3

I have tried setting up this cheap AOZ1280 buck converter to give output of 3.3V from input of 3.5V to 4.2V without success. Please check my schematics below.

What happens is that I can not achieve an output of 3.3V unless the input voltage is at least 4.2V. If my input voltage is at first 4.2V and then I reduce it to 3.9V, for example, the output will stay at 3.3V. But if I initially connect 3.9V to the input, I will get only around 1.95V at the output. The output voltage will slightly increase with increasing input voltage, but at 4.2V there will be a bump and 3.3V at the output would stabilise.

At that point, while 1.95V can be found at the output, feedback pin gives 0.5V - where should be 0.8V (page 3 of datasheet) and LX pin frequency is around 800kHz. Where 3.3V is at the output, feedback pin is at around 0.8V, and switching frequency is around 1.3MHz, all as it should be.

According to my understanding, this misses datasheet specifications in two points:

  • Page 3, Input Under-Voltage Lockout Threshold for VIN raising is 2.9V. In my case it seems to be at the voltage I receive some output voltage, but not what is desired
  • Page 3, recommended operating conditions; Output voltage range: 0.8V to Vin

I even tried replacing with higher inductor with bigger one rated 3.3uH 4A, but with no change (as you could guess). All testing done with 1k resistor as a load. I can not understand at which points my circuit is (partly) failing. It should be quite simple and I stuck closely to datasheet recommendations.

Schematics

Tim Williams
  • 33,670
  • 2
  • 22
  • 95
davaradijator
  • 350
  • 3
  • 13

2 Answers2

5

I use this part in a lot of projects, mostly for applications with higher input voltage, e.g. 9-24V. But when I started trying to use from LiPo I saw the same behavior as you described. I followed the reference design also so I don't think there is anything wrong with your implementation.

I don't have a good explanation for how it behaves when starting from a low voltage compared to starting with a higher voltage. But I think if you increase your load you will see it behaves more closely both ways.

The main reason I believe the low input voltage performance is poor is because the max duty cycle is 87% (something I missed initially). I ended up changing to a different controller with 100% max duty cycle to reduce the dropout voltage.

I agree the specs can be misleading, especially recommended voltage range up to Vin. If you think about it there must be some dropout though since it's simply buck topology, but I also expected to get much closer to Vin.

AngeloQ
  • 2,167
  • 1
  • 10
  • 16
  • 1
    may I ask what part did you replace the AOZ1280 with? – Wesley Lee Mar 12 '17 at 05:50
  • 1
    I used Torex XC9236A30CMR-G (I needed 3V), but they have other fixed voltage and adjustable also. The max input voltage is only 6V and the current rating is half (600mA). It's about twice the cost but still fairly inexpensive. It has 100% max duty cycle which helps bring the dropout down. Note again they show max output voltage 6V, despite the max input voltage being 6V also. Maybe it's marketing 'wank'. I'm still satisfied with and still use plenty of the AOZ parts, but for some applications with low Vin I started using the Torex. – AngeloQ Mar 12 '17 at 13:46
3

I found similar behavior with low input voltage (abt. 4,5V) and output to be 3,3V when using AOZ1280. I found saw-tooth from 0V to 3,3V on output. Oscilloscope shows recurring voltage breakdown at bootstrap pin. So I tried a little schottky (e.g. BAT42) from output (3,3V) to bootstrap pin leaving bootstrap capacitor on place: success, it works! Looks as if internal bootstrap LDO regulator does not work properly at low voltage. I bought may AOZ1280 from chinese supplier(?). May be he sold some suspicious parts, although smd-marking seems to be original.

G. Hippchen
  • 31
  • 2
  • This is an insightful solution, not one contained in the datasheet, and can also potentially improve efficiency when the output voltage is low (notice V(BST, Lx) < 6V abs. max.). – Tim Williams Nov 04 '23 at 00:25