• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Subscribers 62 subscribers
  • Views 901 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LMZ14202 Noisy/unstable at low loads

Chris Westcott
Prodigy 30 points
Other Parts Discussed in Thread: LMZ14202, LMZ14202H, LMZ14201

We are running an LMZ14202 to create 5.2 V from a 12V input. I am experiencing problems with the circuit when running at low loads (<300 mA or so). 

The output voltage is correct but very unstable, lots of noise and variation due to small changes in output current. For example, just turning on an LED that draws 6 mA is causing a 10 mV dip in the output voltage. The bigger issue is that the switching noise on the output is much worse and the switching frequency changes dramatically with load. I can actually hear the parts buzzing, and hear the pitch changing as the LED's turn on and off. The output waveform has periods where the switching cycles are present and then they go away, almost like it is PWM controlled or something similar. 

I tried hooking up a sizable load resistor and all of this went away. The output was stable and there was a continuous switching waveform observable. 

I may be able to load it down enough to solve this problem by adding dummy loads, but I am wondering if changing R_ON and the switching frequency can impact this. We are currently at 100k for ROn which is 400kHz. I was thinking of trying a higher frequency 500-600 kHz. Will this help? 

Also, I could not find anywhere in the datasheet where minimum load currents are mentioned - is there a recommendation for minimum loading for this part?

Thanks for you help.

-Chris

  • 0
    TI__Mastermind 18940 points
    Hello Chris,

    Here is some background on the part.

    At light load the device enters discontinuous conduction mode. In this mode of operation the switching frequency will decrease with the load going down. As a result of the lower frequency, there will be larger ripple on the output voltage when compared to the full load case and full switching frequency (as you have noticed). In terms of regulation though the output voltage should still be regulated.

    The voltage regulation point happens at the valley of the output ripple. The actual DC regulation point will appear to be shifted up by half of the ripple voltage. Since there is more ripple at very light load (very low frequency) the output voltage will appear to be higher at lighter load. You can probably see this in Figure 25 of the datasheet (for 3.3V output).
    This drift at light load will be higher for higher output voltages.

    The LMZ14202H actually improves the regulation at light load. If you think the drift at light load is too much you can try the pin-to-pin LMZ14202H part instead.

    Questions:
    Do you have any scope shots of the output voltage?
    Do you have a minimum switching frequency you need to avoid? (the feedback resistor network can be used to provide some minor loading to the output to maintain some minimum switching frequency)
    Can you share your board layout and schematic for a quick review?

    Cheers,
    Denislav
  • 0 in reply to Denislav Petkov
    Prodigy 30 points

    I have attached the requested pics below. 

    At low loads, it is running at about 3kHz with 80 mV pp noise. When I add 350mA of load current it improves, now runs at the setpoint of 400kHz and noise is down to ~12mV. This is interesting because per my calculations it should still be in DCM, but yet it is running at the setpoint frequency. Does the switching frequency gradually increase until CCM is reached, or does it stabilize to the setpoint at some earlier threshold?

    We actually have some 14202H parts on the same board and they are not showing this problem at all. We used the 14202 for the 5.2 volt rail because there was a chance we would need to set it to a lower voltage (~3.3V) and we thought it would be better not to run it so close to it's bottom limit of 5V. But I think we are fairly set on 5.2V now. So I could swap these out on the next build.

    I would still like to know if using the 1Amp LMZ14201 part would improve things, or if increasing the switching frequency setpoint would help. Our max load without adding any dummy loads is probably ~300-500 mA. 

    Thanks for the quick helpful response!

    -Chris

  • 0 in reply to Chris Westcott
    TI__Mastermind 18940 points

    Hi Chris, 

    I don't think the LMZ14201 will have a different behavior than LMZ14202. Both parts have the same inductance value inside (shown in the block diagram) but with different current limit level. If you are not using the 2A capability, you can go to the 1A part (probably better for cost too) but I don't think the behavior will be different.

    The "H" version will actually work for voltages under 5Vout as well. The non "H" version will be slightly more efficiency at full load (because of the smaller inductor). So, in terms of flexibility you could still use 1 part (the "H") even if you want to go to 3.3V at some point.

    Setting the switching frequency to a higher point may help a bit but I don't think it will be a big change. The sweet spot for these modules for efficiency is setting the frequency at around 450-500 kHz at full load. 

    Regards, 
    Denislav