Hi,
I am designing a buck converter with the MP9928 controller.
My parameters are Vin 20-40 V, Vout 5 V, Iout = 10 A, Fsw 500 kHz
I have an issue with the bottom gate drive. The signal is high for a very short period of time instead of being complementary with the high side. See attached oscilloscope capture (Top : switch node, Bot : low side gate voltage (the waform is not clean since I had no proper probe ground connection)
It happens whatever the input voltage from 20 to 40 V and load from 0 to 10 A. The low side antiparallel diode takes all the current during the expected low side conduction period.
I do not have the issue in MPSmart simulation.
I tested on different samples of the converter and the isssue is the same.
Do you have ideas what is causing the issue ? What could I investigate ?
Thanks in advance,
Téo Robert
Here is the schematic of the converter :
Hello @teo.rbt ,
It seems to me that your converter is going into AAM/ efficient light load mode. You can measure the switching frequency with the switch node, if the frequency is less than the intended 500khz then you are in AAM. Get a scope capture of the inductor current as well, see if the AC component of the inductor is larger than its DC component, which would cause the inductor to go into DCM when AAM mode is enabled. Inductor current is always a good thing to probe when trying to understand what is going on in your circuit.
It could be worth looking into the inductor you are using, make sure both its rated and saturation current are at least 20% larger than peak inductor current. For a buck converter, peak inductor current would be the output current plus its ripple, note that a larger inductor leads to less ripple and vice versa.
Hope this helps.
Regards,
Jonathan Hidalgo
Hi,
Thanks for your answer.
The converter is not supposed to go in AAM mode since I disabled it by making the CCM/AAM pin floating. However, at light loads, i did see that the part does work in PFM which is unexpected I think.
The previous scope capture was done around 5A of output current. I can not easily scope the inductor current due to my setup. However, it is guaranteed considering my ripple that the inductor current never go near zero. The inductor saturation current is around 19A.
I measured constant 500kHz switching frequency, which make me think its more on an DCM than AAM, where the part turns off the low side fet. However, DCM is not required since the inductor current is high.
I tred fixing the AAM voltage to a low value with a resistor (15 kOhms) on CCM/AAM pin and I still have the issue.
Thanks in advance,
Teo Robert
Here is an additional scope capture with AAM voltage in pink and COMP voltage in blue, showing that AAM is not activated since they does not cross each other.
Hello @teo.rbt ,
Apologies, you are correct that that IC should not be going into AAM since the AAM pin is floated. After looking at your schematic again I believe the issue could be that the VCC pin does not have a decoupling capacitor. Try adding a 1uF cap from VCC to GND and try changing the bootstrap cap to 0.1uF instead of 2.2uF as recommended in the datasheet.
VCC is what powers the internal circuit logic, perhaps what is happening is that when the low side switch turns on, VCC is supplying current to charge the bootstrap cap and since the bootstrap cap is so large the current needed to charge it is large as well, leading to VCC drooping down below a value that is required for the internal logic to work.
Also, unrelated to the current issue we are discussing, the bootstrap resistor should have one end tied to BST pin and the other to the node with both the diode and cap. Right now when the Cboot is used to help turn on the HFET, the diode is off so the cap does not see the R and the turn on rate of the HFET is not affected. Try using no more than 2.2 ohms for the bootstrap R as recommended by the DS.
Hope this helps!
Regards,
Jonathan
Hi,
Sorry, i forgot to add the Vcc capacitor on the simulation schematic i provided, i actually have a 2.2uF ceramic capacitor on this pin. Similarly, i forgot on that schematic the 2.2 ohms resistors on the fet gates which allow tuning of the turn on rate. My 10 ohms resistor between Vcc and Bst was there to limit the current at startup since I have a large value bst capacitor.
I changed the BST capacitor to a 0.1 uF, and I noticed no change. I also tried increasing the gate resistors value and noticed no change.
However, a new thing i did notice is that if I increase the current (around 8 A) the low side on time progressively rises and the issue disappear. It seems that the controller is in DCM on a too high range of current. To sum up :
-
I have DCM from 0 to 2A which is expected considering my ripple of around 4A. The low side control signal is too short however (yellow switch node voltage, blue, low side gate voltage) ;
-
above 2A i am out of DCM but the controller still apply the minimum on time ;
-
at around 8A the controller start to rise the on time until it is properly complementary to the high side.
The two possibles causes i identified are :
- noise on my current sensing which could cause mistriggering of the DCM. It is unclear in the datasheet how the controller controls DCM. It seems to be independent from AAM.
- unadapted compensation network values, but in my opinion it would not only affect the low side. My frequency and high side on time are very stable so is the output voltage.
Additionally, I saw on the forum another topic with a similar issue, although it seems that they have solved the problem on their end : Please check the Mos driver waveform of MP9928
If you see any other possible sources of this failing DCM control or if you have more details on how the MP9928 controls it, I am interested.
Thanks in advance.
Teo Robert
Hello @teo.rbt ,
Kindly change your compensation network to match that of the evaluation board:
680pF cap in series with a 51K resistor, remove the parallel cap for now. The parallel cap should only be needed if the ESR condition mentioned on pg 21 is valid.
The stability of the converter determines if you go into DCM or not, generally for DCDC converters the lighter the load the more unstable the control loop is.
You could also try getting and testing our Evaluation Board: EV9928-F-00B | 60V Current Mode, Synchronous Step-Down Controller Evaluation Board | MPS (monolithicpower.com). Once the EVB design works the way you like, you can copy design on your own board.
Let me know if this helps and solves the issue.
Regards,
Jonathan
