Current injection / Port pins HC08 HCS08

[galahad]

Hello,

it was some kind of a shock for us  to see that newer chips of the HC08 (QC) and most (all?) of the HCS08 have a much smaller limit for current injection on port pins.

The limit for HC08 was usually +-2mA

Newer chips now tolerate only 0.2mA

We used 220..470 Ohm resistors to protect ports for years, we never had problems. In many of our our applications you might reverse a connector, and then about 5V are at the port pins without any supply to the MCU. Again, so far no chip failed.

But with 0.2mA this seems to be far more dangerous. We stopped a MC9S08GT16 design already and will use a good old (but expensive) MC908GR16 instead.

Any comments or experiences? Thank you!

[bigmac]

Hello,

The injection current might further be limited by inclusion of dual Schottky diodes connected to Vdd and Vss supply rails, as required.  The Schottky diode should conduct at a lower voltage than the parallel internal diode.

Regards,

Mac

[galahad]

Mac,

we would need this kind of protection 9 times... Needs to much place on the PCB. And i don't know the exact voltage of the internal diode (does anybody?) - to be absolutely save, you would need two resistors and two schottky diodes for each IO port pin in this situation.

After all, it's much easier to use a MCU with better protection.

[bigmac]

Hello,

If your port pins are truly accessible to the "outside world", you will need at least the level of protection described to achieve a robust product.  Simply providing a series 470R resistor will not be enough, whether the maximum injection current is 0.2mA or 2mA.

Even with the Schottky diode approach previously described, I would suggest that the series resistor value should be at least 10k, to further limit the current through the diode, perhaps to allow for an applied static discharge impulse.

Another variation to protect MCU inputs might consist of two resistors, and a zener diode or transient voltage suppressor.  This method will clamp the transient voltage, to limit the injection current through one of the series resistors.

For output signals, I might suggest that the MCU be isolated by use of simple bipolar transistor or MOSFET buffers.

Regards,

Mac

[celsoken]

Dear Mac,

The S08 (QG8/GT16)  datasheet seems to show contradictory information. First it shows +/- 25 mA (@ abs max ratings) as max current input and after (@ DC characteristics) it shows DC current injection as 2mA (greater than VDD) or even worse -0.2mA (negative input).

I don't believe that FS reduced the input clamp diodes, that would be nonsense and the +/- 25 mA info seems to confirm that. I really think that the clamp diodes withstand the 25mA but you have to provide a path for the current flow, e.g. place a shunt regulator (or even a zener diode between VDD and VSS pins, but be careful) so the current that flows thru the internal clamps can be absorbed by the external element, so the resulting voltage between VDD and VSS could be kept under max VDD. Using the DC current injection probe in automotive modules used to work like that.

The designs I've been working with rely on these premises. I hope they are right. Pls do send your comments.

Regards,

Celso

[bigmac]

Hello Celso,

Yes, the datasheet does seem a little contradictory.  I might guess that the maximum values shown in Table A-5 are applicable to steady state conditions.  The absolute maximum value given in Table A-1 may apply to short transient peaks, possibly based on a single event.  I don't really know.  I note that, for the QG8, the maximum is 0.2mA in both directions.

It is also quite possible that the survival of the protection diode is not the only factor.  There could also be a possibility that the injection current might cause latch-up of the device, to be avoided at all costs.

Whether or not a zener diode is required between Vdd and Vss will depend on the duration of any injection current impulses.  For short transient events, the usual bulk capacitor (typically 10uF) might be cabable of absorbing  the energy, without exceeding Vdd maximum.  However, for the automotive environment, the zener is probably a good safeguard.

My personal approach is to design for zero injection current. even though, this may involve additional buffer components.  I do not rely on the presence of the protection diodes.  If a failure should occur, my aim is for it to be one of the buffer components, and not the MCU.

Regards,

Mac

[celsoken]

You are absolutely right in terms of zero current injection. That is perferct and I wish I could afford that. The problem is that I am normally involved in cost-sensitive applications and normally I cannot afford the outside protection but I need to meet the specs, including costs specs.

That's why we keep on using the internal clamp diodes. Latchup is avoided limiting the current with the input series resistor. The problem happens when the current injection generated e.g. by a load dump flows thru several input pins and there is low power consumption. There has to be a dummy load to absorb the current e.g a shunt regulator of any kind. Also ADC readings can be affected by the VDD change if your shunt regulator is not good enough.

Maybe we should ask for a freescaler to help us understand. Do you know how call their attention?

Cheers,

Celso

[galahad]

That would be great...

[BasePointer]

Hi,

Yes, Freescale DC clamp current is so low. Microchip supports 25mA for all MCUs.

I'm using 15Kohm serial resistor to connect 5V SPI output to 3.3V input of LC60. It is OK for me.

Good lucks,

BP.