Go Down

Topic: ADC and a zener  (Read 1 time) previous topic - next topic

LandonW

I'm taking a reading from a voltage divider and using 3.3vdc ADC. I'm wanting to limit the voltage at the ADC with a 3v zener.
My question is, would one clamp that input signal back to the negative side of the voltage divider or to battery neg.
The voltage divider does not share negative with the battery in my circuit.

septillion

Depending on the values use, that zener may mess things up...

Let's start with, why do you want to limit it?

Normally you jut use high enough resistor values and use normal diodes to clamp it against the rails.
Use fricking code tags!!!!
I want x => I would like x, I need help => I would like help, Need fast => Go and pay someone to do the job...

NEW Library to make fading leds a piece of cake
https://github.com/septillion-git/FadeLed

LandonW

To protect against voltage spikes.

septillion

Then I would certainly go for the clamping + a cap :) Zeners leak.
Use fricking code tags!!!!
I want x => I would like x, I need help => I would like help, Need fast => Go and pay someone to do the job...

NEW Library to make fading leds a piece of cake
https://github.com/septillion-git/FadeLed

LandonW

So to answer the question.  Clamp and cap to the voltage divider negative?

LandonW

Clamp and cap in series of parallel

septillion

A clamp diode to positive if you expect positive peaks, diode to GNDif you expect to go below GND. A lot of ADC already have them build in. But in order for it to work you must limit the current enough (aka, use high enough resistors). For example, a pin on an ATmega238p can clamp 1mA max. Stay well below that.

Cap goes between GND and the ADC. Forms a low pass with the resistors of the divider and gives a lower impedance for the ADC to read.
Use fricking code tags!!!!
I want x => I would like x, I need help => I would like help, Need fast => Go and pay someone to do the job...

NEW Library to make fading leds a piece of cake
https://github.com/septillion-git/FadeLed

LandonW

Thanks. It's a 400:1 divider and should be able to compensate for twice the "normal" input voltage. However voltage spikes can be much greater than twice a "normal" input. That's what I'm attempting to protect against.

Wawa

400:1 means you're measuring >1000volt?
Give us the full picture.

Forget about zener diodes.

You could use a 2-stage voltage divider, and clamp to the unregulated supply (5volt, 12volt) of that 3.3volt processor, with normal diodes. That would take care of positive spikes, negative voltage, even when the Arduino is off (unlike a zener).
Leo..


ReverseEMF

#10
Sep 21, 2018, 06:11 pm Last Edit: Sep 21, 2018, 06:23 pm by ReverseEMF
For protection techniques see:-
http://www.thebox.myzen.co.uk/Tutorial/Protection.html

Nice article!  But, there are two things it didn't cover--two things I've wondered about:

  • Some devices, in their datasheet, show a MAXIMUM input voltage of something like VDD + 0.6V.  And, other's it's more like VDD + 0.3V.   In the later case, wouldn't the clamping diode need to be a Schottky Diode?  A regular signal diode, such as a 1N914, would allow the voltage to go too, high, wouldn't it?
  • You show a 5.1V Zener (presumably protecting the input of a device powered by a 5V supply).  I've always thought this a tricky solution, because of voltage mismatch due to tolerances.  Even if the input voltage can go as much as 0.6V above the supply voltage, if the regulator is running at its lower tolerance limit, and that tolerance is 5%, that would be: 5V * 0.95 = 4.75V.  And if that Zener has a tolerance of 5%, putting it's highest voltage at 5.1 * 1.05 = 5.36V, that gonna be too high, because the ABSOLUTE MAXIMUM input voltage is 4.75V + 0.6V = 5.35V.  I mean, it's close, but too much of a Murphy temptation, right?  This isn't as much of a problem with the more modern regulators, like LP2950, with a 0.5% tolerance, but what about those cases where the spec is more like VDD + 0.3V?  I'm not a fan of protecting inputs with zener diodes, for this reason.  But, if I was to use a Zener, I would go with more like a 4.7V zener, like it shows in that DigiKey article.  But, then, that puts more of a load on the device driving that input.

Am I off on my thinking?

http://www.digikey.com/us/en/techzone/microcontroller/resources/articles/protecting-inputs-in-digital-electronics.html
Looks like DigiKey pulled a switch-r-roo on you!  This link no longer points to your intended destination.

UPDATE: I think I found where DigiKey hid that article: https://www.digikey.com/en/articles/techzone/2012/apr/protecting-inputs-in-digital-electronics
"It's a big galaxy, Mr. Scott"

Please DON'T Private Message to me, what should be part of the Public Conversation -- especially if it's to correct a mistake, or contradict a statement!  Let it ALL hang out!!

JohnRob

I personally would not use a zener right at the input pin.  As stated there are zener tolerances and zeners often do not have a very "sharp" transistor from not conducting to conducting.

One of the decision inputs is the energy of the incoming spike.  If there is enough energy to increase the board voltage through the clamp diodes then I would use a zener.

In either case i would add a resistor maybe 1k  or 500 ohms between the clamp and the ADC input.  This will limit any current resulting from the clamp going higher that the device powering voltage (i.e. 5V or 3.3V)


Regarding the 400:1 divider,  without knowing the actual resistances,  I would at least check the zener leakage current to see if it can adversely affect the divider accuracy.

John
Please do not PM me with thread based messages.  If your thoughts are worth responding,  the group should benefit from your insight.

Wawa

Zeners are useless as pin protection.
What does a zener do when the proccessor is off (VCC = 0volt).

I tested once the clamping voltage of an Uno pin by sending a higher voltage with a 1mA current limit into it.
Positive clamping and negative clamping was about 0.65volt, so clamping with a 1N4148 or 1N4004 is almost useless too.
Must use (low leakage) schottky diodes if you want to add extra clamping.

Clamping dumps the spike onto the power rail of the micro.
Must be sure that the decoupling can take that spike, and the rail can use the sustained current (if any).

I don't think there is much of a problem with the 400:1 divider OP uses.
Like to know the values before making any recommendations.
Leo..

Grumpy_Mike

Quote
A regular signal diode, such as a 1N914, would allow the voltage to go too, high, wouldn't it?
Yes you would think so but that is the system used in industry and we test it with an electro static zapping gun to test for compliance to the human body static model. It works, time and time again in the most demanding situations.

https://en.wikipedia.org/wiki/Human-body_model

ReverseEMF

Yes you would think so but that is the system used in industry and we test it with an electro static zapping gun to test for compliance to the human body static model. It works, time and time again in the most demanding situations.

https://en.wikipedia.org/wiki/Human-body_model
Amazing!  OK, I'm schooled  ;)
"It's a big galaxy, Mr. Scott"

Please DON'T Private Message to me, what should be part of the Public Conversation -- especially if it's to correct a mistake, or contradict a statement!  Let it ALL hang out!!

Go Up