Automotive input protection cost-down

[paulie]

Is this automotive input protection excessive?

If this is for personal use then IMO yes. I used a simple 1m resistor (actually 2 470k) instead of all that in a 1986 Fiero 2 seater and it performed perfectly on a daily basis for 21 years until the car was retired as an antique. It still works for the twice a year warmup. Same resistor only circuit currently used for 80v (hundreds transient) magneto and couple other sensors in a GY6 2 wheel vehicle. Goody-goody naysayers will probably have something else to say about this.

If this is for commercial use then your circuit is probably inadequate from a certification viewpoint. The main issue being one of liability and standards as discussed endlessly in the automotive megathread. IIRC record largest thread ever in this forum.

[Psi]

Thanks guys,

The channels also handle 5V analog signals, so 4.7V zeners are a no go.
5.1V zeners also cause too much current flow at 5V and limit the analog max to ~4.8V.
I would have to use a ~7V zener.
That would probably be fine as the internal clamping diodes can handle the difference through the 1k resistor (220R in picture).

[nuno]

I don't remember ever seeing this spec is on their datasheets, but an ATMEL appnote mentions that you want to keep current through internal pin protection diodes below 1mA.

Lower voltage (true zener) types have much softer knees and/or higher ESR.

So that's the difference I saw in 3 zeners I once traced the curves. I thought it was because some were very old types, but the ones with the softer curve are in fact < 5V while the 5V1 or 5V6 in the set is comparatively much more sharp.

[paulie]

The actual reason was because higher voltage "zeners" are not really zeners at all but "avalanche" diodes. A more significant difference is they have positive instead of negative temperature coefficient. At the dividing line we have best of both worlds: sharp knee and zero TC which is why 5.6v parts (6.2v with the back-to-back diode) are used in most precision references.

Anyway it will be interesting to see just how much more complicated and expensive we can manage in this "cost down" effort Compared to replacing the whole mess with a scientifically optimum simple resistor that is.

[nuno]

Some VI curves below just for the record.

When using that protection for an ADC input, beware that as the series resistor raises, it can reach the point where the internal sampling capacitor can no longer be charged up to input voltage, on varying signals.

[paulie]

When using that protection for an ADC input, beware that as the series resistor raises, it can reach the point where the internal sampling capacitor can no longer be charged up to input voltage, on varying signals.

Mhz analog speeds are rare so a small cap at the input will fix those ADC charge problems. In most cases digital inputs don't even need this because the few pf inherent in input pins will filter HF noise. For low frequency interference etc software debounce/noise routines suffice.

ps. Much bigger automotive issues deal with external high voltages and pulse coupling. Large physical size (high wattage) resistors and a good grounded metal case works wonders here which is required anyway for dirt and environmental protection.

[T3sl4co1l]

When using that protection for an ADC input, beware that as the series resistor raises, it can reach the point where the internal sampling capacitor can no longer be charged up to input voltage, on varying signals.

Mhz analog speeds are rare so a small cap at the input will fix those ADC charge problems. In most cases digital inputs don't even need this because the few pf inherent in input pins will filter HF noise. For low frequency interference etc software debounce/noise routines suffice.

No, he's talking about the zener going partially conductive and loading down the signal.  DC effect.

Anyway it will be interesting to see just how much more complicated and expensive we can manage in this "cost down" effort Compared to replacing the whole mess with a scientifically optimum simple resistor that is.

But a single resistor won't protect against ESD.

That's actually a partial lie; one can purchase 1206 size chip resistors rated for 10kV.  Needless to say, they can only be used in vacuum sealed potting compound, on thoroughly cleaned boards.  If your board is getting potted anyway, that's a very real possibility -- but if not, it's not much of a "cost-down" option, and you'll need to clamp the ESD somehow.

Tim

[paulie]

Pin capacitance and built-in clamp diodes do an excellent job on that. Assuming you don't allow long wires entering too far into the case. As far as the resistors expensive special purpose chip type would not be a consideration for me over 2 cent half watts or 5 cent one watters.

I am a little disappointed there hasn't been more uproar against my methods. No personal attacks or thread reports for safety violation. Guess the old days of low post count are over.

[paulie]

But a single resistor won't protect against ESD.

That's probably true. As I mentioned in my first comments here at least two or more in series would be recommended due to potential for high potentials floating around in the engine compartment.

[Niklas]

One problem with larger SMD packages is the mismatch in thermal expansion between the component and the PCB laminate itself. Tests have shown that a 2512 sized resistor and its soldering will not be able to survive a full lifetime of the car, even when placed in the driver's compartment and not under the hood. It does not have anything to do with leaded or lead free solder, both alloys failed, and with several types of plating on the PCB (ENIG, HASL...). The solution is to split up the resistance into several smaller packages, typically 0805 or 1206.

By only relying on increased series resistance, you will also make the signal more sensitive to component tolerances, leakage currents etc. It might be ok for a low frequency, low resolution signal, but the accuracy of a sensor will suffer and can go out out spec.