Circuit ideas for undervoltage protection?

[Dirk Leber]

I recently became aware of a problem which I believe is seriously
underestimated in a lot of designs:
For most standard integrated circuits manufacturers rate the min/max
voltage levels at the pins with something like -0.3V to Vcc + 0.3V,
which is probably derived from the idea of having schottky diodes,
either internal or external, clamping the pin to the above levels. Which
seems so obvious proves problematic when taking a closer look at various
schottky diode datasheets. The forward voltage drop is heavily depending
on the forward current which can easily result in voltage drops of more
than 1 volt. Inacceptable with the above mentioned absolute maximum
ratings for the "protected" integrated circuit.
Only very few schottky diodes guarantee Vfw values of less than 0.3 Volt
even at very low forward current. Apparently there seem to be two
qualities of a schottky diode that seem to be incompatible. Low forward
voltage drop values are apparently not achievable together with high
maximum reverse voltages, which resembles a problem anywhere close to
things like driver lines. How about an example:

TI's hot swap controller TPS249x can safely drive a FET as high side
switch by survising the drain-source current via a shunt resistor and
additionally doing something like a power calculation for the FET by
multiplying the current by the drain-source voltage drop. For measuring
this voltage drop it uses an additional pin OUT which has to be
connected to source. So far for the background.
The pin in question (OUT) since being connected directly to "the world
outside" will have to withstand all the nasty things like surge/burst
pulses, reversed polarity caused by users etc. As a little help, you
might want to add maybe a suppressor diode together with a small serial
resistance to limit currents. Not to forget the must-have schottky diode
to ground against erroneous negative voltages applied to the drivers
outside.
Still, since you cannot spend too much voltage drop over that serial
resistor, the resulting current through the schottky diode will grow to
something like 1.3 ampere when applying negative voltages up to the
suppressor diodes clamping voltage (let's say 40 volt), limited by a
serial resistance of 30 ohm.
The question now is, which schottky diode to choose:
For example Vishay's SS2H10 that is able to stand a reverse voltage of
up to 100V would have a forward voltage drop of approx. 0.68 Volt. The
OUT pin would be driven far out of spec.
The opposite direction would be something like Vishay's SL44 whith a
impressing forward voltage of only approx. 0.31 Volt at 1.3 ampere. The
problem here is the low reverse voltage of only 28 Volt.

Did anybody solve this quest in the past? Any alternative protective
circuit ideas?

Cheerio,
Dirk Leber
HEITEC AG

 

[w2aew]

Maybe a simple answer... If you use the SL44 to clamp a given input to
VCC and GND, then each diode would never see 28V reverse bias, because
the 'other' diode connected to the 'other' rail will clamp the voltage
to .3V of the rail. Unless you're in a system where VCC>=28V, the
diodes won't see a reverse voltage larger than VCC+.3V.

 

[Jim Thompson]

On Mon, 15 Aug 2005 18:50:56 +0200, Dirk Leber <amidumb@orwhat.no>
wrote:

I recently became aware of a problem which I believe is seriously
underestimated in a lot of designs:
For most standard integrated circuits manufacturers rate the min/max
voltage levels at the pins with something like -0.3V to Vcc + 0.3V,
which is probably derived from the idea of having schottky diodes,
either internal or external, clamping the pin to the above levels.

No. This SPEC is derived from requiring that ESD structures NOT be
forward biased other than by ESD events.

Which
seems so obvious proves problematic when taking a closer look at various
schottky diode datasheets. The forward voltage drop is heavily depending
on the forward current which can easily result in voltage drops of more
than 1 volt. Inacceptable with the above mentioned absolute maximum
ratings for the "protected" integrated circuit.
Only very few schottky diodes guarantee Vfw values of less than 0.3 Volt
even at very low forward current. Apparently there seem to be two
qualities of a schottky diode that seem to be incompatible. Low forward
voltage drop values are apparently not achievable together with high
maximum reverse voltages, which resembles a problem anywhere close to
things like driver lines. How about an example:

TI's hot swap controller TPS249x can safely drive a FET as high side
switch by survising the drain-source current via a shunt resistor and
additionally doing something like a power calculation for the FET by
multiplying the current by the drain-source voltage drop. For measuring
this voltage drop it uses an additional pin OUT which has to be
connected to source. So far for the background.
The pin in question (OUT) since being connected directly to "the world
outside" will have to withstand all the nasty things like surge/burst
pulses, reversed polarity caused by users etc. As a little help, you
might want to add maybe a suppressor diode together with a small serial
resistance to limit currents. Not to forget the must-have schottky diode
to ground against erroneous negative voltages applied to the drivers
outside.
Still, since you cannot spend too much voltage drop over that serial
resistor, the resulting current through the schottky diode will grow to
something like 1.3 ampere when applying negative voltages up to the
suppressor diodes clamping voltage (let's say 40 volt), limited by a
serial resistance of 30 ohm.
The question now is, which schottky diode to choose:
For example Vishay's SS2H10 that is able to stand a reverse voltage of
up to 100V would have a forward voltage drop of approx. 0.68 Volt. The
OUT pin would be driven far out of spec.
The opposite direction would be something like Vishay's SL44 whith a
impressing forward voltage of only approx. 0.31 Volt at 1.3 ampere. The
problem here is the low reverse voltage of only 28 Volt.

Did anybody solve this quest in the past? Any alternative protective
circuit ideas?

Cheerio,
Dirk Leber
HEITEC AG

Nothing comes for free. The "impressing" forward voltage comes at a
cost of slowed speed.

...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona Voice480)460-2350 | |
| E-mail Address at Website Fax480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |

I love to cook with wine. Sometimes I even put it in the food.

 

[Don Klipstein]

In article <ddqh05$74s$2@news1.transmedia.de>, Dirk Leber wrote:

I recently became aware of a problem which I believe is seriously
underestimated in a lot of designs:
For most standard integrated circuits manufacturers rate the min/max
voltage levels at the pins with something like -0.3V to Vcc + 0.3V,
which is probably derived from the idea of having schottky diodes,
either internal or external, clamping the pin to the above levels. Which
seems so obvious proves problematic when taking a closer look at various
schottky diode datasheets. The forward voltage drop is heavily depending
on the forward current which can easily result in voltage drops of more
than 1 volt. Inacceptable with the above mentioned absolute maximum
ratings for the "protected" integrated circuit.

This voltage limit is when from a low impedance source. Provided
current is adequately limited, you can violate this voltage limit vith
whatever voltage causes a small amount of current to flow through internal
diodes.

This current limit is typically 1 milliamp. Exceeding 1 milliamp can be
problematic since internal diodes are often parasitic ones that are parts
of unintentional parasitic SCRs, and if one of those SCRs turns on then
you usually get bonding wires ehaving as fuses.

This current limit may not be in the datasheet, but in some sort of
supplemental notes for the IC's family.

A way to get protection: Use external diodes for clamping, and then add
a resistor in series with the IC pin after the point the external diode
does its clamping. The resistor value would be the worst case external
diode voltage drop minus .3 volt, divided by whatever the maximum safe
input current is. If the external diode drops 1 volt and the max. safe
current is 1 mA, then this would indicate a 700 ohm resistor - next higher
common value is 820 ohms. Maybe use 1K ohms to be safe.

- Don Klipstein (don@misty.com)

 

[John Fields]

On Mon, 15 Aug 2005 10:24:02 -0700, Jim Thompson
<thegreatone@example.com> wrote:

Nothing comes for free.

---
My first girl friend did.

--
John Fields
Professional Circuit Designer

 

You need to bear in mind that the current flowing through the
protection diodes may not go directly to the supply rails, and may
provoke the integrated circuit into deviating from the data sheet.

I ran into this when developing a quadruple ramp A/D converter back in
1979 - it was good to about 14-bits once I clamped the ramp with a
dscrete Schotty diode, rather than relying on the free protection
diodes on an IC to do the job - and the current involved was much less
than a milliamp.

Using an external diode and a current limiting resistor is an excellent
scheme, but I wouldn't like to rely on the behaviour of the circuit
being protected while it is being protected - it won't blow up, but it
may behave oddly.

------------
Bill Sloman, Nijmegen

 

[Joerg]

Hello Dirk,

For most standard integrated circuits manufacturers rate the min/max
voltage levels at the pins with something like -0.3V to Vcc + 0.3V,
which is probably derived from the idea of having schottky diodes,
either internal or external, clamping the pin to the above levels. ...

Mfgs usually spec two limits and not just that 300mV limit: The voltage
applied to a pin from a source with supposedly infinitely small source
resistance, then the maximum current into a pin.

If you source or sink 2mA into a pin and the manufacturer states under
abs max that you should stay below 10mA then the voltage of the pin will
go above that 300mV limit. Unless there are Schottkys inside, something
that would be pretty unusual these days.

There recently was a lengthy discussion started by Chris Carlen here in
the group. That is also where ESD measures were discussed. Excessive
current into a substrate diode is what causes grief so that needs to be
limited.

Regards, Joerg

http://www.analogconsultants.com

 

[josh lawton]

Dirk,

You should contact the manufacturer. They will provide you with some
application support. I think the out pin is a high impeadant pin used
for regulating the current. Shorting out to battery should be no prolem
for the pin, it will cause you to loose current limitation. Shorting to
gnd should also be no problem. Talk to an application engineer.

Josh

Dirk Leber wrote:

I recently became aware of a problem which I believe is seriously
underestimated in a lot of designs:
For most standard integrated circuits manufacturers rate the min/max
voltage levels at the pins with something like -0.3V to Vcc + 0.3V,
which is probably derived from the idea of having schottky diodes,
either internal or external, clamping the pin to the above levels. Which
seems so obvious proves problematic when taking a closer look at various
schottky diode datasheets. The forward voltage drop is heavily depending
on the forward current which can easily result in voltage drops of more
than 1 volt. Inacceptable with the above mentioned absolute maximum
ratings for the "protected" integrated circuit.
Only very few schottky diodes guarantee Vfw values of less than 0.3 Volt
even at very low forward current. Apparently there seem to be two
qualities of a schottky diode that seem to be incompatible. Low forward
voltage drop values are apparently not achievable together with high
maximum reverse voltages, which resembles a problem anywhere close to
things like driver lines. How about an example:

TI's hot swap controller TPS249x can safely drive a FET as high side
switch by survising the drain-source current via a shunt resistor and
additionally doing something like a power calculation for the FET by
multiplying the current by the drain-source voltage drop. For measuring
this voltage drop it uses an additional pin OUT which has to be
connected to source. So far for the background.
The pin in question (OUT) since being connected directly to "the world
outside" will have to withstand all the nasty things like surge/burst
pulses, reversed polarity caused by users etc. As a little help, you
might want to add maybe a suppressor diode together with a small serial
resistance to limit currents. Not to forget the must-have schottky diode
to ground against erroneous negative voltages applied to the drivers
outside.
Still, since you cannot spend too much voltage drop over that serial
resistor, the resulting current through the schottky diode will grow to
something like 1.3 ampere when applying negative voltages up to the
suppressor diodes clamping voltage (let's say 40 volt), limited by a
serial resistance of 30 ohm.
The question now is, which schottky diode to choose:
For example Vishay's SS2H10 that is able to stand a reverse voltage of
up to 100V would have a forward voltage drop of approx. 0.68 Volt. The
OUT pin would be driven far out of spec.
The opposite direction would be something like Vishay's SL44 whith a
impressing forward voltage of only approx. 0.31 Volt at 1.3 ampere. The
problem here is the low reverse voltage of only 28 Volt.

Did anybody solve this quest in the past? Any alternative protective
circuit ideas?

Cheerio,
Dirk Leber
HEITEC AG