LNK306 3.3V buck low output voltage
I designed a non-isolated 120V / 240V AC in to 3.3V out buck supply based on the indirect-feedback reference designs in AN-37 and the datasheet and using the PIXIs designer tool. With a load of about 100 mA I'm getting only 2.8V on the output. However, the voltage across the feedback cap is good, at 3.28V. Using a 1.87K feedback resistor.
I was using a 470 uH RFB0807 output inductor, but switched it to an MSS1038 and that made no difference.
I also replaced the feedback diode with the same one I'm using for the freewheeling diode so Vf would be equal. It has a 35ns recovery time (Comchip CSFMT108-HF). This also made no difference.
I looked at the output on a scope and the LNK306 is not stuck in restart mode. I tried adding a soft-start cap of various values anyway and it didn't help.
I have a 10uF feedback cap and 68uF output cap. Total load capacitance is somewhere around 130uF.
I don't know what else to try. Any ideas?
Thanks.
Comments
I changed the feedback circuit to the optocoupler design in AN-37 using an LTV-817s optocoupler and a 2.4V zener with a 100 ohm resistor in parallel with the LED and a 10K from FB to GND. I'm now getting 3.3V with light load and dropping down to 3.1V at 300mA load; drops slowly as components heat up.
You guys might consider putting a note in AN-37 that the +/- 10% spec is not valid at low output voltages with direct feedback, and to use optocoupler feedback for low output voltages. I've seen several other posts here from people experiencing the same issue.
Hi,
Thanks for your feedback. We will revisit AN-37 regarding this issue. Meanwhile Could you please send your schematics.
Regards,
PI-NANO
I've attached both revs - rev A did not regulate well. Rev B works quite well (within about 5%). I've since switched out L2 to a 330uH to meet space and heat constraints without ill-effects. RF1 was increased to reduce inrush current. The 2.0V zener was a better match for Vf of the opto that I ended up using in the final design.
Thanks.
Hi,
Since the output voltage is very low, the output voltage tracking diodes are contributing to this error in case of RevA. We appreciate your feedback. we will consider it in revising application note. How ever if you trying to use this 3.3V for powering up FPGA's or DSP's its good idea to convert 230V AC to 12V or 5V DC and then use a linear regulator. So that you can have stiff supply for the powering devices.
Regards,
PI-NANO
Thanks for your help.
On other designs, I've done just that - LinkSwitch to 12V and then a second switcher or LDO to 5V or 3.3V. It does work very well and is also good when multiple voltage outputs are needed. This particular design was very space-constrained, which is why I went directly to my needed voltage. The 3.3V is powering a 32-bit micro and several other devices and seems to handle those loads quite well.
I just posted another question in the low power forum about a very low power, isolated design (universal input to 5V, 40mA).
Hi,
I hope that 32-bit micro controller has seperate 2.5V VRef and not the 3.3V for ADC's. If you are using 3.3V itself as a reference voltage for ADC's, then you might get significant error if you are monitoring any voltages or currents. Please make sure of that one. We will answer you the other question.
Regards,
PI-NANO
Not doing any ADC functions with this design, but that is a good thing to know for future designs. Thanks.
Hi Mathew,
Could you please share your schematic here? I can see you have said they are attached but there is nothing attached to your comments here.
Hi,
Could you please share the final schematic that Mathew has provided you in here?
Hi fairfix,
Since the schematics belongs to another customer, we are not allowed to share the schematics to anyone outside the company.
If you any questions which are specific to your design, let us know so that we resolve the issue.
Regards,
PI-NANO
I am just designing a power supply using your chip. I need 3.3v 300~400mA output.
I will eventually commercialize this design so I need to have as less component as possible to keep the price down. I also have limited space so it must be as small as possible.
Any suggestins?
Hi,
Please take a look at our LNKTN-2 family of devices.
https://ac-dc.power.com/products/linkswitch-family/linkswitch-tn2/
If you need output current is >360mA, then I would suggest you to use a tapped buck configuartion as shown in following example.
https://ac-dc.power.com/sites/default/files/PDFFiles/der574.pdf
If the output current is <360mA, then you can use regular buck configuration.
Regards,
PI-NANO
Read on this post. After reading the whole post, we can get solutions to deal with the trouble quickly and safely. Thanks for your nice sharing.
If I measure the voltages across the feedback diode and the freewheeling diode, they are different by the same amount that my output voltage is lower than the voltage across the feedback cap. In other words, I measure 3.3V across the feedback cap and 2.9V across my output cap = 0.4V difference. I measure 3.4V across the feedback diode and 3.0V across the freewheeling diode = 0.4V difference. Again, this is using the same part for each diode.