protection diodes

C

Chester

Guest
May be it's been asked a thousand times in the past decade, but here it
is...

ICs with protected I/O pins using internal clamp diodes are, usually,
limited to 25mA through these diodes. In my case, I just want to drive a
CMOS input (Vcc=5V) with a maximum input of +-12V from a sqaure wave
generator through a current limiting resistor to keep that current well
below this limit in the clamp diodes.

According to general specs, inputs cannot exceed Vcc+0.3V and Vss-0.3V. But
this is the voltage only. I assume this is to be sure there is no current
flowing through the internal diodes with 0.3V excess voltage. Stating any
higher voltage like 0.64 could be ok but 0.72 could lead to much higher
current and damage the chip.

But if I add a limiting resistor, the pin will be driven to Vcc+0.7V and
Vss-0.7V, apparently exceeding the spefication by 0.4V. Yet the current is
limited...

So what gives? Is this safe to do this?
Chester
 
On Wed, 26 May 2004 17:35:13 -0400, the renowned "Chester"
<mchester@directlan.net> wrote:

May be it's been asked a thousand times in the past decade, but here it
is...

ICs with protected I/O pins using internal clamp diodes are, usually,
limited to 25mA through these diodes. In my case, I just want to drive a
CMOS input (Vcc=5V) with a maximum input of +-12V from a sqaure wave
generator through a current limiting resistor to keep that current well
below this limit in the clamp diodes.

According to general specs, inputs cannot exceed Vcc+0.3V and Vss-0.3V. But
this is the voltage only. I assume this is to be sure there is no current
flowing through the internal diodes with 0.3V excess voltage. Stating any
higher voltage like 0.64 could be ok but 0.72 could lead to much higher
current and damage the chip.

But if I add a limiting resistor, the pin will be driven to Vcc+0.7V and
Vss-0.7V, apparently exceeding the spefication by 0.4V. Yet the current is
limited...

So what gives? Is this safe to do this?
Chester
Click to expand...
Bottom line, is that it is "safe", but there may be subtle issues with
operation of the chip (increased Idd, small amounts of current
spilling out of nearby inputs, etc.).

Best regards,
Spehro Pefhany
--
"it's the network..." "The Journey is the reward"
speff@interlog.com Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog Info for designers: http://www.speff.com
 
Hi Chester,

That's exactly what I had asked two manufacturers a few months ago. One said
that they couldn't tell (ouch), another replied that only the current into the
pin matters which made sense. After all, this is just a diode path you run into
and there is a certain limit that ensures you stay far enough away from
latch-up. Of course these two responses meant the first mfg had to be spec'd
out and miss out on the business. Pretty much for good, that is.

We can only release parts where all parameters are specified, and if not where
the manufacturer provides a proper "exemption". Which means that if they cannot
provide answers because they skimp on app engineers they will lose business. I
bet their top management doesn't even know what hit them once they go under.

Don't know your CMOS but much of the 4000 series is limited to 10mA.

Regards, Joerg

http://www.analogconsultants.com
 
On Wed, 26 May 2004 23:27:55 GMT, the renowned Joerg
<notthisjoergsch@removethispacbell.net> wrote:

Hi Chester,

That's exactly what I had asked two manufacturers a few months ago. One said
that they couldn't tell (ouch), another replied that only the current into the
pin matters which made sense. After all, this is just a diode path you run into
and there is a certain limit that ensures you stay far enough away from
latch-up. Of course these two responses meant the first mfg had to be spec'd
out and miss out on the business. Pretty much for good, that is.

We can only release parts where all parameters are specified, and if not where
the manufacturer provides a proper "exemption". Which means that if they cannot
provide answers because they skimp on app engineers they will lose business. I
bet their top management doesn't even know what hit them once they go under.

Don't know your CMOS but much of the 4000 series is limited to 10mA.

Regards, Joerg

http://www.analogconsultants.com
Click to expand...
Ever get the feeling those "answers" are not worth much more than your
own judgement? I just dealt with this (contacting the manufacturer
because the part is used in a somewhat unusual way etc.) and they
(M*x*m) hemmed and hawed and came back with a "sure, that's okay".

There is an output with a series resistor connected to a >12V line
under some conditions (and with the power cycled). Well, I don't think
it will latch up because it's under the minimum for that by orders of
magnitude. Fortunately, it's a rare condition so I'm not too worried
about it, but if it was a normal operating condition I don't think I'd
be satisfied.


Best regards,
Spehro Pefhany
--
"it's the network..." "The Journey is the reward"
speff@interlog.com Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog Info for designers: http://www.speff.com
 
Hi Spehro,

Yes, I do get that feeling a lot. Especially if the app engineer appears to be much
less experienced than myself or in cases where it's really just a sales person.

Since many of my projects are medical or otherwise safety critical this kind of
stuff needs to come from the manufacturer in writing and not just over the phone.
That really separates the good from the ugly. Never really a problem with National,
AD or others who maintain knowledgeable app engineers. Then there are the "not so
good". I don't want to name them here because it would be derogatory.

There are some companies that used to be the bellwethers which have now deteriorated
to mere sales outlets. Some of them I used heavily in the 80's and that has dropped
to zilch. Meaning their products will not be used in production. I wrote to a few
CEOs but the responses showed they don't understand. Or don't want to.

Regards, Joerg

http://www.analogconsultants.com
 
Joerg wrote...
Don't know your CMOS but much of the 4000 series is limited to 10mA.
Click to expand...
In case anyone cares, that's wrong.

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)
 
On 26 May 2004 17:57:04 -0700, Winfield Hill
<Winfield_member@newsguy.com> wrote:

Joerg wrote...

Don't know your CMOS but much of the 4000 series is limited to 10mA.

In case anyone cares, that's wrong.

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)
Click to expand...

The early 4000s were notorious for latching up with even tiny
protective diode currents. Modern HC-type parts usually have latchup
thresholds of 100 mA or more, and some designs can't latch at all.

We've had lots of interesting latchup problems with recent vintage
analog or mixed-signal parts from Analog Devices, Burr-Brown, TI, and
National.

John
 
Winfield Hill wrote:

Joerg wrote...

Don't know your CMOS but much of the 4000 series is limited to 10mA.

In case anyone cares, that's wrong.

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)
Click to expand...
Hi Winfield,

10mA is what many datasheets state, for example the one for the Texas
Instrument CD40106B. Same for the CD4020B and others.

Regards, Joerg

http://www.analogconsultants.com
 
Your inquiry with manufacturers is interesting. The ICs I'm talking about
are the well known 4000 series, 74H series and Microchip PICs.
And about the possible latch-up problem: for my current design I limit the
supply current to about 20mA, so I don't think latch-up is possible this
way, and even if latch-up occurs, ICs won't auto-destruct.

Chester

"Joerg" <notthisjoergsch@removethispacbell.net> a écrit dans le message de
news:40B63661.FE6D7029@removethispacbell.net...
Winfield Hill wrote:

Joerg wrote...

Don't know your CMOS but much of the 4000 series is limited to 10mA.

In case anyone cares, that's wrong.

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

Hi Winfield,

10mA is what many datasheets state, for example the one for the Texas
Instrument CD40106B. Same for the CD4020B and others.

Regards, Joerg

http://www.analogconsultants.com
Click to expand...
 
Hi Chester,

Often the overload issue isn't just about latch up. Another concern to me is
whether the circuit continues to function in an orderly fashion. Without
choking within an undefined logic state for a while, as an example.

So I try to keep input current to the uA range. I wish manufacturers would
also spec a max safe capacitive discharge value. For example when the power
abruptly goes and there is a cap connected to an input.

Regards, Joerg

http://www.analogconsultants.com
 
John Larkin wrote...
Winfield Hill wrote:

Joerg wrote...

Don't know your CMOS but much of the 4000 series is limited to 10mA.

In case anyone cares, that's wrong.

The early 4000s were notorious for latching up with even tiny
protective diode currents. Modern HC-type parts usually have latchup
thresholds of 100 mA or more, and some designs can't latch at all.
Click to expand...
Indeed, but the official "spec" was 20 or 25mA to my memory. And
this was dramatically increased in the early 1970s timeframe.
Thankfully,because that's when I began using them by the 10000 lot.

We've had lots of interesting latchup problems with recent vintage
analog or mixed-signal parts from Analog Devices, Burr-Brown, TI,
and National.
Click to expand...
Effective high-current latchup-protection circuits suffer from a
few performance degradation issues: high input capacitance (e.g.
more than 2pF), and "high" input leakage (e.g. more than 0.1pA).

As for the NSC LMC662 (typical input bias current: 2 fA), we are
still trying to figure that one out. Hmm, maybe they *do* suffer
from low-pin-current supply-latchup susceptibility. I've not had
any trouble, but a few tests may be in order. I see the datasheet
says this, Absolute Maximum Ratings: Current at Output Pin ą18 mA,
Current at Input Pin ą5 mA. Yep, looks rather fragile! Watch out!

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top