Digital Output Protection

[Klaus Kragelund]

Hi

We have some 15V digital outputs that are pretty fast. They are
connected to other modules by means of a cable and the worry is that
somehow a cable can be misplaced, so that we could feed 30VDC into the
15V output.

Right now we use a resistor, but due to the low output impedance
requirements, the power dissipation when shorting to 30VDC is very
large (4W).

So, do any of you guys know of either an IC or a simple circuit
solution to feed through the signal with low impedance and also
protect the output from shorts to ground and 30VDC?

(I thought about something crazy like a resettable fuse, or using
protection diodes on the output to the supply rails and a precision
clamp on the rails to quinch the rising rail voltage in case of an
errer)

Thanks

Klaus

 

[James Arthur]

Hi

We have some 15V digital outputs that are pretty fast. They are
connected to other modules by means of a cable and the worry is that
somehow a cable can be misplaced, so that we could feed 30VDC into the
15V output.

Right now we use a resistor, but due to the low output impedance
requirements, the power dissipation when shorting to 30VDC is very
large (4W).

So, do any of you guys know of either an IC or a simple circuit
solution to feed through the signal with low impedance and also
protect the output from shorts to ground and 30VDC?

(I thought about something crazy like a resettable fuse, or using
protection diodes on the output to the supply rails and a precision
clamp on the rails to quinch the rising rail voltage in case of an
errer)

Thanks

Klaus

a) Is the output push-pull, or open collector?
b) What, and how much do you have to drive?

James Arthur

 

[Klaus Kragelund]

a) Is the output push-pull, or open collector?
b) What, and how much do you have to drive?

James Arthur- Skjul tekst i anførselstegn -

- Vis tekst i anførselstegn -

The output is push-pull because we need to re-charge EMC capacitors
quickly on the receiving end. The receiving end impedance is 5k
resistor with 1n in parallel to ground

Thanks

Klaus

 

[Klaus Kragelund]

Hi Klaus,  What are you using for the "15V digital outputs that are
pretty fast"?

The TC4468 CMOS driver (http://www.microchip.com/wwwproducts/
Devices.aspx?dDocName=en010679)

A diode in the 15V VDD line would keep the 30V from back-feeding into
the +15V.

Yes, but it would toast the stuff on the 15V VDD line. I was thinking
about using a crowbar on the VDD line triggered when the power supply
for the VDD line went out of regulation (due to a surge on the VDD
line).

Maybe a current limiter in the ground of the "15V digital outputs that
are pretty fast"?

Do you mean seperate current sensing in the ground line reference
(won't work if multible grounds are used due to several output with
numerous grounds)

Regards

Klaus

 

[Joerg]

Hi

We have some 15V digital outputs that are pretty fast. They are
connected to other modules by means of a cable and the worry is that
somehow a cable can be misplaced, so that we could feed 30VDC into the
15V output.

Right now we use a resistor, but due to the low output impedance
requirements, the power dissipation when shorting to 30VDC is very
large (4W).

So, do any of you guys know of either an IC or a simple circuit
solution to feed through the signal with low impedance and also
protect the output from shorts to ground and 30VDC?

(I thought about something crazy like a resettable fuse, or using
protection diodes on the output to the supply rails and a precision
clamp on the rails to quinch the rising rail voltage in case of an
errer)

Probably something can be concocted around larger depletion mode FETs so
the path starves when the outside wants to drive it past the rail. This
doesn't take care of a short to GND though, that would have to be
sensed, possibly meaning too many parts.

If your outputs are CMOS with body diodes in there, how about PTC
resistors? Just make sure there is a shunt regulator on the 15V rail
which kicks in when the outside wants to drive it past, say, 16V.

 

[Bob Engelhardt]

Hi

We have some 15V digital outputs that are pretty fast. They are
connected to other modules by means of a cable and the worry is that
somehow a cable can be misplaced, so that we could feed 30VDC into the
15V output. ...

Use unique connectors on your cables, so they can't be switched.

 

[Klaus Kragelund]

Probably something can be concocted around larger depletion mode FETs so
the path starves when the outside wants to drive it past the rail. This
doesn't take care of a short to GND though, that would have to be
sensed, possibly meaning too many parts.

Nice. Will try that idea. Actually I have not be told you if we need
to survive a short to ground.

If your outputs are CMOS with body diodes in there, how about PTC
resistors? Just make sure there is a shunt regulator on the 15V rail
which kicks in when the outside wants to drive it past, say, 16V.

Yes, that was my first thought. Simple and provide dual protection of
the power supply also.

Another trick could be to use keep our low ohmic resistor in the
output along with the protection diode.

Then using a micro or whatever circuit to monitor the voltage on both
sides of the resistor. In case of overvoltage the supply clamping
diodes will do the trick and the protection circuit would kick in
before the supply would rise to a too high level or the resistor would
be destroyed. It would then also report if the output is malfunctioned
in some way. Of course this add some failure modes and needs to be
considered.

Regards

Klaus

 

[Klaus Kragelund]

Use unique connectors on your cables, so they can't be switched.

Yes, thats one of the things we have looked at, but there is so many
connectors that it would be pricey to have different ones. Did look
into mating pins also, but thats also troublesome with price and
production

Thanks

Klaus

 

[Klaus Kragelund]

I meant the diode between the driver and the +15V rail.

Maybe a disconnect, such as a variation on...

http://analog-innovations.com/SED/OverAndReverseVoltageProtection.pdf









There are CMOS bus drivers which go tri-state when VDD is dropped OR
if the output is forced above rail.  I don't have a part number off
the top of my head, but I've designed them for Moto/ON-Semi around 10
years ago.

Will try to see if I can find one, thanks for the hint

Regards

Klaus

 

[Klaus Kragelund]

Might this help?

http://www.analog.com/en/switchesmultiplexers/analog-switches/adg465/...

I was looking for a device that would protect an output, this on
protects an input. But that will certainly help me, that perhaps
analog devices has a part that can do the trick.

I found this one from NXP:

http://www.standardics.nxp.com/products/lvc/datasheet/74lvc1g34.pdf

But, its a 3.3V part :-(

Regards

Klaus

 

[westom]

I was looking for a device that would protect an output, this on
protects an input. But that will certainly help me, that perhaps
analog devices has a part that can do the trick.

First you must do the same overvoltage analysis for both common mode
and differential mode. Your definition implies low capacitance
devices are required. Two possible sources for solutions to both
differential mode and common mode overvoltages may be low capacitance
devices from Sidactor and low capacitance avalanche diodes from
http://www.semtech.com

 

[Joerg]

Nice. Will try that idea. Actually I have not be told you if we need
to survive a short to ground.


Yes, that was my first thought. Simple and provide dual protection of
the power supply also.

Another trick could be to use keep our low ohmic resistor in the
output along with the protection diode.

However, that would dissipate more because you must select a low ohms
resistor so it'll keep burning those 4W. A PTC would rise in resistance
when it becomes hot and the total dissipated power would be much less.
It would just sit there at an elevated temperature. In medical we often
aren't allowed to do that because a local hotspot is not allowed.

Then using a micro or whatever circuit to monitor the voltage on both
sides of the resistor. In case of overvoltage the supply clamping
diodes will do the trick and the protection circuit would kick in
before the supply would rise to a too high level or the resistor would
be destroyed. It would then also report if the output is malfunctioned
in some way. Of course this add some failure modes and needs to be
considered.

As long as you can guarantee that the uC will not hang. Because a
smelly/smokey exit is frowned upon by customers and it would trigger a
service case.

 

[Ben Bradley]

Yes, thats one of the things we have looked at, but there is so many
connectors that it would be pricey to have different ones. Did look
into mating pins also, but thats also troublesome with price and
production

Failing that, color-code the cables and connectors.

 

[James Arthur]

The output is push-pull because we need to re-charge EMC capacitors
quickly on the receiving end. The receiving end impedance is 5k
resistor with 1n in parallel to ground

Thanks

Klaus

4W @ 30V says your series resistor is already 220 ohms, so
why not John's idea (Analog Devices part)?--it's only
85 ohms.

85 ohms * 1nF = 85 nS.

Is that fast enough?


Cheers,
James Arthur

 

[Rich Grise]

We have some 15V digital outputs that are pretty fast. They are
connected to other modules by means of a cable and the worry is that
somehow a cable can be misplaced, so that we could feed 30VDC into the
15V output.

Right now we use a resistor, but due to the low output impedance
requirements, the power dissipation when shorting to 30VDC is very
large (4W).

So, do any of you guys know of either an IC or a simple circuit
solution to feed through the signal with low impedance and also
protect the output from shorts to ground and 30VDC?

(I thought about something crazy like a resettable fuse, or using
protection diodes on the output to the supply rails and a precision
clamp on the rails to quinch the rising rail voltage in case of an
errer)

http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=P6KE16CA/54GI-ND

Have Fun!
Rich

 

[Rich Grise]

We have some 15V digital outputs that are pretty fast. They are
connected to other modules by means of a cable and the worry is that
somehow a cable can be misplaced, so that we could feed 30VDC into the
15V output.

Right now we use a resistor, but due to the low output impedance
requirements, the power dissipation when shorting to 30VDC is very
large (4W).

So, do any of you guys know of either an IC or a simple circuit
solution to feed through the signal with low impedance and also
protect the output from shorts to ground and 30VDC?

(I thought about something crazy like a resettable fuse, or using
protection diodes on the output to the supply rails and a precision
clamp on the rails to quinch the rising rail voltage in case of an
errer)

What you really need to do is use a different connector so that it's
impossible to plug +30V into an output.

Good Luck!
Rich

 

[Joerg]

http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=P6KE16CA/54GI-ND

That's a transzorb. 30V -> poof ... phssss ... *BANG*

 

[Joerg]

Joerg's favorite phrase ;-)

Wait until next week. A prototype with a >50kV gen is on the way. So
when I don't post anymore after that you know what happened ... tsk ...
*WHADDABAM*

 

[Jasen Betts]

Hi

We have some 15V digital outputs that are pretty fast. They are
connected to other modules by means of a cable and the worry is that
somehow a cable can be misplaced, so that we could feed 30VDC into the
15V output.

Right now we use a resistor, but due to the low output impedance
requirements, the power dissipation when shorting to 30VDC is very
large (4W).

could you use a larger resistor and a parallel capacitor?

So, do any of you guys know of either an IC or a simple circuit
solution to feed through the signal with low impedance and also
protect the output from shorts to ground and 30VDC?

(I thought about something crazy like a resettable fuse, or using
protection diodes on the output to the supply rails and a precision
clamp on the rails to quinch the rising rail voltage in case of an
errer)

sounds good, how about a polyswitch and the diodes,