How to reduce inductor heating in buck converter?

[vishweshgm]

I am designing a simple buck converter to drive 9 V, 3 A LED. Supply can vary between 20 and 30 V. I used the TI Power Designer Tool to calculate the inductance value and other parameters. Here is the schematic:

After driving the load, everything is fine except that the inductor is getting heated to almost 85°C.

My PCB has to go into an enclosed box so this is not OK.

Strange thing is, the MOSFET is not getting hot at all, so the gate driving is fine, I guess. I have no clue why only the inductor gets heated in a buck converter.

Please suggest what factors concerning the inductor I should look into.

Vin : 20-30V switching frequency : 350kHz , Note that I am driving exactly 3A load (using shunt resistor (0.1ohm,2W) sensing to limit current)

Inductor HCM1A1307V2-330-R : Datasheet

Mosfet SI_4288DY : Datsheet

Diode SS54 : Datasheet

Driver IR2110 : Datasheet
Here is the calculation screenshot of TI tool:

 

[Easy peasy]

To solve diode heating only solution is to replace with a finer low-drop diode which I am yet to search digikey

there is essentially not a finer or lower drop diode ..... but feel free to search ...

--- Updated Jan 11, 2022 ---

paralleling diodes works just fine in a switcher - as long as they are thermally connected and have much the same track resistance to each - however the total watts in losses does not reduce very much at all as Ploss still = Von x I .....

--- Updated Jan 11, 2022 ---

you can also go to a 3 layer or 4 layer pcb and use the extra layers for heatspreading across the pcb - this works fairly well, and keep the fibreglass layers to the min thickness you can get away with e.g. 0.8mm or less.

 

[vishweshgm]

[This message is off-topic of this thread]

Cload limitations at output LM358 -

[Vishwesh] : Honestly, I had to google the terms that you are talking about. As @FvM mentioned, I am using average value of the adc read voltage to map the current drawn by load in SW. First I use rheostat to draw 3A from my board and note down the adc voltage and from next time, I use error-gain correction control system to limit pwm from allowing more than 3A current draw. I adjust the feedback resistor value to get a satisfying adc read and this is working fine. But, indeed I may be lucky here as system has not reached unstable point.

However, I see ringing behaviour at the output of opamp, which I have filtered out in SW. This may be result of high Cload. I now understand and agree that this method to put capacitive load is wrong. If some-one else reading this thread I am here giving links to right theory documents, to understand the points made by danadakk:

1. What is Cload parameter in opamp? Read this : Design aspects in Opamps driving capacitive loads.
2. What is BodePlot and poles? Watch this : Brian Douglas

Also the slew rate of LM358 horrible for trying to make accurate measurements in 350 Khz signal path
for monitoring -

[Vishwesh] : Since I am trying to calculate only the average current which won't increase very fast(as I control pwm duty cycle slowly in SW), 0.5V/us felt good to me. Isn't it?

And I see no caps on supply pins ....? The PSRR of an OpAmp degrades with frequency, in LM358

[Vishwesh] : I thought PSRR as the measure of opamp which tells its stability in case the supply voltage given to it has ripples. With this understanding I thought, since my supply is stable, no worries here.

If the noise is caused at the output, then you should put the RC at the output. But this may cause DC error due to voltage drop at the resistor. To avoid this you may do some kind of DC feedback from the capacitor (not directly from OPAMP_output)

[Vishwesh] : Noise was usually at the output. I will resdesign the circuit with RC filter just to be safe. But I donot understand the DC error part. Since I am using microcontroller to read the voltage, I can just adjust the resistor drop in code to map the right voltage. Isn't it?

--- Updated Jan 11, 2022 ---

I guess you can't save more than 20 - 30 % of diode power disspation by using lowest available forward voltage. A synchronous rectifier build with a second MOSFET would be the preferred solution.

[Vishwesh] : You are right. I looked up parts, but didn't find good SMD diodes for 5A rating that I could use. Synchronous rectifier will need more space which I cannot accomodate.

you can also go to a 3 layer or 4 layer pcb and use the extra layers for heatspreading across the pcb - this works fairly well, and keep the fibreglass layers to the min thickness you can get away with e.g. 0.8mm or less.

[Vishwesh] : Im gonna have to stick to 2 layer due to cost reasons. 3-4 layers pcb manufacturing becomes a lot expensive here in India.
Another idea I got is to use a D-PAK diode VS-50WQ03FN-M3. My PCB Enclosure can have a metal on the bottom side (where mosfet already sitting). If I move this diode to bottom along with mosfet and with enough copper pour and bottom metal I might just be able to get away from problem. This is the only approach I see for now.

From tomorrow I need to look into better inductors, as pointed out earlier from vishay or coilcraft. things gonna get expensive for small unit such as this...! difficult to face customer..

 

[danadakk]

Yes on avg current. The idea of slew rate affects control loop response. With a low rate then it acts as
an integrator of sorts, so that can be good. But if you had a fast device powered by this supply, and a
big transient occurs you could have a situation whereby the device starts misbehaving because the control
loop allows that transient to incur on the supply output. But caps on its supply pins helps that. There are
designs out there that need very fast control loops, I am guessing your design is not one of them.

The PSRR issue comes about becuse as you can see from the curves at highj freq what ever noise/transients
are on the supply buss to the loop control will be affected by lack of good PSRR. But again seems like your
loads aree not too concerned with fast deep (high amplitude) transients.


Regards, Dana.

 

[Taihung]

I am designing a simple buck converter to drive 9 V, 3 A LED. Supply can vary between 20 and 30 V. I used the TI Power Designer Tool to calculate the inductance value and other parameters. Here is the schematic:
View attachment 173763
View attachment 173762
After driving the load, everything is fine except that the inductor is getting heated to almost 85°C.

My PCB has to go into an enclosed box so this is not OK.

Strange thing is, the MOSFET is not getting hot at all, so the gate driving is fine, I guess. I have no clue why only the inductor gets heated in a buck converter.

Please suggest what factors concerning the inductor I should look into.

Vin : 20-30V switching frequency : 350kHz , Note that I am driving exactly 3A load (using shunt resistor (0.1ohm,2W) sensing to limit current)

Inductor HCM1A1307V2-330-R : Datasheet

Mosfet SI_4288DY : Datsheet

Diode SS54 : Datasheet

Driver IR2110 : Datasheet
Here is the calculation screenshot of TI tool:
View attachment 173764

Hi

The software suggested 44.26uH but you selected far away 33uH.
Normally should select higher value as 47uH. I think, should double-check with other software as LTCad.
If chose smaller, should increase frequency to reduce Ipeak-peak, and also reduce heat.
Let try with 47uH and maybe it will be cooler.

The compensation loop for LED driver is not need adaptive fast as normal power source. But it still need stable, if not, it make higher lost power, higher heat. Current loop control vs Voltage loop control.

To measure Ipp through inductor, just solder a small resistor 10mR 2512 in serial with inductor then use OSC measure voltage drop on it.
I often design with ferrite core than Alloy powder core material which use in inductor because it often has lower core lost vs high frequency > 100kHz.

If design for automotive application, I suggest design solution with Isw current monitor cycle by cycle to increase safety as short circuit, overload, EFT effect.

I don't know why your design with discrete solution that make more expensive and bigger.
I just designed similar application with input 15-32Vin 12Vout@6A 250kHz with 30x30mm size. Vout ripple 9mVpp@5A 15mVpp@<4A. The temperature of inductor just 42 degree, but IC 65 degree @ full load.

My design use with 10uH inductor instead of high as 33uH, output cap only 150uF rated Tantan + 33uF rated Ceramic:

 

[danadakk]

Phase margin versus time domain response in a control loop -

Too much and our step response takes on RC exponential like behavior. Too little
and things can blow up.

Pole locations versus time domain -

LHP stuff decays, RHP grows without bounds. On the jw axis stuff stable oscillations.



Regards, Dana.

 

[vishweshgm]

If chose smaller, should increase frequency to reduce Ipeak-peak, and also reduce heat.
Let try with 47uH and maybe it will be cooler.

[Vishwesh] : I have tried upto 750kHz actually. But still heating exists. From discussions done in this thread I believe it is due to Inductor core loss rather than Inductor DCR loss. I agree ferrite core materials are better for this, but my application has strict space and cost constraints. I can go max upto 17mmx17mm area. But low-cost ferrite core materials need minimum 20mm footprint which I cannot accomodate in 50mm dia PCB. So precise thermal calculation necessary here.

I have now found another part SRP1770TA-560M (I couldn't go for wurth or coilcraft due to cost ). Can Anybody help me precise calculate core loss for my application (and thus help me understand how to do it)? I found the following formula online but I am unable to find x&y parameters in datasheet

The compensation loop for LED driver is not need adaptive fast as normal power source. But it still need stable, if not, it make higher lost power, higher heat. Current loop control vs Voltage loop control.

[Vishwesh] : I want to discuss more on this point please. Here I attach the video of pwm control from power-on. It slowly increases duty cycle and loop stays at 3A DC output. However since its control loop there is small correction always going on in pwm and this small instability is forever running. In my application it is around 50-100ns (at 00.28 timestamp in video). Are you talking about this stability?
(In video its 200kHz, but its same for 350 or 500kHz)

I don't know why your design with discrete solution that make more expensive and bigger.

I actually earlier tried MAX20078 and failed horribly. (blewup 2 LED arrays). MAXIM support team were very much unresponsive to my queries and I understood pure analog design is not my strength. I am good at SW, so I tried this. If not for heating, this is good lowcost design for other lowpower led drives.

--- Updated Jan 12, 2022 ---

If design for automotive application, I suggest design solution with Isw current monitor cycle by cycle to increase safety as short circuit, overload, EFT effect.

Nop. Its not automotive but it will be used in few places to run 24/7 continuously..

 

[vishweshgm]

Coil resistance:
Resistance is usually is the DC resistance. Thus the calculated 0.36W (0.37W) is the DC loss.
As cupftea mentiones .. you need to add the core loss.

@KlausST , you had mentioned I need to add core loss. I realize not completely understand this point.

[Vishwesh] : Can you please explain a bit more. From datasheet I could find this graph but unable to understand it. What is Bp-p value?

@FvM told:

See datasheet page 2, note 5.

But in datasheet Pag2, note5, Bp-p meaning is not given (I mean how to find value in watts)

I have now found another part SRP1770TA-560M (I couldn't go for wurth or coilcraft due to cost ). Can Anybody help me precise calculate core loss for my application (and thus help me understand how to do it)? I found the following formula online but I am unable to find x&y parameters in datasheet

I googled and found one calculation formulae but unable to understand and link that here. Can you help me with calculation please? (just to remind, 9V,3A buck system 350kHz, delta I is within 18%)

I need calcualtion example on how to do it on this inductor (which is not ferrite core, but alloy powder core)

 

[FvM]

But in datasheet Pag2, note5, Bp-p meaning is not given (I mean how to find value in watts)

Everything is given, read again. You calculate Bpp according to given formula and K factor of the respective inductor, than get loss from the loss curve.

 

[Easy peasy]

We all look forward to hearing the path you chose to solve this thermal issue ... ?

 

[vishweshgm]

You calculate Bpp according to given formula and K factor of the respective inductor, than get loss from the loss curve.

Oh okay, I was clearly missing some obvious thing here. Sorry for troubling you again. I understand it now.

We all look forward to hearing the path you chose to solve this thermal issue ... ?

Yes, It is very clear that I need to change the inductor. This time Im gonna use higher inductance value 68uH. I have chosen IHLP6767GZER680M11 - https://www.vishay.com/docs/34282/ihlp-6767gz-11.pdf

I donot know which type of core this device has, but they say its patented and their website shows only 11degC temp rise(screen attched). If this happen, that would be great. It might take a week or 2 to reach me( Things are usually slow here .. and everyone knows pandemic made it worse).

There was another inductor MSS1583-683MED which I am interested, as it is of same size as my current inductor and its ferrite core wire-wound. Coilcraft calculation tool predicts 33degC rise, which might be just ok forme. But its not available dealers that I know in India, I am checking with few other places.

Anyhow I'll update the this thread after my tests. Thanks.

 

[Easy peasy]

for 0.9 watts in the choke - you're going to see more than 11 deg C rise unless a lot of copper is attached to the pads ....

 

[KlausST]

Hi,

You are focussed on the inductor temperature only. But your circuit is in a closed box, thus every power dissipation will heat up what's inside the box.
Now we have seen that the diodes have a lot of power dissipation.. thus I recommend to reduce it by using a synchronous (rectifier) switch mode circuit.

Mind:
In a closed box the average temperature or better say the internal ambient (to the single parts) temperature is determined by the total power power dissipation, not by the single devices temperature.
Thus - in my eyes - you need to take care total power dissipation.
In other words.
A small diode with 0.5W power dissipation will get 60°C hot and raises the in_box_temperature to 35°C.
While a big diode with 0.5W will get 50° hot, but still the in_box_temperature will be 35°C.
So a big diode reduces the "hot spots"(which surely has it's benefit) , but not the average temperature.

Also mind: inside a box there is reduced air flow. Thus you can't well compare temperatures with and without box.

Klaus

 

[FvM]

As long as you don't change the enclosure geometry or add venting, total dissipated power inside the enclosure is the parameter of interest, not hypothetical temperature rise numbers. With the latest inductor change, you manage to at least double the inductor power dissipation, presumed the datsheets don't lie.

 

[cupoftea]

If you want a cheap solution, you can out an NTC temperature sensor in the enclosure...and just dim the led down when the temperature rises too high....probably people wont notice.

 

[Easy peasy]

contrary to the knowledgeable writers above - if you have a sealed box, and the thermal resistance of the box in degC/watt to the out side air is on the high side, say 40degC/watt, then for modest power dissipation inside the box the temp will rise to that temp needed to give thermal equilibrium

If the outside air is 35 degC say and you have 1W in the box, then for the above example the internals will rise to 75 deg C for the 1 W dissipated.

for 3 W the temp will be 155 deg C inside the box,

There are some caveats, as the box external wall temp rises, its ability to shed heat goes up - as convection effects increase - assuming there is free air around the box, so the internal rise with power is not linear ( slight advantage ).

Such thermal equilibrium can take half to one hour to stabilise

You can test all this by gluing a square/rectangular ceramic 5W resistor ( typically the white ones ) to a bare pcb, putting say 3 watts into it, attach a temp sensor to the R and to the pcb edge, and another to measure air temp - seal up the box and let it run - the results will be instructive

you can vary the power to see the effect on the pcb temp away from the heat source and the general air temp inside the box.