Surge Pulse Clamping with Ceramic Capacitors

[Klaus Kragelund]

Hi

For an application I need to protect the electronics of an RS485
interface from surges (8/20us 1kV pulse)

The normal way be to use a transzorb, in a SMA housing or even a SOT23
device.

That is not possible since we have to be able to withstand up to 30V
DC on the bus also (that is handled by a special RS485 IC)

The problem is that the selected breakdown voltage of the transzorb
therefore is high (>30V) and a lot of energy is dissipated into the
transzorb

In another product I have used a diode from the affected node to a
ceramic capacitor with a bleeder resistor in parallel to clamp the
energy and dissipate the energy into the bleeder and that worked fine.
The diode sees very little energy and the capacitor is just charged
during the pulse

I never saw any problems doing that, but I would like to know if
anyone here has tried the same and has any inputs into failure cases
or even a better way to clamp the pulse?

One "feature" of the diode-capacitor clamp is that closely spaced
pulses will eventually destroy the capacitor, but anyhow closely
spaced pulses in a tranzorb will also destroy that one....

Thanks

Klaus

 

[Joerg]

Hi

For an application I need to protect the electronics of an RS485
interface from surges (8/20us 1kV pulse)

What, you also? I am just looking at RS485 protection for an industrial
system.

The normal way be to use a transzorb, in a SMA housing or even a SOT23
device.

Normally you need special ones with low capacitance or TVS in series
with regular diodes. Ordinary transzorbs with their huge capacitance can
muffle the data signal because of the lowpass effect. They'd "kill it
dead", as John Wayne would have put it.

That is not possible since we have to be able to withstand up to 30V
DC on the bus also (that is handled by a special RS485 IC)

The problem is that the selected breakdown voltage of the transzorb
therefore is high (>30V) and a lot of energy is dissipated into the
transzorb

In another product I have used a diode from the affected node to a
ceramic capacitor with a bleeder resistor in parallel to clamp the
energy and dissipate the energy into the bleeder and that worked fine.
The diode sees very little energy and the capacitor is just charged
during the pulse

I never saw any problems doing that, but I would like to know if
anyone here has tried the same and has any inputs into failure cases
or even a better way to clamp the pulse?

It works but can blunt the edges of your signal in times when the cap is
not yet charged at all. You can also limit the charge on the cap via a
big TVS in parallel to it.

One "feature" of the diode-capacitor clamp is that closely spaced
pulses will eventually destroy the capacitor, but anyhow closely
spaced pulses in a tranzorb will also destroy that one....

Why not clamp directly to the 30V DC rail? You just have to make sure
that this rail won't start to float up when lots of pulses appear and
the load on the DC rail is too light. A TL431 can help with that
although it'll be marginal in this case with its 36V limit. Maybe
another homemade shunt regulator.

 

[Klaus Kragelund]

What, you also? I am just looking at RS485 protection for an industrial
system.


Normally you need special ones with low capacitance or TVS in series
with regular diodes. Ordinary transzorbs with their huge capacitance can
muffle the data signal because of the lowpass effect. They'd "kill it
dead", as John Wayne would have put it.

Yes, we are using a low capacitance type (in other case put a diode in
series to quinch the tranzorb capacitance)

It works but can blunt the edges of your signal in times when the cap is
not yet charged at all. You can also limit the charge on the cap via a
big TVS in parallel to it.

I have no room for it, don't even have room for 3 SMAs

Why not clamp directly to the 30V DC rail? You just have to make sure
that this rail won't start to float up when lots of pulses appear and
the load on the DC rail is too light. A TL431 can help with that
although it'll be marginal in this case with its 36V limit. Maybe
another homemade shunt regulator.

The 30V is only available since some idiot would connect 30V power to
the A/B line of the RS485 interface

One way to do it to add a set of biasing resistors, so one cap is
charged to +12V and another to -7V (the RS485 CM range) and connect
that surge capacitor to the bus via the diode.

Another idea is to use the Bourns CDSOT23-SM712, specifically designet
for surge protection for RS485 devices, but add ceramic caps in series
with each line connection so it can tolerate 30V without creating
wonderful smoke, but will be able to clamp surge pulses without
affecting the high speed bus.

Regards

Klaus

 

[Klaus Kragelund]

Yes, we are using a low capacitance type (in other case put a diode in
series to quinch the tranzorb capacitance)













I have no room for it, don't even have room for 3 SMAs



The 30V is only available since some idiot would connect 30V power to
the A/B line of the RS485 interface

One way to do it to add a set of biasing resistors, so one cap is
charged to +12V and another to -7V (the RS485 CM range) and connect
that surge capacitor to the bus via the diode.

Another idea is to use the Bourns CDSOT23-SM712, specifically designet
for surge protection for RS485 devices, but add ceramic caps in series
with each line connection so it can tolerate 30V without creating
wonderful smoke, but will be able to clamp surge pulses without
affecting the high speed bus.

Regards

Klaus

Like this:

www.electronicsdesign.dk/tmp/RS485_cap_protection.pdf

 

[Joerg]

Klaus Kragelund wrote:

[...]

It works but can blunt the edges of your signal in times when the cap is
not yet charged at all. You can also limit the charge on the cap via a
big TVS in parallel to it.

I have no room for it, don't even have room for 3 SMAs

Almost the same here

The 30V is only available since some idiot would connect 30V power to
the A/B line of the RS485 interface

One way to handle that would be to sense A and B. The instance either
one goes beyond 12V cut the data line electronically or shunt to ground
with Polyfuses in line. I don't like the Polyfuse approach because they
become hot, if possible I'd try to electronically switch open. Two
back-to-back P-channels or something like that, maybe use an array if
space is really tight. That avoids dissipation issues.

One way to do it to add a set of biasing resistors, so one cap is
charged to +12V and another to -7V (the RS485 CM range) and connect
that surge capacitor to the bus via the diode.

That would be a very nice way to do it. In my case I don't have any
negative supply, and also very little space. But you'd have to make sure
it can't run away. So either the dividers have to contain low enough
resistor values or there have to be clamps. With the chance of some dude
connecting a hard 30VDC that's next to impossible. Better to use an
electronic protection like above.

Another idea is to use the Bourns CDSOT23-SM712, specifically designet
for surge protection for RS485 devices, but add ceramic caps in series
with each line connection so it can tolerate 30V without creating
wonderful smoke, but will be able to clamp surge pulses without
affecting the high speed bus.

Sure, but then you'd have DC drift on the bus. Might be ok but it can
get iffy.

 

[Robert Macy]

Hi

For an application I need to protect the electronics of an RS485
interface from surges (8/20us 1kV pulse)

The normal way be to use a transzorb, in a SMA housing or even a SOT23
device.

That is not possible since we have to be able to withstand up to 30V
DC on the bus also (that is handled by a special RS485 IC)

The problem is that the selected breakdown voltage of the transzorb
therefore is high (>30V) and a lot of energy is dissipated into the
transzorb

In another product I have used a diode from the affected node to a
ceramic capacitor with a bleeder resistor in parallel to clamp the
energy and dissipate the energy into the bleeder and that worked fine.
The diode sees very little energy and the capacitor is just charged
during the pulse

I never saw any problems doing that, but I would like to know if
anyone here has tried the same and has any inputs into failure cases
or even a better way to clamp the pulse?

One "feature" of the diode-capacitor clamp is that closely spaced
pulses will eventually destroy the capacitor, but anyhow closely
spaced pulses in a tranzorb will also destroy that one....

Thanks

Klaus

what's the impedance of the pulse? if low, wont work.

or, rather how many joules do you expect in the pulse?

I once had to protect telco equipment whose pulse went to around 500V,
BUT could supply around 500A With protection like you describe you
didn't lose a tranzorb, you'd lose the PCB!

 

[Joerg]

[...]

Another idea is to use the Bourns CDSOT23-SM712, specifically designet
for surge protection for RS485 devices, but add ceramic caps in series
with each line connection so it can tolerate 30V without creating
wonderful smoke, but will be able to clamp surge pulses without
affecting the high speed bus.

Regards

Klaus

Like this:

www.electronicsdesign.dk/tmp/RS485_cap_protection.pdf

But that would not protect against a hard 30VDC applied because an
installer miswired something. It could cause your RS485 chip to go PHUT
unless it has internal protection against this.

 

[Tim Williams]

Can give latches a try. There are SIDACs made for protection duty.
That'll short the line down to a few volts within a microsecond or so (if
your bitrate isn't too high, you could filter out the remaining fractional
microseconds of overshoot). Should still work if the line remains
active -- RS485 levels of 200mV won't keep a SIDAC on, so you could use
two (one per line to ground) or three (full delta).

MOVs handle gobs of energy, but of course, they are rather high
capacitance, no good for a terminated line.

There are low-capacitance TVSs, made from some sort of plastic I think,
that'd do this sort of job, but I don't know that any will handle the kind
of energy a 1kV spike will most likely deliver. Here's an example:
http://www.littelfuse.com/products/...ata Sheets/Littelfuse_PulseGuard ESD_PGB2.pdf
Yeah, no mention of avalanche energy... hmm...

Other than that, diodes into a rail or conventional TVS -- as suggested --
is your best alternative.

Tim

 

[Klaus Kragelund]

Klaus Kragelund wrote:

[...]

[Snip]

One way to do it to add a set of biasing resistors, so one cap is
charged to +12V and another to -7V (the RS485 CM range) and connect
that surge capacitor to the bus via the diode.

That would be a very nice way to do it. In my case I don't have any
negative supply, and also very little space. But you'd have to make sure
it can't run away. So either the dividers have to contain low enough
resistor values or there have to be clamps. With the chance of some dude
connecting a hard 30VDC that's next to impossible. Better to use an
electronic protection like above.

The electronic protection would probably be sensitive to ESD and
burst, and is a pain to design with a possible 30V voltage applied

Sure, but then you'd have DC drift on the bus. Might be ok but it can
get iffy.

The Bourns part has 7V/12V transzorbs back-to-back, so it will handle
the CM range of the bus. A resistor should be added in the schematics
in parallel with each capacitor

Cheers

Klaus

 

[Klaus Kragelund]

what's the impedance of the pulse? if low, wont work.

The impedance is 40ohms, so the peak current as test reveals is about
20A

or, rather how many joules do you expect in the pulse?

I once had to protect telco equipment whose pulse went to around 500V,
BUT could supply around 500A  With protection like you describe you
didn't lose a tranzorb, you'd lose the PCB!

I have a surge test earlier passed using the CDSOT device, without the
requirement for 30V applied

Regards

Klaus

 

[Klaus Kragelund]

[...]

Another idea is to use the Bourns CDSOT23-SM712, specifically designet
for surge protection for RS485 devices, but add ceramic caps in series
with each line connection so it can tolerate 30V without creating
wonderful smoke, but will be able to clamp surge pulses without
affecting the high speed bus.
Regards
Klaus

Like this:

www.electronicsdesign.dk/tmp/RS485_cap_protection.pdf

But that would not protect against a hard 30VDC applied because an
installer miswired something. It could cause your RS485 chip to go PHUT
unless it has internal protection against this.

The RS485 IC has +/-60V protection rating, so its ok

Regards

Klaus

 

[Klaus Kragelund]

Klaus Kragelund wrote:

[...]

[Snip]

One way to do it to add a set of biasing resistors, so one cap is
charged to +12V and another to -7V (the RS485 CM range) and connect
that surge capacitor to the bus via the diode.

That would be a very nice way to do it. In my case I don't have any
negative supply, and also very little space. But you'd have to make sure
it can't run away. So either the dividers have to contain low enough
resistor values or there have to be clamps. With the chance of some dude
connecting a hard 30VDC that's next to impossible. Better to use an
electronic protection like above.

The electronic protection would probably be sensitive to ESD and
burst, and is a pain to design with a possible 30V voltage applied

Sure, but then you'd have DC drift on the bus. Might be ok but it can
get iffy.

The Bourns part has 7V/12V transzorbs back-to-back, so it will handle
the CM range of the bus. A resistor should be added in the schematics
in parallel with each capacitor

.... resistor added to avoid stair casing of the voltage on the
capacitor from multiple surges (discharge time of less than a second,
the standard defines wait times between pulses of maximum 60 seconds,
which of course does not mimic real life pulses)

 

[Joerg]

The RS485 IC has +/-60V protection rating, so its ok

Which one do you use? The ones on mine (clients's choice) have abs max
ratings of -8V to +12V, and no internal circuitry given.

Also, make sure for how long it can take that. Sometimes there is only a
poly-resistor that eventually hisses out.

 

[Klaus Kragelund]

Can give latches a try.  There are SIDACs made for protection duty.
That'll short the line down to a few volts within a microsecond or so (if
your bitrate isn't too high, you could filter out the remaining fractional
microseconds of overshoot).  Should still work if the line remains
active -- RS485 levels of 200mV won't keep a SIDAC on, so you could use
two (one per line to ground) or three (full delta).

Looks like a nice part, had not heard of them before. Seems it will be
a little too big, AFAICS the same size as SMA tranzorbs.

MOVs handle gobs of energy, but of course, they are rather high
capacitance, no good for a terminated line.

MOVs derate over time, every pulse applied reduces the breakdown
voltage, so after some use the product self-destructs

A ceramic TVS, like the CT0805K14 is better, so they say, but it has
very soft knee and large capacitance

Regards

Klaus

 

[Joerg]

Klaus Kragelund wrote: [...]

[Snip]

One way to do it to add a set of biasing resistors, so one cap is
charged to +12V and another to -7V (the RS485 CM range) and connect
that surge capacitor to the bus via the diode.

That would be a very nice way to do it. In my case I don't have any
negative supply, and also very little space. But you'd have to make sure
it can't run away. So either the dividers have to contain low enough
resistor values or there have to be clamps. With the chance of some dude
connecting a hard 30VDC that's next to impossible. Better to use an
electronic protection like above.

The electronic protection would probably be sensitive to ESD and
burst, and is a pain to design with a possible 30V voltage applied

You'd need a 40V or so TVS upfront but I don't see the 30V presenting a
problem. I was thinking about something like this:

http://www.vishay.com/docs/71433/71433.pdf

The Bourns part has 7V/12V transzorbs back-to-back, so it will handle
the CM range of the bus. A resistor should be added in the schematics
in parallel with each capacitor

Yes, it'll handle it alright but the bus will meander around a bit
depending on the data transmitted. But how would it prevent 30VDC from
being sent down a data line?

 

[Klaus Kragelund]

Klaus Kragelund wrote:
[...]
Another idea is to use the Bourns CDSOT23-SM712, specifically designet
for surge protection for RS485 devices, but add ceramic caps in series
with each line connection so it can tolerate 30V without creating
wonderful smoke, but will be able to clamp surge pulses without
affecting the high speed bus.
Regards
Klaus
Like this:
www.electronicsdesign.dk/tmp/RS485_cap_protection.pdf
But that would not protect against a hard 30VDC applied because an
installer miswired something. It could cause your RS485 chip to go PHUT
unless it has internal protection against this.

The RS485 IC has +/-60V protection rating, so its ok

Which one do you use? The ones on mine (clients's choice) have abs max
ratings of -8V to +12V, and no internal circuitry given.

LTC2862, can take 60V indefinitely, but you need to add clamping
circuit to the VDD node since it will dump current into that node if
it is in transmit mode and is subjected to back fed voltage.

Cheers

Klaus

 

[Klaus Kragelund]

Klaus Kragelund wrote:
Klaus Kragelund wrote:
[...]

[Snip]

One way to do it to add a set of biasing resistors, so one cap is
charged to +12V and another to -7V (the RS485 CM range) and connect
that surge capacitor to the bus via the diode.
That would be a very nice way to do it. In my case I don't have any
negative supply, and also very little space. But you'd have to make sure
it can't run away. So either the dividers have to contain low enough
resistor values or there have to be clamps. With the chance of some dude
connecting a hard 30VDC that's next to impossible. Better to use an
electronic protection like above.

The electronic protection would probably be sensitive to ESD and
burst, and is a pain to design with a possible 30V voltage applied

You'd need a 40V or so TVS upfront but I don't see the 30V presenting a
problem. I was thinking about something like this:

http://www.vishay.com/docs/71433/71433.pdf

The Bourns part has 7V/12V transzorbs back-to-back, so it will handle
the CM range of the bus. A resistor should be added in the schematics
in parallel with each capacitor

Yes, it'll handle it alright but the bus will meander around a bit
depending on the data transmitted. But how would it prevent 30VDC from
being sent down a data line?

Our device will survive, but other devices on the same bus from other
suppliers will probably be destroyed.

A lot of the device I have reverse engineered has no protection at all
and will be destroyed from this

Cheers

Klaus

 

[Klaus Kragelund]

Klaus Kragelund wrote:
Klaus Kragelund wrote:
[...]

[Snip]

One way to do it to add a set of biasing resistors, so one cap is
charged to +12V and another to -7V (the RS485 CM range) and connect
that surge capacitor to the bus via the diode.
That would be a very nice way to do it. In my case I don't have any
negative supply, and also very little space. But you'd have to make sure
it can't run away. So either the dividers have to contain low enough
resistor values or there have to be clamps. With the chance of some dude
connecting a hard 30VDC that's next to impossible. Better to use an
electronic protection like above.

The electronic protection would probably be sensitive to ESD and
burst, and is a pain to design with a possible 30V voltage applied

You'd need a 40V or so TVS upfront but I don't see the 30V presenting a
problem. I was thinking about something like this:

http://www.vishay.com/docs/71433/71433.pdf

It would need levelshifting and another supply rail to control the
FETs at all input voltages. be immune to bursts and allow to be turned
on from regular signals from the RS485 device

I tried to draw it up, became a nightmare

Cheers

Klaus

 

[Joerg]

Klaus Kragelund wrote:
[...]
Another idea is to use the Bourns CDSOT23-SM712, specifically designet
for surge protection for RS485 devices, but add ceramic caps in series
with each line connection so it can tolerate 30V without creating
wonderful smoke, but will be able to clamp surge pulses without
affecting the high speed bus.
Regards
Klaus
Like this:
www.electronicsdesign.dk/tmp/RS485_cap_protection.pdf
But that would not protect against a hard 30VDC applied because an
installer miswired something. It could cause your RS485 chip to go PHUT
unless it has internal protection against this.
The RS485 IC has +/-60V protection rating, so its ok

Which one do you use? The ones on mine (clients's choice) have abs max
ratings of -8V to +12V, and no internal circuitry given.

LTC2862, can take 60V indefinitely, but you need to add clamping
circuit to the VDD node since it will dump current into that node if
it is in transmit mode and is subjected to back fed voltage.

Wow, that sure is the Rolls-Royce of RS485 chips. With a corresponding
price tag

But isn't that bleed-through only an issue if the ground of the LTC2862
has come off? Otherwise this would really be a problem because you'd
just have moved the dissipation from one place to another. I think the
only real protection for 30V continuously is some sort of cut-off,
whether inside or outside a chip. At least a partial one where the
current becomes very small.

 

[Joerg]

Klaus Kragelund wrote:
Klaus Kragelund wrote:
[...]
[Snip]
One way to do it to add a set of biasing resistors, so one cap is
charged to +12V and another to -7V (the RS485 CM range) and connect
that surge capacitor to the bus via the diode.
That would be a very nice way to do it. In my case I don't have any
negative supply, and also very little space. But you'd have to make sure
it can't run away. So either the dividers have to contain low enough
resistor values or there have to be clamps. With the chance of some dude
connecting a hard 30VDC that's next to impossible. Better to use an
electronic protection like above.
The electronic protection would probably be sensitive to ESD and
burst, and is a pain to design with a possible 30V voltage applied

You'd need a 40V or so TVS upfront but I don't see the 30V presenting a
problem. I was thinking about something like this:

http://www.vishay.com/docs/71433/71433.pdf

It would need levelshifting and another supply rail to control the
FETs at all input voltages. be immune to bursts and allow to be turned
on from regular signals from the RS485 device

I tried to draw it up, became a nightmare

Yeah, you do need a negative supply voltage. Sources and gates tied
together, zener between them, resistor from gates to negative supply.
The OVP trigger could then shunt gate-source via an optocoupler, or you
could have a transistor at the base (in that case there needs to be also
a gate-source bleed resistor).

There may even be a ready-to-go OVP protector chip with all this in
there, but probably expensive.

Then there are always the Supertex high voltage mux chips but they will
probably be too big. I wish they made something like this for RS485
level trip points but haven't seen it yet:

http://www.supertex.com/pdf/datasheets/MD0100.pdf