The circuit above functions as expected and gives output of 12V when supplied with a voltage greater than about 15V.
I have tried gradually increasing the input voltage upto about 35V with no problems. However, when the supply is switched on at 32V or above the chip fails and ..smokes!
Considering that, according to the datasheet, Vin can go as high as 40V, I cant explain why it fails at this level.
The idea is for the above circuit to be supplied from a 24V transformer which gives after rectification and smoothing about 34V DC.
Did you look at the input voltage with a scope? How smooth is the output of your rectifier at the loads your testing?
If your peak voltage is above that 40V, this might explain the issue (35V RMS -> 49V Peak). You might not want to go above 28V peak (40V RMS). If you really need to get up to 40V peak, you might consider another device, just to be safe and over-design.
How tight is the layout - switch-mode converters need strict attention to avoiding too
much stray inductance in the main current paths, otherwise large induced voltages will play
havoc (perhaps this has happened here?). 35V + 5V induced pulse = 40V, for instance.
How tight is the layout - switch-mode converters need strict attention to avoiding too
much stray inductance in the main current paths, otherwise large induced voltages will play
havoc (perhaps this has happened here?). 35V + 5V induced pulse = 40V, for instance.
The fact is that the layout is quite tight. Although I payed attention in the layout itself to keep critical nets short and used copper fills for ground on both top and bottom layer. As a result, and unlike previous attempts, I now get perfect waveforms across C36 with no distortion.
Do you think there is a way to verify if this is a case of stray inductance? Anyway to limit it now? (without designing another PCB!)
What current do you expect to draw? Inrush current is probably your problem somehow. Can you increase R36 to limit maximum current to lower values? I did not make calculations how much peak current would be in your case but I think it is very possible to grow over limit when Vout is 0V and Vin is 34V. If ~0.3 V over R36 is exceeded the CE current does not stop instantly - charging of C36 is increased and until it is charged the current grows.
Whats the minimum load current?
I suspect that L2 is too low and is saturating at switchon as Smajdalf has indicated, due to the current inrush caused by the switcher chip ramping up its Ton time to maximum to
minimise the voltage differance between 0 and 12V.
Slowly winding up the input voltage wont cause this as the switcher will gradually ramp its PWM duty cycle down.
And get rid of C34.
Why is it wrong? In datasheet of LP2950 (a LDO linear redulator, not switched) of TI they use a cap in exactly the same way to "reduce output noise". Is there an important difference?
What current do you expect to draw? Inrush current is probably your problem somehow. Can you increase R36 to limit maximum current to lower values?
The current doesn excreed 300 mA. I thought the same and increased R36 to 0.5Ω but didnt make any difference.
Whats the minimum load current?
About 50mA
I suspect that L2 is too low and is saturating at switchon as Smajdalf has indicated, due to the current inrush caused by the switcher chip ramping up its Ton time to maximum to
minimise the voltage differance between 0 and 12V.
Slowly winding up the input voltage wont cause this as the switcher will gradually ramp its PWM duty cycle down.
I am thinking of the possibility this problem is due to switching spikes during switch on ? Is that possible? Anyway to limit it?
So you are suggesting I should increase L2? 220uH was well over the min value suggested by design tools
The chip has an absolute maximum rating of 40V, you are sailing very close to the wind. The only
thing I can think of is adding a fast acting spike-clamp circuit on the input, assuming that switchon
transient voltages are causing the failure (which is likely).
The supply you are using to test with may be generating voltage spikes on switch on, and it sounds like
you have no oscilloscope to diagnose this.
The chip has an absolute maximum rating of 40V, you are sailing very close to the wind. The only
thing I can think of is adding a fast acting spike-clamp circuit on the input, assuming that switchon
transient voltages are causing the failure (which is likely).
The supply you are using to test with may be generating voltage spikes on switch on, and it sounds like
you have no oscilloscope to diagnose this.
Can you suggest some sort of "fast acting spike-clamp" ?
An yes, I have an oscilloscope but I was unable to catch any spikes. What I did was I had the power supply already switched on and then I plugged in the above circuit. I have also tried this with 24v AC RMS as input from a mains transformer with the same result. (24V AC = 33 V DC Peak)
In an effort to isolate the problem, I have tried a few different types of inductors.
At the end I found out that the circuit works just fine with this inductor while at the same time the problem above replicates with this inductor!
Both inductors are rated approx the same Irms current of 500mA. For the second one (which also causes the problem) the Saturation current is also specified at 900mA while for the first one (which works just fine) Isat is not specified. I would expect, it would be the other way round: ie the inductor with the saturation current specified, would work fine and the other one create the issue above - hence the confusion!
I now have the above regulator circuit supplied with 25 V AC and loaded with 300mA at 12V DC output on an electronic load for testing for about an hour. It looks stable.
By the way, my target application would have the circuit above permanently supplied with 24 V AC (which is readily available as control voltage in industrial panels) and supply a stand alone arduino based controller with 12V DC and 5V DC.
