Buck driver design advice needed

I have just finished the first version of my custom buck driver design.

Driver

I would like to drive a 6V XHP-70 led from 2s-3s lipo.
It works, but the inductor gets very very hot.
I use a coilcraft XAL 6060-333, it is 33uH and 5A rated.

Is it a normal behaviour that it gats really hot?

I was suggested to replace it. I also mage a simulation ( it might be wrong) which shows ~30W dissipated heat at 12.6V input and 5A led current.

a typical inductor size for 2-3S to 6V at 5A is 4.7uH

33uH is far too big, how did you calculated the needed parts?

with such high inductance the ripple is also very low which is not ghood for the regulation feedback, depending how the output current is sensed

an usual unsynchronized buck driver from 3S to 6V at 5A shoulc dissipate at most 3W heat

What’s the resistance of the coil? It might be “rated” 5A, but that might be saturation current, not design current. That’ll at least get you the raw wattage burning up heat in the coil (I²R losses).

One (design) failure mode of inductors is when the core goes into saturation, so then you’re burning up current that’s not adding to magnetic energy, just adding heat.

here my LM3409 calculation based on the formulas in the buck chips datasheet

this is 13mm full sized inductor with 8.4mOhms, a XAL7070 should get about 16mOhms

also from what I see the input caps are not even close to be sufficient, usually at 500kHz you need about 20uF

so add more safety 3x 10uF 1206 25V capacitors, best is X7R

output cap should be 10uF 16V

22mm

23mm single sided Klarus

I use an MAX 16820 LED driver.

Maximintegrated has a tool to calculate the inductor and the sensing resistor.

The result:
MAX 16820 calculation

It gives 4.62 uH, it is technically the same as that you suggest. But I was also suggested to use ~10 times higher inductance than the calculated one.
And if you see the graphs in the datasheet (page 4), 47 uH is used.
It was the way somehow 33uH was finally selected.

I had only a 22uH coilcraft XAL 6060-223 at home, so I used it instead of 33uH. I was also very hot.

I also made a simulation and it shows ~30W heat dissipated by the inductor. It does not matter what inductance I choose. I might be wrong, I do not know. bBut the led current is more or less OK. (pwm 500 Hz 100% duty)
Simulation

Coilcraft XAL 6060-223 22uH was finally used
Maximum DC Current: 5 A
Maximum DC Resistance: 60.63 mOhms

This is the current design:

I build over two hundred Buck drivers so trust me 4.7uH is right there is no take 10 times the inductance
I managed to get up to 96% efficiency on 4S to 12V builds, so about 72W to the LED and 3W on the driver, the parts stay nice an cool

for a good regulation you need about 0.5-1A ripple depending on the use buck chip, this is what 4.7uH gives you
also such an inductor has less DC and far less AC resistance so the heat will be significant lower

usually I use coils that I use are at least 3-4 times higher rated than the output current to keep losses low,
you cant use a 5A max rated coil for 5A output

also noticed on your design the input caps are missing, no wonder the inductor has a hard time

same with the capacitors there is a calculation that defines minimum which with like 10uF at 500kHz for 3S—>6V,
you can simply calculate it fromm the offtime C=I*toff(/1000000000 for ns)/1,44****(2 minimum safety value, better x4)
I doubled itat least to be on the safe side,
the more input caps the better for efficiency as the voltage stays higher so we get less current losses

you can trust the tool, the results are almost identical to my LM3409 calculation

0.8A ripple current is perfect

also at 200kHz you need input caps

the thing is its dynamic switching frequency like most high side sensing resistor buck controllers
no way to modify it besides Inductor selection, this leads you down more or less to measure switching frequancy and then if its not around 300-500kHz change inductor value

I took your values and calculated the fwitching frequency
I get about 800kHz with 4.7uH, which is a bit on the high side

the thing is that switching frequency should be calculated by the tool but its not, very odd

you could even drop the LDO as the chip has a 5V 10mA LDO integrated you can put the MCU on it

I calculated the switching frequency now and would even drop the inductor to 3.3uH

this gives me 550kHz switching frequency

with 33uH that thing would run on 5.5MHz no wonder it cookes like crazy

Thank you very much.

I will try it with a 3.3uH inductor and add input capacitor 3x10uF.
Shall I also add a 10uF capacitor accross the LED terminals? Or is it only optional? No calculation of it in the datasheet.

I also calculated the switching frequency according to the formula that can be found in the manual of the 16820 chip.
According to this, if you increase the inductivity the frequency gets lower.
And if you change the switching frequency in the calculator of maximintegrated the min. inductivity do not change.

I am not sure with low side switched buck, likely no caps across the LEds

Can you have a look at the modified design?
Would you suggest more modifications?

Design3

(R4 or R3 is optional, only for flexibility, different sizes and power)

I have just had time to build the modified driver.
With the modification you suggested the driver got cooler but it is still hot. Especially the FET.
driver

I connected it to an adjustable power supply and measured the followings:

Input 8.4V 4.16A >> 34.9W
LED 6V 5.04A >> 30.24W
So the efficiency is about 86%

I tried out what happens if I use it without cooling (the driver was not attached to any cooling plate, only the led).
The input current increased and when the efficiency decreased to about 83%, the fet got smooking hot and died.
In this case the driver should have dissipated ~6.5W

So I think that my driver less efficient and much hotter than it should be.

And one thing that is also strange, I ajusted the input voltage from 6 to 12 V. i thought that the led current should be constant. But is changed parallel to the input voltage.
I thought that a buck driver provide constant output if the input voltage is bigger than the led voltage.

Which FET do you use, sounds it has way too high switching losses
You need one with low gate charge, usual decent fast N-Mosfets have 4-7mOhms but only about 5-10nC total gate charge

Has this regulator a fixed switsching frequency or do you need to set it with a resistor?

It likely switches way too fast

The only way to see whats going on is using a decent digital memory oscilloscope

Also looked in the datasheet and according to it the buck has no low dropout, so its not allowed to drive 6V LED with 2S battery configuration

Increasing voltage should stay constant current so something is buggy

I use PSMN013-30MLC115 powerfet.
FET

13.6 mOhm RdsOn
8nC Qg

The XAL7070-332ME is only 3.3 uH, this is a mistake on my grawing.
XAL7070-332ME