Series Resistor and TVS diode placement

[mahen]

If I have to put ESD diode and current limiting series resistor to protect the IO pin of MCU from ESD and over current, which is the suggested way?

MCU Pin -------------- Series R -- TVS to GND -- Connector

or

MCU Pin -------------- TVS to GND -- Series R-- Connector

With the 2nd option, doesn't the R limit the surge into TVS needing only lower wattage ESD diode compared to 1st option?

-mjnk

 

[Klaus Kragelund]

If I have to put ESD diode and current limiting series resistor to protect the IO pin of MCU from ESD and over current, which is the suggested way?



MCU Pin -------------- Series R -- TVS to GND -- Connector



or



MCU Pin -------------- TVS to GND -- Series R-- Connector



With the 2nd option, doesn't the R limit the surge into TVS needing only lower wattage ESD diode compared to 1st option?

With the second option all your surge current is running in the series R and it would need to be either of very small resistor value or large power handling value to cope with the surge energy

This way:

MCU Pin ---- Series R ------External ESD diode ------ Series R -- TVS to GND -- Connector

TVS takes blunt of the surge pulse. Resistor limits current into series ESD diode. Second resistor limits current input MCU pin to avoid latchup (normally you need to stay below 1mA depending on the die technology)

Cheers

Klaus

 

[RobertMacy]

I used zeners as it limits the voltage to both +zener voltage (use for
example a 5.6V zener for 5V micro), and -0.7V.
Or transzorbs.

It also depends on speed required, and power environment, 24V, 110V,
230V etc..

Jim, be careful of using zeners as circuit protection. They turn on
EXTREMELY slowly and during that time can let a lot through.

For grins, take two 5.6V zeners and put in series 'facing each other' as
feedback on a high speed OpAmp. Then drive the thing and you won't get a
square wave out, you'll get an extremely 'spikey' square wave out. We're
talking audio frequencies, too. I've seen 20kHz spikes.

Do again with tranzorbs and you can really see the difference.

Or, you can do what we used to do, slightly turn the zener on ahead of
time, that helps. but takes a fast switching diode into the turned on
zener. Today, too many parts, then, the only way.

 

[RobertMacy]

With the second option all your surge current is running in the series R
and it would need to be either of very small resistor value or large
power handling value to cope with the surge energy

This way:

MCU Pin ---- Series R ------External ESD diode ------ Series R -- TVS to
GND -- Connector

TVS takes blunt of the surge pulse. Resistor limits current into series
ESD diode. Second resistor limits current input MCU pin to avoid latchup
(normally you need to stay below 1mA depending on the die technology)

Cheers

Klaus

How about RF beads instead of R's, they're almost the same physical size. ?

Then you get the equivalent of over 1k to 10k ohm series resistance, but
at the operating frequencies of the component, very low impedance, so you
don't end up compromising waveform shape, too much.

 

[RobertMacy]

And burn out the micro and diodes with any DC.

Right. is that before or after the esd diode burns up? or when the series
resistor burns up, what are these 0402's 1/64 w ??

I usually use series R and RF Bead, to get a 'well-rounded' response.

 

[RobertMacy]

..snip...
I have used zeners in big data networks that way, no problem.
Zeners have quite a bit of capacitance too.

If u want 100% protection use optocouplers.

What you can 'get away with' and what happens are two different things.
For example, I don't use an anti-static wrist band, nor am very careful
about handling components, no problem. Sure, no problem ...today! but
probably shortened their little lives so they'll fail in anywhere from 2,
6 months out to one year, instead of lasting 10 years.

And, optocouplers? not quite the panacea one would expect. kind of work
for AC mains isolation and some on voltage protection, except capacitive
coupling high voltage can still 'punch' through. and worst of all magnetic
pulses [even the pulses caused by HV discharge from HV protection], just
go right around an optocoupler. EMP's can do a LOT of damage. All a
manner of degree.

For total protection, I like to think in terms of analog filtering
components. the uP has C and limit [low Z], therefore place a high-Z in
series, then place a low-Z to GND, and then a high-Z out to the unknown
origin. [this is a judgment call, because using a low-Z here can result in
a spike of current injected into the GND plane and/or inject a current
pulse into adjacent circuitry] And, be sure to design for at least 3GHz
spectrum and you should be ok.

 

[RobertMacy]

...snip...
I usually use series R and RF Bead, to get a 'well-rounded' response.

[diode]C and L will resonate somewhere...

NO!, not if done CORRECTLY. Also, RF Beads are NOT L, they're very lossy,
very low Q inductors.

Again, DESIGN the solution. Don't 'throw' parts at the problem and wait
until it "worked this time."

 

[RobertMacy]

700$
eeh
Euro

Thank you. Accolade indeed.

Jan, if you ever want some of the LTspice models and designs for 'pure'
bypass filtering on uC chips done for n ADC mounted in and about
Bluetooth, let me know. When I say pure, I mean the bypass looks
resistive in AND out, all that happens is 'rerouting' the energy to GND.
That way, you don't get that pesky ringing on the power terminals caused
by the impedance of the bypass at some spectrum suddenly going up to 10,
and even 100 ohms reactance! Had to abandon many of the manufacturer's
recommended bypass components to do it, but gives BETTER performance.

 

[[email protected]]

This one, but perhaps add an extra R in case somebody connects it directy to a power source:
MCU Pin --- Series R1 -- zener to gnd -- series R2 ---- Connector

Disagree. The resistor between the TVS and I/O pin is better for the
reason stated (current limiting) and the capacitance of the TVS also
helps (assuming the R and C don't cause functional problems).

I used zeners as it limits the voltage to both +zener voltage (use for example a 5.6V zener for 5V micro), and -0.7V.
Or transzorbs.

So do unidirectional TVS diodes. They're a faster than zeners and
will take a lot more abuse. Low voltage zeners are horrible.

It also depends on speed required, and power environment, 24V, 110V, 230V etc..

Of course. The R gets problematic at high currents. ;-)

 

[mahen]

On a sunny day (Thu, 9 Jan 2014 04:20:32 -0800 (PST)) it happened mahen


<[email protected]>:







This one, but perhaps add an extra R in case somebody connects it directy to a power source:

MCU Pin --- Series R1 -- zener to gnd -- series R2 ---- Connector



I used zeners as it limits the voltage to both +zener voltage (use for example a 5.6V zener for 5V micro), and -0.7V.

Or transzorbs.



It also depends on speed required, and power environment, 24V, 110V, 230V etc..

Thanks Jan. may be a dump Q. Series R2 b/w zener and connector already limiting the current into the pin, correct? Why series R1 again?

 

[Klaus Kragelund]

Tell it to the PLC guys who use optos by the thousands.

I have used optos too, where spikes made other solutions impossible.

They use optos for galvanic isolation, to combat common mode noise, since they handle signals in industrial environments with a lot of ground bounce etc.

For DM noise, use the tranzorb, resistor, diode, resistor combo

Cheers

Klaus