccup' mode. You have one cycle with an overly strong current burst, followed by one or two cycles without any current. This can create audible noise. The likely cause is lack of tight feedback.

Looking at C10 across the sense pins. The datasheet shows it used in certain applicattions. When it is absent, the load is a simple resistive load.

However C10 apparently is needed for a load which would quickly gobble all available current, until voltage drops to its threshold level (LED, supercapacitor)... unless you deliberately limit it. And you state this project is a current source.

The value of C10 could make a difference in the tightness of feedback. I suggest you try different values.

Thank you BradtheRad, i will try to change C10 tommorow.

Edit: I tried different values of C10, it didnt help... And without C10 its oscilating even without load.