improve efficiency of my DC/DC converter

Thread Starter

bug13

Joined Feb 13, 2012
2,002
Hi guys

I am having troubles of managing the heat (too hot to touch) from my DC/DC converter. So I would like to improve the efficiency of my DC/DC converter. I think I have narrowed it down to either my PCB layout, or my inductor selection, but I am not sure.

Current efficiency is about 80%, generating about 2.6w of heat @ 5.3V 2A load, 12VIN. I am hoping to improve the efficiency to about 90%.

So I thought I can post it here and ask for any idea.

I am suing this DC/DC converter ACT4533BYH-T and this inductor, I am using low ESR input and output cap.

Here is the circuit (straight out of the datasheet, with minor modification)
schematic.PNG

Here is the pcb:
Capture2.PNG

Capture.PNG
 

ronv

Joined Nov 12, 2008
3,770
Hi guys

I am having troubles of managing the heat (too hot to touch) from my DC/DC converter. So I would like to improve the efficiency of my DC/DC converter. I think I have narrowed it down to either my PCB layout, or my inductor selection, but I am not sure.

Current efficiency is about 80%, generating about 2.6w of heat @ 5.3V 2A load, 12VIN. I am hoping to improve the efficiency to about 90%.

So I thought I can post it here and ask for any idea.

I am suing this DC/DC converter ACT4533BYH-T and this inductor, I am using low ESR input and output cap.

Here is the circuit (straight out of the datasheet, with minor modification)
View attachment 177344

Here is the pcb:
View attachment 177345

View attachment 177346
Usually there are only a few places you lose much power.
The FET in the IC. Probably not your problem.
The clamp and blocking diode. A possibility.
The inductor saturating. Looks okay
The calculation/measurement?
What part is the hottest.
 

Thread Starter

bug13

Joined Feb 13, 2012
2,002
Usually there are only a few places you lose much power.
The FET in the IC. Probably not your problem.
The clamp and blocking diode. A possibility.
The inductor saturating. Looks okay
The calculation/measurement?
What part is the hottest.
What exactly do you mean by calculation/measurement. You mean the theoretical calculated efficiency? The 80% efficiency is my measured number.

The thing is, the components are so closed together, I don't know which part is hottest, this properly doesn't help. Would be good if I have access an thermal image camera...

Probably should replace D2 with a PFET- maybe save 400-500 mw right there?
Pretty spot on, D2 draws about 390mW according to my measurement.

If you are careful to not connect the input voltage backwards you can replace D2 with a wire jumper.
I do need input protection, may put a FET or put the diode parallel on the input.
 

MrSoftware

Joined Oct 29, 2013
2,188
Get a thermocouple (cheap) and try to determine which part is getting the hottest. Also get a scope on it and make sure it's running at the frequency and duty cycle that you're expecting for your given output. Note that if you see spikes in the waveforms on the scope, they might not actually be there. It could be your probe or probe-ground wire acting as an antenna and catching the EMF from the inductor.

I didn't study the datasheet close enough to pick out the differences; can you explain how you deviated from the datasheet circuit and why?

Also your cap footprints look like you might be using electrolytic, can you scab some ceramics on there and see if that helps?
 

drc_567

Joined Dec 29, 2008
1,156
The datasheet for the ACT4533 converter indicates that the chip efficiency, at about 2 amps load current, should be approximately 90%, at 25 C. (p. 13) However, it will likely have a lower efficiency at an elevated operating temperature. So locating the heat source and reducing the heat input ... should be goal. A finned heat sink for the converter chip could make a difference.
 

djsfantasi

Joined Apr 11, 2010
9,156
Get a thermocouple (cheap) and try to determine which part is getting the hottest. Also get a scope on it and make sure it's running at the frequency and duty cycle that you're expecting for your given output. Note that if you see spikes in the waveforms on the scope, they might not actually be there. It could be your probe or probe-ground wire acting as an antenna and catching the EMF from the inductor.

I didn't study the datasheet close enough to pick out the differences; can you explain how you deviated from the datasheet circuit and why?

Also your cap footprints look like you might be using electrolytic, can you scab some ceramics on there and see if that helps?
Would a digital thermometer work? Like a digital meat thermometer? Under $10US
 

Thread Starter

bug13

Joined Feb 13, 2012
2,002
I didn't study the datasheet close enough to pick out the differences; can you explain how you deviated from the datasheet circuit and why?
It's straight out of the datasheet on page 12. Except I disable the output short protection as I am testing it with a BIG 2ohm resistor, and this resistor always trigger the output short protection.

And then just make sure the inductor is not saturated (well that what I intended anyway)

Also your cap footprints look like you might be using electrolytic, can you scab some ceramics on there and see if that helps?
Yes there are two input caps and two output caps, there is a 1206 10u ceramic bypass cap (C5) very close to the VIN, and 47uF (C4) electrolytic cap as datasheet suggested.

I did however replaced the 47uF input cap with a low ESR one, still very hot.
 

Sensacell

Joined Jun 19, 2012
3,432
That's where the thermal camera comes in- helps you see it in clear perspective.
The damn thing was expensive but has become one of my favorite tools.
FLIR0044.jpg
 

Thread Starter

bug13

Joined Feb 13, 2012
2,002
Also, is there a way to measure/tell the inductor is saturated?? I want to make sure it is not saturated. Thanks team!
 
Top