ARM experimentation/developments

[commie]

Hi all,

I just thought I'd let you know how my building of an ARM platform went today, I started pcb design on Thursday just gone, I finished the board design yesterday and went ahead and etched 2 boards, built one today. When I'd finished, I was quite proud of my work.I'm using STM32F030C6 48 lqfn soldered onto a square adapter board which gives two rows of 6x2.54mm each of the four sides to give a total of 48 pins.

When the testing time came, I plugged in the power and the STlink/V2, and straight away I could amell burning, switched everything off and unplugged the processor then checked the supply rails only to find everything normal. So tried again under control conditions and found the processor taking enormous amounts of current.I then started tracing the pins and names and found I had gotten all evens pin numbers mixed up with all odd numbers and vice versa, in other words when connecting the processor to the psu it was polarity reversed hence my processor was/is completely fried.

Indeed, I wouldn't have minded if I had detected earlier because I spent about 3 days designing/building it, I was essentially building a pile of pooh but wasn't aware.

[bson]

Errors happen.  I don't know how many times I've been saved from letting the smoke out by my PSU going into CC mode and the voltage gone down to a single diode drop (0.7V) - a telltale sign of a short over a protection diode.

[Bob F.]

Bad luck!  But as bson says, these things happen.  I once got all the data lines to the program memory (which was in eprom - going back a bit here) reversed so D0 -> D7, D1 -> D6 etc.  I somehow got the pin numbering confused with the bus numbering in my head as the pinout went in that direction  .

I sold it to my boss as a "security feature" to make it difficult for rivals to steal the code 'cos they would assume the data lines were the right way around and read garbage.

He bought it... 



It was only a prototype so, no worries, as they say in Daveland.

[v8dave]

This is one of the reason I love a proper bench power supply with current limiting. It saves you so many times.

Those nice new adapters to use a PC power supply scare me as the current limiting on them is not the same and you can dump a high current into a short circuit for some nice explosions.

On the swapped data bus of the other posting I did that last year on a VFD clock I built. I had to use a lookup table to convert the bits into the right format but it worked. Luckily it was just a home project but little silly mistakes like this make you think harder next time and one thing I now do is share my design with a friend to look it over for anything obvious. He does the same with his designs to me.

[commie]

Well, spent the whole of last night re doing the pcb and have just finished soldering all the parts in. Guess what?, it works so I have qfn-48 and qfn-32 ARM capability. The amazing thing is, stm32f030c6 48 qfn I thought fried yesterday is still working even after reversing the supply across it. This little adventure, learning how to hand solder qfn-48 and my incompetent mistakes cost me about £50.

Cheers
Commie

[Brutte]

Hi all,(..) I'm using STM32F030C6

Typical current consumption
3.6V, execution from flash, HSI 8MHz default clock, draws 4.4mA to 5.2mA max

Not sure what currents you expected.

in other words when connecting the processor to the psu it was polarity reversed hence my processor was/is completely fried.

5mA and 0.7V is 3.5mW and it is definitely not enough to fry a uC by just reversing polarity.
Would you reveal what was the current limiter set to?

 

[Christopher]

I always use 10ma current limit now and increase while watching the voltage cos of silly mistakes like this. Also put a 100r resistor on the end of the power leads for a bit of protection

I also just buy Dev boards. Saves my precious time

[commie]

5mA and 0.7V is 3.5mW and it is definitely not enough to fry a uC by just reversing polarity.
Would you reveal what was the current limiter set to?

Hi Brutte,

I didn't have any current limit on testing( perhaps I should have done because I do have cc psu), the uC was supplied via a 3V3 regulator and the STLINK/V2. To my amazement the chip survived, I couldn't believe it, really.

[commie]

Also put a 100r resistor on the end of the power leads for a bit of protection

Snap, so do I.

I also just buy Dev boards. Saves my precious time

I actually bought 2x dev boards of Ebay at £3.40 each all inclusive, I checked the Farnell price for that type of uC and the uC alone cost   £4.20 each. That said though, Ebay purchasing/sourcing would be unreliable for product design(large quantities on demand), so it would be better to develop your own.

[Brutte]

I didn't have any current limit on testing( perhaps I should have done because I do have cc psu), the uC was supplied via a 3V3 regulator and the STLINK/V2.

You did have. It was most likely LD1117 and it limits current in the range of 0.8A to 1.3A
Even though the chip did survive, scrap any abused chips to save yourself frustration in the future.

[Howardlong]

Agreed with all regarding the current limiting, it becomes a natural part of everyday life to set the voltage _and_ the current limit.

However... sometimes you're on a project and typically you're incrementally adding new functionality or changing performance, this can come and bite you on the butt. There have been a few hours of my life lost on MCU projects where every thing works fine, then I do something like ramp the core clock up or start using a peripheral that takes the PSU into current limit, and the MCU resets with a brown out in a matter of 100us or so. Because it happens so quickly, it's not immediately obvious what's happened: after the reset the PSU still shows all's OK, the DMM measures the supply OK, but that tiny PSU glitch where it went into CC is long gone. It's just part of life's rich EE experience.

[bson]

Well, just so you don't feel too bad about doing something dumb...  I just got a batch of boards back and realized I had consistently used the wrong footprint for four of a critical part - I misread the datasheet. 

Sigh.
Fixed.
New gerbers generated.
Reordered.

It's the "stupid" tax.

[dannyf]

straight away I could amell burning

it pays to start your circuit, for the very first time, with a serial resistor of the appropriate values.

[commie]

it pays to start your circuit, for the very first time, with a serial resistor of the appropriate values.

Hi,

I did have a series resistor in series with the input of the 3V3 regulator, it was this resistor which I could smell. BTW, the chip survived.

Cheers
Commie

[Scrts]

I always limit the current when trying to power the board for the very first time

[Sal Ammoniac]

I always limit the current when trying to power the board for the very first time

This. I also use a multimeter to check a new board for obvious errors such as power shorted to ground before I solder anything onto it. The MCU is the last thing I solder to the board and only after I've checked everything else out.