LNK306 converter will not regulate

Posted by: Malcolm_OSX on 11/24/2009

Hello,

I have build an offline buck converter around an LNK306 regulator. I have used the PI design expert and build an exact copy of the typical application circuit found in the datasheets.

It does manage to get to the right output voltage and it looks like it does regulate itself. The device seems to switch it's internal MOSFET but not on the specified 66KHz frequency but on a 2,5KHz frequency causing a high pitched fairly loud audible noise to be heard. This frequency is dependant on the load current and the input voltage.
Could this relatively low frequency be caused by the large amount of cycles skipped to maintain output regulation at these low loads? So the main question here is: is this noise normal?

It should output a voltage of around 42V. To meet this voltage the feedback circuit consists of a 2K bias resistor and a 47K feedback resistor. I use a 10K load resistor(4mA). I have used a 470nF feedback capacitor.
This is what i make of the feedback circuit:
It stacks the dc output voltage on top of the source pin voltage, which is seen as ground by the LNK306 itself, using the diode and Fb capacitor. The Rbias and Rfeedback resistors divide this voltage to 1,65V on the feedback pin. Is this correct?
Another thing i do not understand is why the feedback capacitor is advised to be 10uF. Isn't this an absurdly large value? Making the regulation response very slow?

Any advise is welcome.
Thanks in advance!

Jeroen

Comments

I'm sorry for heaving bothered you. I found the solution for the instability myself. A larger output capacitor did the trick.

Just one question still remains, why does the feedback capacitor need to be 10uF?

You are correct that the low frequency is due to the large number of skipped cycles. The noise is usually caused by the output inductor. A varnished or potted inductor will reduce the noise. A possible alternative is to use a smaller LinkSwitch TN device. Using a LNK304 (or possibly even a LNK302) will reduce the number of skipped cycles and help keep the operating frequency above the audible range. You will need to check that the smaller device will deliver the required output power. If you are at 2.5 kHz at full load, then the smaller device will be able to deliver the power.

Your feedback evaluation is correct. The output voltage is represented by the voltage on the 10 uF feedback capacitor. The large value is required to hold the voltage between switching cycles. In your case the frequency is 2.5 kHz. A smaller capacitor will allow the voltage to droop too much and cause a regulation error at low switching rates (low load currents).

Z. Cochrane

Thank you very much for your helpfull reply. I will be using the converter for larger loads so i will be needing the LNK306 device. I have switched to optocoupler feedback and this now also works like a charm.
The load for the converter will be about 90% variable so i think i will need to get used to some audible noise at the light loads, or switch to a varnished inductor.
(Or wait around for about 25 years to let my hearing go down ;-))

Thanks again, the forum proved to be a helpfull tool!

Jeroen