Go Down

Topic: Protecting Analog Inputs: How often and how bad will it fail? (Read 2319 times) previous topic - next topic

tms8c8

I'm looking at developing a commercial product and I need to measure the voltage across an n-channel MOSFET when the MOSFET is conducting (i.e., the voltage between source and drain). This voltage is expected to be between 0.02V and 0.12V. With AREF at 1.1V, this is measurable and safe for the inputs. However, when the MOSFET is not conducting, the Arduino input sees 48 volts. When the MOSFET is "off", I do NOT need to measure any voltage, I just want to prevent my analog pins from being damaged.

My thought was to use a zener diode. When the voltage is less than the zener voltage (1.1V) then the inputs see the real voltage. When the voltage rises above the zener voltage, it conducts and the input sees the zener voltage. How often do zener diodes (chosen properly for current and voltage) fail? If it fails, what kind of damage can I expect to the board? A single pin fried or something more serious?  Would using a relay to completely isolate the input be a better choice? Unfortunately, I need to protect 12 analog pins so using relays represents a non-negligible increase in cost and PCB size.

Any thoughts? Is there a better way to do this?

CrossRoads

Why not use a P-channel instead, sourcing current into a 10K resistor to Gnd.
When the P-channel opens up, the resistor pulls the analog pin low.

Oherwise, post a schematic so we can see where this 48V is coming from.
Designing & building electrical circuits for over 25 years.  Screw Shield for Mega/Due/Uno,  Bobuino with ATMega1284P, & other '328P & '1284P creations & offerings at  my website.

be80be

I would limit current to under the clamping diodes spec. At the 48 volts

tms8c8

#3
Jan 19, 2013, 04:29 am Last Edit: Jan 19, 2013, 04:33 am by tms8c8 Reason: 1

Why not use a P-channel instead, sourcing current into a 10K resistor to Gnd.
When the P-channel opens up, the resistor pulls the analog pin low.

Oherwise, post a schematic so we can see where this 48V is coming from.


I need to supply up to 10 amps (the load varies but can be as small as ~5 ohms). I'm controlling a variety of current hogging devices and I'm using a sort-of "multiplexing" system where I switch grounds on/off with N-channel MOSFETs and I switch V+ on/off with P-channel MOSFETs. I'm working under contract as a "consultant" and I can't provide too many specifics. I'm sorry - I know ambiguous questions get ambiguous answers but that's why I tried to ask a specific question regarding the reliability of zeners.

Basically, I am using the source-drain resistance of the MOSFET as a current-sense resistor. A crude measurement of the current is all that is necessary.

Jack Christensen

I certainly can't quote MTBF specs or anything, but regarding the reliability of Zener diodes, I would expect them to be more or less as reliable as any other semiconductor, which is to say very reliable. Like anything else, they will tend to be more reliable the more conservatively they are used, i.e. not close to their limits.

Not sure I'd lose too much sleep over it, unless every other part in the project is orders of magnitude more reliable than semiconductors. I don't recall seeing too many reliability specs in datasheets; if hard numbers are needed, contacting the manufacturer is probably in order.
MCP79411/12 RTC ... "One Million Ohms" ATtiny kit ... available at http://www.tindie.com/stores/JChristensen/

SirNickity

Well, a Zener is only going to clamp the voltage if it can bring the supply side down to its cutoff voltage.  In other words, it will try to pull your 48v rail to 5v (or whatever).  If it's a high-current rail, it will probably fail to do so, catastrophically.  If it succeeds (a BIG Zener!), your 48v rail, isn't.

The solution to this is to have a series resistance before the Zener.  But then you have resistance (other than the current sensing "resistor" formed by the Rds of the FET) in your sensing circuit that you have to deal with.

If this is not news to you, you've found some suitable way to account for that series resistance, and you're able to to keep the shunt current within limits of the part, then I'm with the other guys here.  The manufacturer should be able to supply reliability specs.  :-)

tms8c8

#6
Jan 19, 2013, 06:12 am Last Edit: Jan 19, 2013, 06:14 am by tms8c8 Reason: 1

Well, a Zener is only going to clamp the voltage if it can bring the supply side down to its cutoff voltage.  In other words, it will try to pull your 48v rail to 5v (or whatever).  If it's a high-current rail, it will probably fail to do so, catastrophically.  If it succeeds (a BIG Zener!), your 48v rail, isn't.

The solution to this is to have a series resistance before the Zener.  But then you have resistance (other than the current sensing "resistor" formed by the Rds of the FET) in your sensing circuit that you have to deal with.

If this is not news to you, you've found some suitable way to account for that series resistance, and you're able to to keep the shunt current within limits of the part, then I'm with the other guys here.  The manufacturer should be able to supply reliability specs.  :-)


Thanks, SirNickity. You are quite correct I need a resistor before the zener! :) However, it has a negligible contribution to the "measurement circuit" because the zener is non-conducting (except for leakage) when I'm measuring. It's not until the voltage rises (the MOSFET turns off) that the zener starts to conduct and then there is a voltage drop on the resistor. (No current means no voltage drop).

Just to be sure I was correct in my thinking, I tested in multi-sim ;)


Thanks Jack! I figured they would be reasonably reliable, probably better than any relay I would put in there, but I don't have much practical experience with zeners for some reason...

be80be

#7
Jan 19, 2013, 06:14 am Last Edit: Jan 19, 2013, 06:19 am by be80be Reason: 1
Quote
Basically, I am using the source-drain resistance of the MOSFET as a current-sense resistor. A crude measurement of the current is all that is necessary.


That's not a real good idea you could read the copper trace resistance would work better and save you a headache

That's the way cheap DVM read it better one's use a wire thats made to have constant reading with temperature.

dc42

A resistor of 100K or 220K between the mosfet and the analog input will provide sufficient protection, and will not significantly affect the reading unless the input signal is changing quickly. There is an Atmel application note that says the pin protection diodes are good for up to 2mA.
Formal verification of safety-critical software, software development, and electronic design and prototyping. See http://www.eschertech.com. Please do not ask for unpaid help via PM, use the forum.

silverxxx

if you only measure 0.12 volts, you can just use a normal diode to clamp the voltage (or 2). they should be cheaper then a zener diode. Most diodes fail short circuit but that is impossible after 220kohms.

SirNickity

Is there any way you could provide a drawing of just the 48v power rail, Zener, transistor, and analog pin?  Maybe I'm not understanding your intention.  It sounded to me like your V+ rail (48v) would flow through the FET and the load, to ground.  In order to keep the analog input safe, you would need to clamp that 48v rail to less than 5v.  The only way I see to do this is to put the Zener between V+ and Gnd -- that is, across the transistor and load.  Of course, doing this will guarantee that your 48v rail either never reaches 48v, or that the Zener will vaporize shortly after power-up.  So, I hope I'm misunderstanding what you intend to do.  :-)

fungus


Of course, doing this will guarantee that your 48v rail either never reaches 48v, or that the Zener will vaporize shortly after power-up. 


Not if you put a resistor in... eg. limit the current arriving at the Zener to 0.5 mA.

No, I don't answer questions sent in private messages (but I do accept thank-you notes...)

Go Up