2
\$\begingroup\$

UPDATE: Just found some weird behavior. When measuring VFB on the resistor pads with a voltmeter, I shorted VFB to ENABLE (4.36V) with the voltmeter probe. This caused the buck to start correctly outputting 12.36V. Connecting the voltmeter probes to GND and VFB again caused VOUT to go back down to tens of mV. This behavior consistently repeats when I do this.

Datasheets: Buck regulator: AOZ2261NQI-12 Inductor: PQ2614BLA-150K

I am using the buck regulator to convert 24V down to 12V. However, I am seeing weird behavior where VFB is much lower than expected. The datasheet states it should be 0.6V +/- 6mV, but I measure ~18.5mV on mine. The voltage divider circuit is functioning correctly, I measure VOUT as 390mV which is the correct magnification of the feedback voltage. Below are my schematic and board layout. This is my first time designing a buck regulator circuit so I do not really know where to begin debugging this. Any insights for which voltage/current levels to monitor or R/C values to mess with would be greatly appreciated!

I do notice that the maximum feedback input bias current is given as 200nA on page 5 of the data sheet. Of course the rest of that current would be travelling through the 10kOhm resistor to GND, but how exactly does the feedback input bias current play into that? Could that be limiting my VOUT level?

More info:

VIN measures at 23.54V. VVCC measures at 4.99V. VEN measures at 4.15V. AGND and PGND have a 0.0V offset voltage. COUT is actually 47uF, not 88uF. I realized after I ordered the PCBs that it needed to be closer to 22uF but couldn't find any in the right package. 47uF should still be fine.

The first time I powered on my circuit, VFB measured at 4.5mV. I applied a little more solder paste around the edges of the IC and reflowed again. Now VFB is 18.5mV. So possibly it is a soldering problem? The IC comes in a QFN-23L package, so the leads are pretty inaccessible with the tools I have. I have verified that none of the pins are shorted with each other, but it is possible one or multiple have an insufficient connection to the pad. My next step is to solder up a second circuit on another PCB and see if I get similar behavior.

I can post my calculations for frequency, inductor ripple, etc. if that would be helpful.

PCB layout

Schematic

Here are some waveform captures:

https://i.sstatic.net/SO5jZ.jpg

steady-state waveforms

enter image description here

waveforms at power-on

\$\endgroup\$
14
  • \$\begingroup\$ What do the waveforms look like? \$\endgroup\$ Feb 2 at 17:03
  • \$\begingroup\$ FYI, the edit to add waveform images failed to do so. \$\endgroup\$ Feb 2 at 18:41
  • \$\begingroup\$ @MathKeepsMeBusy thanks, classic haha. I'm working on it. \$\endgroup\$
    – InBedded16
    Feb 2 at 18:44
  • \$\begingroup\$ @TimWilliams are there other values I can grab data of that would be useful? \$\endgroup\$
    – InBedded16
    Feb 2 at 20:06
  • 2
    \$\begingroup\$ As far as measurements, all of them would be good. It looks like it's shorted, or not starting up, so something with the startup or fault circuitry would be of interest. \$\endgroup\$ Feb 5 at 16:38

1 Answer 1

1
\$\begingroup\$

So possibly it is a soldering problem?

If you aren't getting output, check the startup. Look at Vin Vout and the switching gate if it's not internal.

If you are worried about solder on many QFN's they will have a very small piece of metal on the top OR you can hit the top of the exposed pad and see if you are actually making a connection. You'll need probes with needle tips to do that and a DMM with continuity mode.

But really it would be good to have a sim or eval board of the buck and be able to compare the voltages and startup sequence. Make sure the components are the same between them. If you changed anything from the reference design I would make your design like the reference design and then change component values.

\$\endgroup\$
3
  • \$\begingroup\$ Interesting,. might look into ordering some of those as we move into more QFN packages. Updated my question with this info, but just found some pretty weird behavior when shorting Vfb to Ven. I'll look into what the startup behavior is telling me as well, thank you. \$\endgroup\$
    – InBedded16
    Feb 5 at 18:14
  • \$\begingroup\$ Your edit is a really solid idea too - I had poked around for sample circuits but hadn't thought to actually grab an off the shelf eval board for testing. Definitely adding that to the circuit prototyping procedure from now on. \$\endgroup\$
    – InBedded16
    Feb 5 at 20:26
  • \$\begingroup\$ It's easier to have a dev board or something that you know works when DC/DC's don't work. There are a lot of moving pieces and you need to cross the things that are working off the list. \$\endgroup\$
    – Voltage Spike
    Feb 5 at 20:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.