High voltage protection for ATD module

[RChapman]

This message contains an entire topic ported from a separate forum. The original message and all replies are in this single message. We have seeded this new forum with selected information that we expect will be of value to you as you search for answers to your questions.

Posted: Tue Jan 10, 2006 12:02 pm  

Hi,
I am using ATD module of MC9S12DJ64.
Appnote AN2429 mentions about on-chip protection diodes, which are supposed to protect the analog input from accidental occurance of high voltage on the pin. It also describes an example which discusses occurance of -12V on the pin. In this case, voltage on the pin comes out to be -0.7V, whereas the absolute maximum rating for any pin is -0.3V. Is that OK?

Does this mean, I can use the circuit for overvoltage protection if I limit the current to 2.5mA max?
Is this available on all the port pins (to be used as digital inputs)?

Posted: Mon Jan 16, 2006 4:45 am    

HI

I have checked the ATD by loading it to +12V and -12V. Of course I have taken care that the current does not exceed 2.5mA

For positive overvoltage, it gets clamped to about 5.6V and for negative overvoltage, it gets clamped to -0,7V.

Posted: Mon Jan 16, 2006 6:57 am    

If you intend to connect an ATD input to the "outside world", either directly or via a resistive voltage divider, you will need to allow for possible electrostatic transients in addition to other over-voltage considerations, and I think it would be much safer to adopt at least the following precautions:

1. Install Schottky diodes between the analog input and each supply rail - these are capable of handling much greater currents than the internal protection diodes, and will start conducting before the internal diodes. (If surface mount, you could use a suitable dual diode package, with anode-to-cathode junction connected to the input.)

2. If you do not have a voltage divider, include a series resistor (say 10k) to limit current.

3. If possible, include a capacitor between the analog pin and ground, to help absorb short transients. The size of the capacitor would be governed by your analog sampling rate.

An alternative arrangement would be to buffer the analog input using a "rail-to-rail" op amp, operating from the same supply rail as the MCU. This may also potentially achieve greater conversion accuracy (depending on amplifier offset voltage), because it provides a lower source impedance to the ATD input. Use a "voltage follower" configuration for positive input voltages, or an "inverting amplifier" configuration to handle negative input voltages.

For the protection of an unbuffered digital input pin (low speed and active low assumed), I would normally do the following:

1. Connect an external pull-up resistor from the input pin to MCU supply rail, say 10k.

2. Connect a Schottky diode in series, with anode connected to the input pin. The cathode of the diode is now the external input. (An ordinary small signal silicon diode should be OK if you are using a 5V MCU supply rail.)

3. Additional protection would be provided by a Zener diode (400 mW) between the cathode of the diode and ground (Zener anode). If the normally applied input voltage was, say nominally 12 volts, I might choose a 22V Zener type.

Regards,

Posted: Sat Jan 21, 2006 9:35 am    

New modifications I have done to my hardware after our communication:

Analog Inputs:

1. Cannot afford to use external schottky diodes (when internal diodes are present).
2. I am using external RC low pass filter, R= 10K - this also acts as current limiting resistor. C = 0.22uF - this will prevent input from small transients.
Any further suggestions?

Digital Inputs:
1. My application has active high as well as active low inputs.
2. Right now, I plan to use potential divider - to scale down 12V inputs to 5V level, and RC filter to take care of switch debounce.
3. Over / Under-voltage care will be taken by on-chip diodes.
Any more suggestions?

Thanks in advance and Regards,

Posted: Sun Jan 22, 2006 3:53 am    

When considering protection for MCU inputs (or the inputs of any chip for that matter), I would usually consider that there are three possible categories of interface -

1. The input is derived from the output of another device on the same PCB.
2. The input is derived from the output of another device on a different PCB, but within the same enclosure.
3. The input is derived from a source outside the equipment enclosure, what I have referred to as the "outside world".

For the first category, no additional protective measures may be necessary, other than to match voltage ranges with a voltage divider. For an analog input of the second category, and when a voltage divider is not required, you might include an input filter and perhaps a further shunt resistor to ground on the input side of the filter - assuming the other board provides a low output impedance drive, as it should, for the source signal.

The suggestions of my previous post would apply to the third category, where the designer has little control over what an input may be (accidently) subjected to. You will need to decide in which category your requirements fall.

The internal protection diodes are of very small geometry, and have a correspondingly low current rating (2.5 mA). For a robust interface, my personal preference is not to rely on them. So the necessary protective measures to limit voltage excursions would be external to the MCU. I think I would prefer that the repair of a PCB did not involve the replacement of a 112- or 80-pin SM package.
Now to some specific points you have raised -

Quote:

1. Cannot afford to use external schottky diodes (when internal diodes are present).

If you are speaking in a monetary sense, you need to compare the cost of the protection with the replacement cost of the device you are trying to protect. Since you are using the HCS12, I presume this is a "higher end" application.

I cannot see any technical reason why external diodes should not be connected in parallel with internal protection diodes. A Schottky diode will commence conduction at a voltage 0.3-0.4V, whereas the internal diode commences conduction at 0.6-0.7V, as you have found.

Quote:

Digital Inputs: 1. My application has active high as well as active low inputs. 2. Right now, I plan to use potential divider - to scale down 12V inputs to 5V level, and RC filter to take care of switch debounce.

For external active high inputs, you might consider using a bi-polar transistor (NPN) as a buffer, with a voltage divider at the base connection to control the switching threshold level. The collector would connect to the input pin of the MCU, with either an external pull-up resistor, or the internal pull-up enabled. The alternative would require a voltage divider, as you propose, but also a pair of Schottky diodes for the reasons stated above.

I assume your switches are externally connected, otherwise you could operate at the MCU supply level.

I would not recommend using a CR circuit for switch debounce. This would imply a slowly changing input, so the input circuit would need to incorporate a Schmitt trigger, with hysteresis (not sure if this is the case for HCS12 inputs). I guess the debounce period would be somewhere between 20-100 milliseconds, so the capacitance values would be relatively large. If a mechanical switch is used, you will also have to consider the minimum required current through the switch contact, dependent on the contact materials in the switch - this could amount to "a few milliamps" for acceptable long term reliability. But that is another subject. In my opinion, it is far simpler to allow for switch debounce within the firmware.

I hope this clarifies some of the issues.

Regards,