Regular power in fries it, SPI power in sails it

[kellogs]

Made my first PCB, yay! a MCU and a RF transmitter mainly. Programmed it OK through the SPI interface (which has a power pin feeding it stable 3.3V). All worked fine.
The circuit is intended to run off its currents at 12V obtained from a car lighter socket. So I have soldered the car plug, plugged it in and within 2-3 seconds there was smoke coming out. Pulled it off and visually inspected. Nothing gave any clues as being fried. Done this a second time with brand new PCB and parts, same deal.

A multimeter verification revealed that Vcc and GND were shorted. Took out everything except the MCU and RF transmitter - the board was still shorted.

My voltage regulator is a good old TO-92 L78L33AB which I have used just like in the datasheet example: one 0.33 uF capacitor on the 12 V rail and another 0.1 uF capactior on the Vout = 3.3V rail,

My first thoughts were heat dissipation problems, but after doing the maths it doesn't seem like this would be the case. Got about 25 mA drawn by the circuit and the car socket reads 12.3 V. Let's round it to 30 mA. With a thermal junction-ambient coefficient of 200 deg_C/W this would give:

(12-3 - 3.3) * 0.03 * 200 = 54 degrees rise.

Not that much to cause the failure I have witnessed twice. And it has worked fine if stable 3.3V was fed to it through the programming interface... Any pointers ?

Thanks!

 

[Harald Kapp]

A multimeter verification revealed that Vcc and GND were shorted. Took out everything except the MCU and RF transmitter - the board was still shorted.

• Which component(s) are shorted?
• Did you try the same circuit operating from a 12 V lab bench supply?
• Did you ensure that there are no short circuits anywhere from the PCB to the car lighert socket?
• Is the 0.33µF capacitor on the input rated for >12V supply voltage?
• Last not least: is the voltage regulator inserted correctly?

 

[kellogs]

• Which component(s) are shorted?

I believe either the MCU or the RF transmitter. I currently can not take those apart with my toolset.

• Did you try the same circuit operating from a 12 V lab bench supply?

I lack such a piece of equipment. If it helps, the cigarette socket can charge various phones OK.

• Did you ensure that there are no short circuits anywhere from the PCB to the car lighert socket?

I did use the multimeter on it and found nothing wrong. Now maybe just when I have plugged it into the socket it made a short, but... twice ? and with different car plug part types ?

• Is the 0.33µF capacitor on the input rated for >12V supply voltage?

Yes, 100V.

• Last not least: is the voltage regulator inserted correctly?

Yes, I believe so.
I also believe I have connected plus to plus and minus to minus

Many thanks!

 

[Harald Kapp]

I lack such a piece of equipment. If it helps, the cigarette socket can charge various phones OK.

Try a 9 V battery?

I did use the multimeter on it and found nothing wrong. Now maybe just when I have plugged it into the socket it made a short, but... twice ? and with different car plug part types ?

Try connecting wires only to the plug and measure the output voltage (magnitude and polarity) without anything else connected to the wires.

 

[kellogs]

Will do! I have taken out the RF chip - this is the one getting shorted. Board now does not short any longer.

Checked the specs again
- RF chip max voltage: 3.6V
- voltage regulator max V_out: 3.465V

*edit*

done the measurement on the PCB that I have taken the RF chip out. I have managed to take a 12.3V reading just before more smoke risen to my scent detector. Pulled it out and felt the parts. It was the voltage regulator the one overheated. The PCB hasn't shorted again this time. So my math above must be wrong...

 

[Harald Kapp]

my math above must be wrong...

Your math looks fine to me.

Yes, I believe so.
I also believe I have connected plus to plus and minus to minus

Your believe may be wrong.

 

[kellogs]

Your believe may be wrong.

I don't get it... I have triple checked now.

Also, measured the current drawn when fed through the 3.3V programmer interface:

- a PCB with the RF chip in place max 8.3 mA DC seen.
- a PCB* without the RF chip max 7.6 mA DC seen.

*This one has a different MCU - ATmega168PA instead of ATmega88PA, but still, at this amperage I am just baffled. Will try tomorrow feeding 5V / 9V through the car plug interface.

 

[davenn]

I don't get it... I have triple checked now.

Also, measured the current drawn when fed through the 3.3V programmer interface:

- a PCB with the RF chip in place max 8.3 mA DC seen.
- a PCB* without the RF chip max 7.6 mA DC seen.

*This one has a different MCU - ATmega168PA instead of ATmega88PA, but still, at this amperage I am just baffled. Will try tomorrow feeding 5V / 9V through the car plug interface.

it's well beyond time that you showed us photos of your setup and what is connected to where
sharp and well lit photos

 

[kellogs]

Rgiht!

- fed with 5.1V - measured 14 mA max
- fed wtih 9.7V - measured 0.01 mA - maybe the multimeter sensed danger and limited the current ?

It can be seen here how the fat reddish cap's left foot goes into the voltage regulator through a wide trace. I am feeding the PCB through that fat cap.

Have I really put the voltage regulator backwards ?

 

[davenn]

I am feeding the PCB through that fat cap.

well that cannot be correct as DC doesn't go through caps

can we please get another pic looking down on the board, showing the power connections

where this connected to....

- fed with 5.1V - measured 14 mA max
- fed with 9.7V - measured 0.01 mA

you existing pic just shows a small section and yes according to the silkscreen J1 appears to be connected correctly

do you have schematics or something you followed to build this up ?
can we see them please

 

[Harald Kapp]

Have I really put the voltage regulator backwards ?

Yes indeed. The datasheet explicitly notes that the TO92 package is shown in bottom view (figure 2, page 4).

 

[Harald Kapp]

according to the silkscreen J1 appears to be connected correctly

mounted correctly, layed out wrong - at least if he considers the trace at the lower left going fron the regulator to the red cap to be the input.

 

[kellogs]

Yes indeed. The datasheet explicitly notes that the TO92 package is shown in bottom view (figure 2, page 4).

FML... I am puting it the other way now

 

[kellogs]

Eh...reversed the voltage regulator and fed it 9.7 V through the multimeter - again got back that 0.1 mA reading (and a non-functional circuit). Took out the part and replaced it with a new one thinking it might have got damaged during the previous tests. Same deal.

Switching the multimeter to "2 mA scale" - the 0.1 mA current actually starts at about 0.8 mA and gradually decreseas to 0.58 mA or so where it stabilizes.

What now... ? More pictures coming up tomorrow. If my phone's camera won't die.

 

[davenn]

Eh...reversed the voltage regulator and fed it 9.7 V through the multimeter - again got back that 0.1 mA reading (and a non-functional circuit). Took out the part and replaced it with a new one thinking it might have got damaged during the previous tests. Same deal.

Switching the multimeter to "2 mA scale" - the 0.1 mA current actually starts at about 0.8 mA and gradually decreseas to 0.58 mA or so where it stabilizes.

What now... ? More pictures coming up tomorrow. If my phone's camera won't die.

more pics as I asked for earlier

 

[Harald Kapp]

plus the layout of at least the voltage regulator section. There may be more errors.

 

[kellogs]

Yep, camera's dead :|
Hope these are fine. The red fat cap is C1 on upper left side.

 

[Harald Kapp]

The layout confirms Vin and Vout of the regulator are swapped.
Apart from that I can see no obvious fault with the layout.