ESD

[Eric_t]

Hi, everybody

I am using the mc9s12dg256b controller.

I have some problems with electrostatic discharge; sometimes the controller resets.

I am applying 8KV air discharge (PCB distance from metal surface 10cm)

Had anyone faced the same problem?

Best Regards

Nikos

[Lundin]

What you should perhaps do is to add a small series resistor to all inputs from connectors, as those are most likely to catch ESD. For example, the CPU reset is present on your BDM connector.

For the reset itself, it needs a decoupling cap of around 47pF - 100pF. Larger caps like 100n will fail to give the steep edge on the reset that the S12 needs. And of course, you also need a pull-up on the reset line, and an external brown-out protection circuit (voltage supervisor of 4.6V or so) connected to the reset line, as DG256 doesnt come with internal LVD.

Is it possible for you to upload a schematic of your reset and BDM circuitry to the site?

Out of curiousity, why are you using DG256B and not DG256? DG256B is an early silicon mask full of bugs, which is also now obsolete, you can't buy it from Freescale (but from Rochester I think). Most likely you shouldn't use this mask neither for development nor production.

[bigmac]

Hello,

---

Lundin wrote:

For the reset itself, it needs a decoupling cap of around 47pF - 100pF. Larger caps like 100n will fail to give the steep edge on the reset that the S12 needs. And of course, you also need a pull-up on the reset line, and an external brown-out protection circuit (voltage supervisor of 4.6V or so) connected to the reset line, as DG256 doesnt come with internal LVD.

---

Surely the maximum allowable shunt capacitor value at the reset pin would depend on the external pullup resistor value used.  If I were to assume a resistor value of 10k in conjunction with 100pF capacitor, the time constant would be 1 microsecond, giving a rise time of about 3 microseconds at the reset pin.

If it is possible to reduce the pullup resistor value to, say 1k, the capacitor value could be increased to 1nF for the same rise time.  This could be beneficial in two ways - a much larger amount of induced energy to discharge the capacitor, and a higher amount of induced current to be maintained once the capacitor is partially discharged.  Of course, the low voltage reset device would require to sink 5mA.

This assumes that the reset pin is the problem, and not some other reset source.  If the interference is severe enough to affect the Vdd line, you will get a LV reset anyway.

Regards,

Mac

[Lundin]

Yeah my bad, in these days of power saving I silently assumed that all pull ups are 10k or larger. The point is, you can't just smack on some 100nF cap there like you would on some generic IC, or the MCU might get stuck in resets or not start at all.

But the best solution for ESB is surely to add a 100 ohm series resistor to every signal routed to any form of connector.

[Eric_t]

I  have used mc9s12dg256B and  mc9s12dg256 with the same results.

I apply 8KV air discharge (pcb distance from metal surface 10cm). After few discharges the controller resets.

I have used microchip PIC and NEC controllers and I have never had such problems.

Unfortunatelly I have not much experience in freescale controllers.

I suspect that something is going wrong with the PLL.

I have uploaded the schematic, pcb and the code which simple blinks a led.

Best Regards

Nikos

[Lundin]

Yep, as I suspected, you are using a way too large cap on the reset pin. Change it to 100pF.

Also follow all other advise given in this thread. For example you must have a voltage supervisor circuit on the reset pin, or the CPU will go bananas in case of drops in the supply voltage.

---

I should refrain from commenting the PCB layout as I'm a clueless software guy, though I suspect some of the EMC problems may be related to the ground plane. Umm... where is the ground plane, more exactly? There's some blue stuff under the oscillator, but I'll assume the rest of the board have a big ground plane too.... right?

 And are you actually trying to cad a 112LQFP on a 2-layer PCB with just one component side? You have attached antennas to each CPU pin... I'd either mount all pull-downs on the bottom of the board, or preferrably: cad a 4 layer PCB and ground them as close to the pin as possible. But then I'm just a software guy and know little of PCB design.

[astro_goto]

Electric arc discharges have very fast rise times inducing very high frequencies in exposed wires. Even
short leads of bypass capacitors present a high impedance allowing very little current to flow.

Dealing with ESD problems involves as much art as science as you'll learn to appreciate. Full enclosure
shielding is known to be effective and you could give it a try. I've also used series 100 ohm resistors
with pull up and pull down schottky diodes on the inputs with much success.

[fanziyu]

what do you mean "on the inputs ",micro input  pin or micro output  pin?or just IO pin.

i meet this problem:+15kv air discharge at the metal surface.and no shielded enclosure for som reason.

PCB distance from metal surface 5cm:reset

PCB distance from metal surface 3cm:micro break down

can you help me?thank you!

[astro_goto]

Your electrical discharge is producing an electro-magnetic pulse in its vicinity and will induce
a voltage pulse in every unshielded wire connected to your microprocessor, including power supply
and I/O wires. If the micro itself is not in a shielded enclosure, every trace on the PCB will
be similarly affected.

[Eric_t]

If the micro itself is not in a shielded enclosure, every trace on the PCB will
be similarly affected.

What do you mean with "shielded enclosure"?

Do you refer to PCB or to controller package?

Best Regards

Nikos

[astro_goto]

A shielded enclosure is a metal box. The micro and as much of the system must be in the metal box. If the electro-magnetic
disturbances are severe, the metal box cover should employ conductive strips to ensure complete continuous contact
between mating edges. Any external wires connecting to the micro must be shielded and pass into the enclosure through
shielded connectors. The shorter the wires, the better.