Surge / overvoltage input protection?

[MRW]

Hi. I've read some articles about surge input protection; I think it
was regarding switches. Basically, they had a diode's anode connected
to the input terminal and the cathode connected to the voltage source.
It looks like this:

http://bayimg.com/kaebDaaBj

My question is that most diode datasheets that I've seen shows the
minimum forward voltage drop as somewhere around 700mV. Another article
that I read (I think this is about opamps this time), mentioned that
the input is capable of handling a maximum signal amplitude that is VCC
+ 0.3V.

In this situation, does that mean that I would need a diode with a
lower forward voltage drop?

Also, I seem to be confused as to how this type of diode protection
scheme works. Using the picture link, if the signal is higher than the
VCC voltage by an amount greater than the diode forward voltage, then
the diode conducts. So that would mean that the input is now at VCC
voltage levels? If so, then wouldn't the diode "open" up again since
the anode is now at the same voltage level as VCC?

Thanks!



--

 

[John Popelish]

Hi. I've read some articles about surge input protection; I think it
was regarding switches. Basically, they had a diode's anode connected
to the input terminal and the cathode connected to the voltage source.
It looks like this:

http://bayimg.com/kaebDaaBj

Often a second diode is connected between the input and the
ground rail to do the same thing for negative spikes.

My question is that most diode datasheets that I've seen shows the
minimum forward voltage drop as somewhere around 700mV. Another article
that I read (I think this is about opamps this time), mentioned that
the input is capable of handling a maximum signal amplitude that is VCC
+ 0.3V.

Usually, the 300 mV limit is an operational one (possible
malfunctions if exceeded), and there is a higher absolute
maximum (possible damage if exceeded) spec.

In this situation, does that mean that I would need a diode with a
lower forward voltage drop?

A lower drop diode, such as a Schottky or germanium, is
sometimes used, but these also have higher leakage current,
so may not by useful for very low current signals.

But in many cases, the 300 mV limit can be exceeded a bit,
without damage, if the input current is limited. This might
be accomplished by using an ordinary silicon diode (that may
drop a volt or so while it is passing a surge current) and
adding a current limiting resistor between the diode and the
chip input. Sometimes it is possible to choose such a
resistor that will safely limit the surge current from a
source 1 volt higher then Vcc but below enough to not
interfere with normal signals.

Also, I seem to be confused as to how this type of diode protection
scheme works. Using the picture link, if the signal is higher than the
VCC voltage by an amount greater than the diode forward voltage, then
the diode conducts.

Right. And if the surge source impedance is considerably
higher than the impedance of the Vcc source, most of the
rest of the surge voltage is dropped across the surge source
impedance. However, if the Vcc source impedance is high,
the surge may just jack up Vcc to damaging levels. For this
reason, some designs use a zener diode to ground, that
breaks down and conducts the surge current to ground when
the voltage gets close to Vcc. I have seem 4.8 volt zeners
used with 5 volt Vcc rails, this way, especially for logic
signals that don't have to go all the way to Vcc to work.
For negative spikes, the zener acts like a normal diode in
forward conduction. The weakness of the zener approach is
that it lets the input go all the way to zener break down
voltage, even if the chip being protected is unpowered. In
that case, the chip ends up seeing an input that is a lot
more than a volt higher than its Vcc.

So that would mean that the input is now at VCC
voltage levels?

No, Vcc plus the diode forward drop voltage.

 

[Eeyore]

Hi. I've read some articles about surge input protection

And why do you think your equipment's inputs needs protecting from these
'surges' ?

Graham

 

[MRW]

Often a second diode is connected between the input and the ground
rail to do the same thing for negative spikes.


Usually, the 300 mV limit is an operational one (possible
malfunctions if exceeded), and there is a higher absolute maximum
(possible damage if exceeded) spec.


A lower drop diode, such as a Schottky or germanium, is sometimes
used, but these also have higher leakage current, so may not by
useful for very low current signals.

But in many cases, the 300 mV limit can be exceeded a bit, without
damage, if the input current is limited. This might be accomplished
by using an ordinary silicon diode (that may drop a volt or so while
it is passing a surge current) and adding a current limiting resistor
between the diode and the chip input. Sometimes it is possible to
choose such a resistor that will safely limit the surge current from
a source 1 volt higher then Vcc but below enough to not interfere
with normal signals.


Right. And if the surge source impedance is considerably higher than
the impedance of the Vcc source, most of the rest of the surge
voltage is dropped across the surge source impedance. However, if
the Vcc source impedance is high, the surge may just jack up Vcc to
damaging levels. For this reason, some designs use a zener diode to
ground, that breaks down and conducts the surge current to ground
when the voltage gets close to Vcc. I have seem 4.8 volt zeners used
with 5 volt Vcc rails, this way, especially for logic signals that
don't have to go all the way to Vcc to work. For negative spikes, the
zener acts like a normal diode in forward conduction. The weakness
of the zener approach is that it lets the input go all the way to
zener break down voltage, even if the chip being protected is
unpowered. In that case, the chip ends up seeing an input that is a
lot more than a volt higher than its Vcc.


No, Vcc plus the diode forward drop voltage.

Thanks, John!


--

 

[MRW]

And why do you think your equipment's inputs needs protecting from
these 'surges' ?

Graham

Hi Graham. Oh, I was just curious. I figured it'd come handy sometime
in the future when I encounter something like this.


--

 

[Eeyore]

Hi Graham. Oh, I was just curious. I figured it'd come handy sometime
in the future when I encounter something like this.

Powerline 'surges' (let's call them transients shall we?) if that's what you
meant are very unlikely to be an issue for equipment inputs (or outputs).

The biggest problems for your ins and outs are likely to be electrostatic
discharges and RF pickup.

Graham

 

[MRW]

Powerline 'surges' (let's call them transients shall we?) if that's
what you meant are very unlikely to be an issue for equipment inputs
(or outputs).

The biggest problems for your ins and outs are likely to be
electrostatic discharges and RF pickup.

Graham

Gotcha. I was actually thinking more of the CMOS switch application
notes that I've seen. They had were depicting a sine wave input that
was much greater than VCC. One solution they mention was using diodes.
I've probably seen that configuration before in some textbook, but I
guess I paid attention to them the second time around. Thanks!


--

 

[Eeyore]

Gotcha. I was actually thinking more of the CMOS switch application
notes that I've seen. They had were depicting a sine wave input that
was much greater than VCC. One solution they mention was using diodes.
I've probably seen that configuration before in some textbook, but I
guess I paid attention to them the second time around. Thanks!

There's a couple of things worth noting here.

Firstly it's not a good idea to use CMOS switches with voltages larger than
their power supplies !

Secondly, mamy ICs have an internal 'parasitic' diode structure to the supply
rails on input nodes. Combined with a modest value current limiting resistor
these may be able to handle at least transient excursions beyond the power
supply voltage.

If you regularly expect excessive voltage inputs then it's a good idea to fit
additioanl diodes to deal with it. Do bear in mind that the input has to be
current limited for these to be effective.

It's good to see you're aware of this issue.

Graham

 

[Manoj]

There's a couple of things worth noting here.

Firstly it's not a good idea to use CMOS switches with voltages larger than
their power supplies !

Secondly, mamy ICs have an internal 'parasitic' diode structure to the supply
rails on input nodes. Combined with a modest value current limiting resistor
these may be able to handle at least transient excursions beyond the power
supply voltage.

If you regularly expect excessive voltage inputs then it's a good idea to fit
additioanl diodes to deal with it. Do bear in mind that the input has to be
current limited for these to be effective.

It's good to see you're aware of this issue.

Graham- Hide quoted text -

- Show quoted text -

Dear friends

Surges may come due to many reasons. they can be as high as hundred of
volts. Suppose the connection is made to wires going outside the
equipment. these wires might be long. the longer the worst is the
case. when the switch opens (the worst condition is opening and not
closing), due to inductance of the wire, there will be a spike good
enough to damage the cmos ics. surges can also be due to noisy
envoirnment.

please correct me if i am wrong.

the internal protective diodes of the cmos are not very great. if you
expect surges / spikes you should always put external diodes. but a
series resistor before the diode is also a must. without this series
resistance a spike of >50 Volts would mean the whole current passes
through the diode with negliglible source resistance. this can damage
the diode itself.

with regards

manoj

 

[Eeyore]

Dear friends

Surges may come due to many reasons. they can be as high as hundred of
volts. Suppose the connection is made to wires going outside the
equipment. these wires might be long. the longer the worst is the
case. when the switch opens (the worst condition is opening and not
closing), due to inductance of the wire, there will be a spike good
enough to damage the cmos ics. surges can also be due to noisy
envoirnment.

please correct me if i am wrong.

You're quite correct about that possibility.

the internal protective diodes of the cmos are not very great.
Correct.


if you
expect surges / spikes you should always put external diodes. but a
series resistor before the diode is also a must. without this series
resistance a spike of >50 Volts would mean the whole current passes
through the diode with negliglible source resistance. this can damage
the diode itself.

Exactly.

Graham