Processor is sensitive to moderate Temperature

[T J]

Posted on December 29, 2015 at 03:33

I am using the STM32F072RBT6  64pin LQFP

My code seems quite stable, but this chip is showing signs of being overly sensitive to temperature...

Since I have so much capacitance around the board on the same power lines,

I didn't put so much capacitance as suggested in the data sheet.

We are running a little hot on the board, approx 70C 2'' away, ambient maybe 60C.

should I change the package on the board ?

which package has the best temp dissipation ?

#!junctiont-!stm32f072-!unstable

[gregstm]

Posted on January 03, 2016 at 02:03

Lots of capacitance is good - but the most effective capacitance will be close to the micro. So a surface mount, 0.1uF ceramic capacitor (X7R) as close to each power/ground point is essential. Otherwise it is like asking a person to drink their thickshake through a three foot long straw. It's less of an issue if the micro is running very slow - but it sounds like your chip is screaming along.

For those wanting to create a product for sale, a good understanding of bypassing is essential - especially if you want to reduce electromagnetic emmissions (or if there is any analog circuitry on the board).

The heat is a separate issue and would have to be investigated in a systematic way.

[re.wolff9]

Posted on January 03, 2016 at 14:56

Although ''all bets are off'' if you don't put the datasheet-recommended capacitances around the CPU.... in practise the '072 works just fine for me with a few missing capacitors. I recently came across the datasheet again and realized that my PCBs had a missing 4.7uF capacitor that hte datasheet recommends. So my most recent PCBs have the footprint, but so far I haven't had any trouble.... 

You are saying the chip seems temperature sensitive. But you're not saying what it IS or is not doing that you were expecting.... 70 degrees at 5cm (2'') away should be quite tolerable. Actually I have a PCB that works just fine with a 70+ capacitor at around 5cm from the '072RB.... Yeah, the capacitor getting that hot was not planned.... (the 10x10 mm capacitor was not up to the job, Hopefully the new 35mm diameter 50mm high capacitor is enough.)

Your '072 should work just fine at 70 degrees, even if it itself is that hot. But that wouldn't be normal. 

[jpeacock]

Posted on December 29, 2015 at 10:35

When bypassing power to the IC proximity is important.  Each power/ground pair on the STM32 should have a bypass cap, plus a bulk cap close by.  Do not ignore the datasheets.  The purpose of the caps is to manage current spikes from transistors switching inside the IC as instructions are executed and peripherals change state.  Capacitors somewhere else on the PCB are not the same.

70C at the package seems high, you better check if some IC pins are trying to drive a ground.  Does it get this hot if stopped or only while the firmware is running?

[Amel NASRI]

Posted on December 29, 2015 at 13:27

Hi marsh.nick,

I suggest you have a look to the application note 

http://www.st.com/st-web-ui/static/active/en/resource/technical/document/application_note/DM00051986.pdf

 ''Getting started with STM32F0x1/x2/x8 hardware development''.

It provides some recommendation on how to design your PCB and set power connections besides to required capacitance to be added.

You need to respect them with the AMR conditions of the datasheet.

-Mayla-

[T J]

Posted on December 30, 2015 at 03:42

ok, ok, I will add the 3 x 4.7uF beside the existing 3x 0.1uF and the 2 primary 1uF caps nearby.

Literally there are 2x 18uF Electrolytics, 10x 0.1uF and 4x 1uF caps already on the same power line only used for 3.3V only used by the processor.

I think the heat may be due to the 20cm SPI bus lead going to the LCD running at 5MHz

The display is very busy,,, 

Thanks for the input, I will order the industrial high temp part too...

[T J]

Posted on December 18, 2016 at 03:25

Hi guys,

sorry for the late reply, this site was shocking to log into....

seems ok now...

ok, I found the problem.

I found that driving any substantial length of cable from a processor pin generates heat inside the processor.

OMG its a revelation...(not)

The 10MHz SPI port seemed to be the problem and fix.

I had 2x 50cm cables running 2x   4.3' LCDs from 2x   SPI ports.

which all worked fine until the ambient reached 60-70C

I put 27R in series with three SPI lines, MISO, MOSI and CLK ( probably don't need resistor for MISO)

Then I made the Dual 4.3' LCD setup run off 1 SPI port. (that's a diode trick for 8c)

The cable is still 2x 50cm running from a single 10MHz SPI port,  all running happily now around 40-60C ambient.

All tested ok on the '072 and the '091.

(Shifted now to the STM32F091RC, double Flash, double Ram, slightly cheaper ?? with 2 pin changes...

:(

)

again, sorry for the delayed response

Nick