DC-DC MOSFET switching signal frequency is not stable

Thread Starter

calvin0908

Joined Mar 9, 2010
28
I designed a DC-DC supply circuits based on LM5116, during debugging, I tested the switching signal of MOSFET pin 4 of Q18 and found the frequency is not stable as I expected 400Khz, it is changing, the switching signal waveform I captured is as below: from the waveform you can see the duty cycle of the signal is always changing.
1673514628980.png




The schematic circuit is as below:
1673514542541.png
 

Thread Starter

calvin0908

Joined Mar 9, 2010
28
I also found the inductor of L3 is very hot (the temperature of L3 I measured is exceed 68 degree C at room temperature 25 degree C at very low current load like 0.5A
Is there anyone can tell me why the switching signal frequency is always changing and what is wrong with the design? is it the root cause that caused the inductor L3 HOT?
Thanks
Calvin
 

Papabravo

Joined Feb 24, 2006
21,159
I also found the inductor of L3 is very hot (the temperature of L3 I measured is exceed 68 degree C at room temperature 25 degree C at very low current load like 0.5A
Is there anyone can tell me why the switching signal frequency is always changing and what is wrong with the design? is it the root cause that caused the inductor L3 HOT?
Thanks
Calvin
What range of frequencies can you measure? I would expect the duty cycle to change, but not the frequency.
Inductors get hot for several different reasons:
  1. What temperature rise do you expect from the DC resistance of the wire used to fabricate the inductor?
  2. What is the peak inductor current implied by the operating conditions? It may be larger than either the input current or the output current.
  3. Do you know if the inductor core is saturating?
  4. How much time does the MOSFET gate spend on the Miller plateau?
The answers to those question may lead you to the root cause.
 

ronsimpson

Joined Oct 7, 2019
2,986
Are you certain the frequency changes. Looks like the duty cycle changes.
Looks like the loop is instable. Try slowing down the error amp by 10x and see if that helps.

60C is not hot for an inductor.
 

Thread Starter

calvin0908

Joined Mar 9, 2010
28
Are you certain the frequency changes. Looks like the duty cycle changes.
Looks like the loop is instable. Try slowing down the error amp by 10x and see if that helps.

60C is not hot for an inductor.
Yes, the duty cycle changes caused the frequency measured by oscilloscope changes also, I selected 3 pulse of switching signal from the waveform randomly to do the time measurement, the time is different, then calculate the frequency by 1/T, the frequency changes.
slow down the error amp? do you mean to try adjusting the R8/C174?
Thanks
 

Thread Starter

calvin0908

Joined Mar 9, 2010
28
What range of frequencies can you measure? I would expect the duty cycle to change, but not the frequency.
Inductors get hot for several different reasons:
  1. What temperature rise do you expect from the DC resistance of the wire used to fabricate the inductor?
  2. What is the peak inductor current implied by the operating conditions? It may be larger than either the input current or the output current.
  3. Do you know if the inductor core is saturating?
  4. How much time does the MOSFET gate spend on the Miller plateau?
The answers to those question may lead you to the root cause.
Yes, the duty cycle changes caused the frequency measured by oscilloscope changes also, I selected 3 pulse of switching signal from the waveform randomly to do the time measurement, the time is different, then calculate the frequency by 1/T, the frequency changes.
So far from the simulation the inductor is not saturated, simulated I inductor current is close to the I load current, is less than the inductor saturation current.
  • How much time does the MOSFET gate spend on the Miller plateau? The MOSFET is not hot, but the inductor, is this related to MOSFET Miller effect?
 

Ian0

Joined Aug 7, 2020
9,667
Is your output voltage more than half your input voltage? (it would appear so from the duty cycle)
Then you should be using slope compensation.
 
After reading the Data Sheet, it recommends some special consideration of biasing the RT/Sync and the AGND pins of the LM5116.
"Specifically, The RT/SYNC pin can be used to synchronize the internal oscillator to an external clock. The external clock must be a higher frequency than the free-running frequency set by the RT resistor. The internal oscillator can be synchronized to an external clock by AC coupling a positive edge into the RT/SYNC pin. The voltage at the RT/SYNC pin is nominally 1.215 V and must exceed 4 V to trip the internal synchronization pulse detection. A 5- V amplitude signal and 100-pF coupling capacitor are recommended. The free-running frequency should be set nominally 15% below the external clock. Synchronizing above twice the free-running frequency may result in abnormal behavior of the pulse width modulator."
Could you be getting some abnormalities due to the bias?
 
Top