Step Down DC DC Converter

[mahen]

Hi Guy,
The product that I'm handling is using a buck DC DC Converter, whereby the output will be around 13.8 V and the input is able to vary from 11V to 26V. The prodcut was soaked in the temperature chamber whereby it goes thru a temperature cycle from +85C to -45C for 2 days which is part of the test profile and was found to have the switching transistor (PNP) burnt.
Could anyone tell me what could have caused the problem. The voltage used at that time was 17V and it normally drains around 3A but we found out that the current went up to 14A before the transistor got burnt.

What are the potiential causes of the transistor to get burnt?

 

[SkyHigh]

Hi mahen,

temperature cycle from +85C to -45C

1. Is the PNP transistor specified on the datasheet to operate in sub-zero temperature such as -45°C or even below -20°C?

2. Did the datasheet recommend you to place a heatsink to the PNP when used above room temperature at the case (not the junction)?

the current went up to 14A before the transistor got burnt

3. When you detected input current went up to 14A to the PNP transistor, what was the temperature at that time or slightly before the incident happened?

I suspect the PNP transistor "melted down" because of the extreme heat at the reversed-biased base-emitter diode junction in the PNP. If I am correct, check the pins of your PNP used in the buck converter. Are the pins at the base and emitter burnt or "oxidised" more obvious than the one at the collector?

 

[IanP]

-45°C to + 85°C is quite wide temperature range ..
Most likely the damage had occured at +85°C, or close to this limit.
What you have to check is the temperature rating of the components.
Pretty common commercial rating is somewhere between 0°C and ≈85°C.
If you think the heatsink of the switching transistor was adequate in size then I would blame rather other components, mostly caps, for the circuit failure ..
Regards,
IanP

 

[mohamedtaha]

How can a buck converter output 13.8V from an 11V input, it can't be a buck regulator then.

 

[mahen]

Hi Mohamedtaha,
Sorry but when it is 11V it will gives out around 11 V as well....when it is highete then 13.8 then it will constantly gives out 13.V. Sorry for the confusion.

Added after 5 minutes:

HI IanP,
The problem is that this PNP transistor doesn't come with a heat sink.;( Previously before we qualify this product it was not having such problem. Nothing changed in the design. But after 1.5 years of selling this product recently only we were facing this problem on the newly manufactured product.

Added after 15 minutes:

Hi SkyHigh,
The temperature cycle from +85C to -45C is the test profile we normally use for our product for years. And as mentioned this product was initially having no issue but recently only. We have checked if there is anywhere shorting due to the process issues but everything looks ok.

Sorry but we didnt manage to monitor the temperature when the current went up to 14A as we run 30 units of products at once and it is difficult to get data like that. But i believe it would have been +85C that time.

Normally the the Emitter and Collector pin is shorted and sometimes burnt off. Have attached the Data sheet and the burnt PNP Transistor. Hope it will help.
Thanks for all your support.

 

[SkyHigh]

I have seen the photo taken for the burnt PNP transistor and I have read the datasheet for this transistor.

1. By visual inspection of the photo, the emitter is seriously burnt. The collector and base are fine. Although dark brown stain appears on the collector's pin, it is the burnt plastic coated on the collector pin as it is next to the emitter (pin and the plastic case package).

2. Based on the datasheet given and from your description on how the PNP transistor is used in your product, the transistor is not operating well below the maximum ratings.

Refer to the datasheet, -15A is the maximum (-ve sign indicates that the collector is the input of PNP), but -8A is the typical rating when the PNP is saturated (Vce) at room temperature of 25°C.

Refer to page 3 of the datasheet, look at plot "Forward Bias Safe Operating Area"
Ic is limited to a maximum of 11A when the Vce is 2V at room temperature.

14A will definitely "fry" this transistor even at room temperature, needless to even mention about operating it at higher temperture and higher switching frequency.

Notice that the switching frequency also causes increase in power intake. What is frequency used?

Reminder: This PNP used here has being designed such that the collector is able to take up higher current because it is the input, usually with a current-limiting resistor before it to protect the collector from high-voltage and high-current supply. Usually since the base is protected by limiting resistors and the hfe or gain of the PNP to ensure very low output current from the base, it is safe. But if the load side, where the emitter is connected, does not offer sufficient current-limiting protection, the emitter will "fry" under such circumstance.

If the PNP transistor is suddenly switched off, which results the input voltage at the PNP collector at the PNP input is lower than the emitter and load, the back current injection from the load will surge the PNP from the emitter. This back injection effect is made stronger from the inductor due to back e.m.f. (Lenz's Law).

Don't forget that the capacitor between emitter and load can discharge to the load and to the emitter of the PNP since it is the common path or node at the load.

I suspect there is no protection diode in series between the emitter of the PNP and the inductor.

There is however one diode which is usually used in step-down or buck converter but this diode is connected in parallel to the inductor and capacitor (capacitor parallel to load). However this doesn't protect the PNP emitter.

You can have the following protection method:

Connect a parallel power diode (1N4003 for example) between the emitter and collector of the PNP such that the cathode of the diode is connected to the PNP collector, anode of the diode connected to the PNP emitter. This is to ensure a flywheel circuit to provide a discharge path from the charged capacitor at the load to discharge stored energy via the PNP in a recursive loop. This also provides a dicharge path for the back e.m.f. from the inductor due to Lenz's Law. This diverts the huge injection current via the diode.

This PNP transistor only reported to fail recently because it has be used for some time. It worked for some time, but during the period when this PNP is working, the emitter bond wire between the die and the package pin is slowly fusing due to back injection until one day this problem occurs.

 

[IanP]

Maybe you should consider power PNP transistor with lower Vcesat.
Take a look at, for example, BD744: **broken link removed**
From its characterisitc Vcesat/Ic you will find that this transistor will have Vcesat<1V@10A ..
Possibly you just need more robust transistor ..
Regards,
IanP