6
\$\begingroup\$

In my design, I have a +3.3 V SMPS that provides power to digital circuits (microcontroller, EEPROM, etc.). The microcontroller allows applying a 100 Hz PWM signal with variable duty cycle and with an amplitude of +24 V to a resistive load. This is done with the following circuit:

PWM driver circuit

Where:

  • HEATER_PWM is a 100 Hz PWM signal with an amplitude of +3.3 V coming from the microcontroller.
  • HEATER_A and HEATER_B are the connections to the resistive load.
  • HEATER_ISENSE+ is the voltage signal used for current sensing.

+3.3 V power supply

+3.3V power supply circuit

+24 V power supply

This is the plot of the +24 V power supply when the PWM is activated.

+24V power supply output plot

Issue

When the 100 Hz PWM is enabled (and only when it is enabled), the +3.3 V power supply produces an audible noise which seems to be at the PWM frequency. I suspect that this is due to the power inductor "vibrating" at 100 Hz.

I tried to increase the value of R116 to 100 kOhm without any effect.

How could I get rid of this acoustic noise?

\$\endgroup\$
5
  • \$\begingroup\$ Where did you connect your oscilloscope ground when doing the measurement? \$\endgroup\$
    – Andy aka
    Commented Dec 11, 2013 at 11:00
  • \$\begingroup\$ @Andyaka I did this measurement on the +24V power supply input connector terminals with a low length twisted ground probe. \$\endgroup\$
    – johsey
    Commented Dec 11, 2013 at 11:06
  • \$\begingroup\$ You might try a diode in series with the 24V feed to C11/C12. The caps will become charged to the peak voltage (minus diode drop) and with luck you won't see so much ripple onto the LM3103. It's solvable. \$\endgroup\$
    – Andy aka
    Commented Dec 11, 2013 at 11:18
  • \$\begingroup\$ @Andyaka To be honest, I don't see the value of adding a diode in series. Could you explain a little bit further the impact of the diode on this 100Hz ripple and on the spikes? \$\endgroup\$
    – johsey
    Commented Dec 11, 2013 at 12:32
  • \$\begingroup\$ I'm trying to isolate the 24V noise so that it doesn't appear on the supply to the 3.3V converter. The ripple will still remain on the main 24V supply but should be somewhat less on the supply that feeds C11/C12 \$\endgroup\$
    – Andy aka
    Commented Dec 11, 2013 at 15:30

1 Answer 1

Reset to default
2
\$\begingroup\$

The first question is, what is switching 5A or 7A at 100Hz doing to your +24V supply? It will be adding some ripple to it, can you measure how much on a scope? I'm guessing the ripple looks a bit like a square wave.

The second is, how much current do you need from that 3.3V supply? 0.1A, 1A, 10A?

Assuming less than 1A or less than 3W total, so the current drawn from the 24V supply is 0.1 or 0.2A...

Then you can feed the 3.3V supply from 24V via a low value resistor : 1 to 5 ohms, dropping less than an volt across the resistor - max 0.1 or 0.2W wasted power.

And replace C11 with a much larger capacitor - 100 to 1000 uf. Then R-C provide a low pass filter to reduce the PWM ripple, or at least slow its edges to a ramp, so that the regulator can respond fast enough to maintain control of the 3.3V output without "singing".

\$\endgroup\$
3
  • \$\begingroup\$ The load is a PTC that can be as low as 3.5Ohm (~7A) and as high as 20Ohm (1.2A). Indeed, the ripple on the +24V power supply looks like a square wave at 100Hz (I updated my post with the plot). The absolute maximum current drawn by the +3.3V power supply is 0.7A. But the average current is more like 0.1 to 0.3A. In my application, a electrolytic capacitor of maximum 100uF would be acceptable for size constraints. Do you think a low-pass filter with a cut-off frequency of 1.6KHz will reduce significantly this audible noise? \$\endgroup\$
    – johsey
    Commented Dec 11, 2013 at 11:04
  • \$\begingroup\$ I think you are hearing high frequency spikes from the inductor at a repetiton rate of 100Hz, so, yes. \$\endgroup\$
    – user16324
    Commented Dec 11, 2013 at 13:19
  • \$\begingroup\$ Ok, I will add such a RC filter. \$\endgroup\$
    – johsey
    Commented Dec 11, 2013 at 13:59

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.