TVS breakdown voltages for higher currents

[legg]

That is exactly what I meant. It is a non-issue _if_ you know how much
it will conduct at 80V, worst case maximum. And I don't, hence my
inquiry and my post here.

If you don't want 80V conduction, use the next higher voltage in the
range. This is guaranteed not to conduct below 85V.

If the aim is to protect your hardware, design it to do so. A dead TVS
is worse than no TVS.

RL

 

[legg]

Zen wasn't possible, I had about the square-footage of a fly to clamp at
two places, no chance to have opening devices because they'd have to be
quite major.

Anyhow, there will never be much energy above 80V and all I want is
something that is guaranteed not to conduct for than a few tens of
milliamps at 80V. The normal voltage range is less than 32V and the 80V
peaks will be very short and a rare occurrence. I am fairly sure a 70V
TVS can do that but a bench test isn't good enough here.

Characterize the ringing for worst case energy content, and stop using
vague generalizations.

Can you damp the 'low energy' ringing?

Real estate reduction possibility = two signal diodes and a single
clamp / energy storage rail that means business, or
capacitively-couple a thyristor-type structure with more benign
single-fault characteristics.

RL

 

[Joerg]

If you don't want 80V conduction, use the next higher voltage in the
range. This is guaranteed not to conduct below 85V.

As I mentioned, exactly that is what I can't do. A 85V TVS will not
conduct in any significant way until the voltage is way past 110V, and
tha's above abs max here.

If the aim is to protect your hardware, design it to do so. A dead TVS
is worse than no TVS.

Can't be done differently, no space for a cut-out circuit. A TVS does
not have to die here, provided its current at 80V is modest enough. The
spike bursts will be short and very low duty cycle, as in "once in a
blue moon". Energy above 80V is small, but not quite small enough for a
zener.

All I need is hard data, like "guaranteed not to exceed 50mA at 80V". I
am sure that data exists but it didn't make it into the datasheets. Same
with zeners but there I was able to obtain such data from a small mfg.

 

[Joerg]

Characterize the ringing for worst case energy content, and stop using
vague generalizations.

When clamped hard to 90V with it is about 10A into the clamp, for around
60usec.

Can you damp the 'low energy' ringing?

Unfortunately not in this case. It is due to stuff outside my circuitry.

Real estate reduction possibility = two signal diodes and a single
clamp / energy storage rail that means business, or
capacitively-couple a thyristor-type structure with more benign
single-fault characteristics.

The clamp device is the problem, it'll usually require a D2PAK or
something that is similarly beefy.

 

[Joerg]

Gas discharge?
<http://www.mouser.com/ProductDetail...=sGAEpiMZZMsdWQmBKEIh8nxzCboN3V1ERArz/wH/83A=>

Thanks, but those don't work because of the "90V +/-30%". At -30% that
would result in kablouie and at +30% the electronics behind it would be
dead.

 

[Joerg]

You could try a gas discharge tube as a first line of defence, there are
some tiny surface mount ones around now which trip around 70V, and then
clamp around 20V. Can take amps for much longer than a TVS.

That's what MArtin suggested but they huge tolerance makes those not
useful. I only have the range of 80V to 100V to work in.

 

[[email protected]]

[email protected] wrote:

Hello Folks, Does anyone know a manufacturer that provides
a wee bit more info in datasheets for their TVS? The issue:
We have SMAJ78CA in a board because long cables, filters
and such can cause nasty ringing. The board must tolerate
several 100msec bursts of 80V amplitude in a row, but only
on rare occasions and then such a burst is a one-time
event. Long story short, these things won't clamp well
before 100V is reached in the ringing and then some stuff
with 100V abs max ratings can go phhhhut.
Many of these flea-sized TVS have really weak average power
handling capability, usually just 10s mW, so those 100msec
burst times of yours have a good chance of blowing the
SMAJ78CA. You better start with the repetitive rate power
handling of the part before you waste a bunch of time looking
for V/I limit guarantees.
SMA isn't exactly flea-sized. They can take a watt or two for a
few hundred msec. That's all I am asking them to do. [...]
No they can't. The Vishay ds says repetitive power handling of
300W at 0.01% duty which comes to 30mW.




Of course not continuously. That's why I wrote "for a few hundred
msec".

As mentioned before, the duty cycle will be extremely low, and
it'll be

less than 0.01%.



I am not going to sink 300W pulse into them but maybe 1-2W pulses.



--

Regards, Joerg



http://www.analogconsultants.com/

Okay , from Fig 5.

http://www.mouser.com/catalog/specsheets/smaj50.pdf the transient

thermal impedance is less than 10oC/W (~7) for the 100ms-1000ms

pulses, so it's a non-issue. But anything approaching continuous and

Rja goes towards 120oC/W in that flea package.

That is exactly what I meant. It is a non-issue _if_ you know how much

it will conduct at 80V, worst case maximum. And I don't, hence my

inquiry and my post here.



--

Regards, Joerg



http://www.analogconsultants.com/

You could combine the precision of a zener with the peak energy handling ofthe TVS to get your results- the zener sets the peak firing voltage, TVS is low voltage not critical except that its threshold for significant conduction exceeds steady state worst case maximum DC on line with sufficient margin so that SCR shuts off by holding current deprivation.


Please view in a fixed-width font such as
Courier.

..
..
..
..
..
.. ----------+-----------+------------
.. | |
.. | |
.. | |
.. ___/ ___/
.. // \ // \
.. --- ZD --- TVS
.. | |
.. | |
.. | |
.. | -----
.. +---------/ \ SCR
.. | -----
.. | |
.. [390] |
.. | |
.. | |
.. | |
.. ----------+-----------+--------------
..
..
..
..
..
..
..
..
..
..
..
..
..
..

 

[Joerg]

Hello Folks, Does anyone know a manufacturer that provides
a wee bit more info in datasheets for their TVS? The issue:
We have SMAJ78CA in a board because long cables, filters
and such can cause nasty ringing. The board must tolerate
several 100msec bursts of 80V amplitude in a row, but only
on rare occasions and then such a burst is a one-time
event. Long story short, these things won't clamp well
before 100V is reached in the ringing and then some stuff
with 100V abs max ratings can go phhhhut.
Many of these flea-sized TVS have really weak average power
handling capability, usually just 10s mW, so those 100msec
burst times of yours have a good chance of blowing the
SMAJ78CA. You better start with the repetitive rate power
handling of the part before you waste a bunch of time looking
for V/I limit guarantees.
SMA isn't exactly flea-sized. They can take a watt or two for a
few hundred msec. That's all I am asking them to do. [...]
No they can't. The Vishay ds says repetitive power handling of
300W at 0.01% duty which comes to 30mW.
Of course not continuously. That's why I wrote "for a few hundred
msec".
As mentioned before, the duty cycle will be extremely low, and
it'll be
less than 0.01%.
I am not going to sink 300W pulse into them but maybe 1-2W pulses.
--
Regards, Joerg
http://www.analogconsultants.com/
Okay , from Fig 5.
http://www.mouser.com/catalog/specsheets/smaj50.pdf the transient
thermal impedance is less than 10oC/W (~7) for the 100ms-1000ms
pulses, so it's a non-issue. But anything approaching continuous and
Rja goes towards 120oC/W in that flea package.

That is exactly what I meant. It is a non-issue _if_ you know how much

it will conduct at 80V, worst case maximum. And I don't, hence my

inquiry and my post here.



--

Regards, Joerg



http://www.analogconsultants.com/

You could combine the precision of a zener with the peak energy handling of the TVS to get your results- the zener sets the peak firing voltage, TVS is low voltage not critical except that its threshold for significant conduction exceeds steady state worst case maximum DC on line with sufficient margin so that SCR shuts off by holding current deprivation.


Please view in a fixed-width font such as
Courier.

.
.
.
.
.
. ----------+-----------+------------
. | |
. | |
. | |
. ___/ ___/
. // \ // \
. --- ZD --- TVS
. | |
. | |
. | |
. | -----
. +---------/ \ SCR
. | -----
. | |
. [390] |
. | |
. | |
. | |
. ----------+-----------+--------------
.
.

That is a pretty good idea. Well, if I had the space for a big enough
SCR. But there is another problem: The SCR will not let go until the TVS
has largely come out of conduction and the decay of the 80V surges is
going to be very slow. That would cook out the TVS.

So essentially I have to leave everything under 80V untouched or sink
only a very small current whenever the circuit is below 80V. Just to
avoid overheating any of the protective devices.

The main issue is the fast onset of the 80V surge which creates ringing.
It's only this ringing I have to cap. So potentially I cold do something
with a high-pass but ... no space.

 

[Jamie]

[email protected] wrote:


On Saturday, September 22, 2012 9:06:34 PM UTC-4, Joerg wrote:

[email protected] wrote:

On Saturday, September 22, 2012 7:59:55 PM UTC-4, Joerg wrote:

[email protected] wrote:

Hello Folks, Does anyone know a manufacturer that provides
a wee bit more info in datasheets for their TVS? The issue:
We have SMAJ78CA in a board because long cables, filters
and such can cause nasty ringing. The board must tolerate
several 100msec bursts of 80V amplitude in a row, but only
on rare occasions and then such a burst is a one-time
event. Long story short, these things won't clamp well
before 100V is reached in the ringing and then some stuff
with 100V abs max ratings can go phhhhut.

Many of these flea-sized TVS have really weak average power
handling capability, usually just 10s mW, so those 100msec
burst times of yours have a good chance of blowing the
SMAJ78CA. You better start with the repetitive rate power
handling of the part before you waste a bunch of time looking
for V/I limit guarantees.

SMA isn't exactly flea-sized. They can take a watt or two for a
few hundred msec. That's all I am asking them to do. [...]

No they can't. The Vishay ds says repetitive power handling of
300W at 0.01% duty which comes to 30mW.

Of course not continuously. That's why I wrote "for a few hundred
msec".
As mentioned before, the duty cycle will be extremely low, and
it'll be
less than 0.01%.
I am not going to sink 300W pulse into them but maybe 1-2W pulses.
--
Regards, Joerg
http://www.analogconsultants.com/

Okay , from Fig 5.
http://www.mouser.com/catalog/specsheets/smaj50.pdf the transient
thermal impedance is less than 10oC/W (~7) for the 100ms-1000ms
pulses, so it's a non-issue. But anything approaching continuous and
Rja goes towards 120oC/W in that flea package.




That is exactly what I meant. It is a non-issue _if_ you know how much

it will conduct at 80V, worst case maximum. And I don't, hence my

inquiry and my post here.



--

Regards, Joerg



http://www.analogconsultants.com/

You could combine the precision of a zener with the peak energy handling of the TVS to get your results- the zener sets the peak firing voltage, TVS is low voltage not critical except that its threshold for significant conduction exceeds steady state worst case maximum DC on line with sufficient margin so that SCR shuts off by holding current deprivation.


Please view in a fixed-width font such as
Courier.

.
.
.
.
.
. ----------+-----------+------------
. | |
. | |
. | |
. ___/ ___/
. // \ // \
. --- ZD --- TVS
. | |
. | |
. | |
. | -----
. +---------/ \ SCR
. | -----
. | |
. [390] |
. | |
. | |
. | |
. ----------+-----------+--------------
.
.

That is a pretty good idea. Well, if I had the space for a big enough
SCR. But there is another problem: The SCR will not let go until the TVS
has largely come out of conduction and the decay of the 80V surges is
going to be very slow. That would cook out the TVS.

So essentially I have to leave everything under 80V untouched or sink
only a very small current whenever the circuit is below 80V. Just to
avoid overheating any of the protective devices.

The main issue is the fast onset of the 80V surge which creates ringing.
It's only this ringing I have to cap. So potentially I cold do something
with a high-pass but ... no space.

80 volt rail or what ever.

+
|
|
+-+---+
z |
A |
| |
| |
.-. |
| | |
| | |
'-' +
| |/
+-+-| some higher voltage type smt
.-. |>
| | +
| | |
'-' |
+-----+

low side or back feed to the supply rail
(created by AACircuit v1.28.6 beta 04/19/05 www.tech-chat.de)

All that can be fitted in a small area..

I've used such circuit and it wells very well.
Jamie

 

[[email protected]]

[email protected] wrote:

Hello Folks, Does anyone know a manufacturer that provides
a wee bit more info in datasheets for their TVS? The issue:
We have SMAJ78CA in a board because long cables, filters
and such can cause nasty ringing. The board must tolerate
several 100msec bursts of 80V amplitude in a row, but only
on rare occasions and then such a burst is a one-time
event. Long story short, these things won't clamp well
before 100V is reached in the ringing and then some stuff
with 100V abs max ratings can go phhhhut.
Many of these flea-sized TVS have really weak average power
handling capability, usually just 10s mW, so those 100msec
burst times of yours have a good chance of blowing the
SMAJ78CA. You better start with the repetitive rate power
handling of the part before you waste a bunch of time looking
for V/I limit guarantees.
SMA isn't exactly flea-sized. They can take a watt or two for a
few hundred msec. That's all I am asking them to do. [...]
No they can't. The Vishay ds says repetitive power handling of
300W at 0.01% duty which comes to 30mW.
Of course not continuously. That's why I wrote "for a few hundred
msec".
As mentioned before, the duty cycle will be extremely low, and
it'll be
less than 0.01%.
I am not going to sink 300W pulse into them but maybe 1-2W pulses.
--
Regards, Joerg
http://www.analogconsultants.com/
Okay , from Fig 5.
http://www.mouser.com/catalog/specsheets/smaj50.pdf the transient
thermal impedance is less than 10oC/W (~7) for the 100ms-1000ms
pulses, so it's a non-issue. But anything approaching continuous and
Rja goes towards 120oC/W in that flea package.




That is exactly what I meant. It is a non-issue _if_ you know how much

it will conduct at 80V, worst case maximum. And I don't, hence my

inquiry and my post here.



--

Regards, Joerg



http://www.analogconsultants.com/

You could combine the precision of a zener with the peak energy handling of the TVS to get your results- the zener sets the peak firing voltage, TVS is low voltage not critical except that its threshold for significant conduction exceeds steady state worst case maximum DC on line with sufficientmargin so that SCR shuts off by holding current deprivation.
Please view in a fixed-width font such as

Courier.

.

.

.

.

.

. ----------+-----------+------------

. | |

. | |

. | |

. ___/ ___/

. // \ // \

. --- ZD --- TVS

. | |

. | |

. | |

. | -----

. +---------/ \ SCR

. | -----

. | |

. [390] |

. | |

. | |

. | |

. ----------+-----------+--------------

.

That is a pretty good idea. Well, if I had the space for a big enough

SCR. But there is another problem: The SCR will not let go until the TVS

has largely come out of conduction and the decay of the 80V surges is

going to be very slow. That would cook out the TVS.



So essentially I have to leave everything under 80V untouched or sink

only a very small current whenever the circuit is below 80V. Just to

avoid overheating any of the protective devices.



The main issue is the fast onset of the 80V surge which creates ringing.

It's only this ringing I have to cap. So potentially I cold do something

with a high-pass but ... no space.



--

Regards, Joerg



http://www.analogconsultants.com/

The SCR can be a micro-dot, there's not enough voltage across it to cause any heating at your currents.

 

[[email protected]]

[email protected] wrote:

Hello Folks, Does anyone know a manufacturer that provides
a wee bit more info in datasheets for their TVS? The issue:
We have SMAJ78CA in a board because long cables, filters
and such can cause nasty ringing. The board must tolerate
several 100msec bursts of 80V amplitude in a row, but only
on rare occasions and then such a burst is a one-time
event. Long story short, these things won't clamp well
before 100V is reached in the ringing and then some stuff
with 100V abs max ratings can go phhhhut.
Many of these flea-sized TVS have really weak average power
handling capability, usually just 10s mW, so those 100msec
burst times of yours have a good chance of blowing the
SMAJ78CA. You better start with the repetitive rate power
handling of the part before you waste a bunch of time looking
for V/I limit guarantees.
SMA isn't exactly flea-sized. They can take a watt or two for a
few hundred msec. That's all I am asking them to do. [...]
No they can't. The Vishay ds says repetitive power handling of
300W at 0.01% duty which comes to 30mW.
Of course not continuously. That's why I wrote "for a few hundred
msec".
As mentioned before, the duty cycle will be extremely low, and
it'll be
less than 0.01%.
I am not going to sink 300W pulse into them but maybe 1-2W pulses.
--
Regards, Joerg
http://www.analogconsultants.com/
Okay , from Fig 5.
http://www.mouser.com/catalog/specsheets/smaj50.pdf the transient
thermal impedance is less than 10oC/W (~7) for the 100ms-1000ms
pulses, so it's a non-issue. But anything approaching continuous and
Rja goes towards 120oC/W in that flea package.




That is exactly what I meant. It is a non-issue _if_ you know how much

it will conduct at 80V, worst case maximum. And I don't, hence my

inquiry and my post here.



--

Regards, Joerg



http://www.analogconsultants.com/

You could combine the precision of a zener with the peak energy handling of the TVS to get your results- the zener sets the peak firing voltage, TVS is low voltage not critical except that its threshold for significant conduction exceeds steady state worst case maximum DC on line with sufficientmargin so that SCR shuts off by holding current deprivation.
Please view in a fixed-width font such as

Courier.

.

.

.

.

.

. ----------+-----------+------------

. | |

. | |

. | |

. ___/ ___/

. // \ // \

. --- ZD --- TVS

. | |

. | |

. | |

. | -----

. +---------/ \ SCR

. | -----

. | |

. [390] |

. | |

. | |

. | |

. ----------+-----------+--------------

.

That is a pretty good idea. Well, if I had the space for a big enough

SCR. But there is another problem: The SCR will not let go until the TVS

has largely come out of conduction and the decay of the 80V surges is

going to be very slow. That would cook out the TVS.



So essentially I have to leave everything under 80V untouched or sink

only a very small current whenever the circuit is below 80V. Just to

avoid overheating any of the protective devices.



The main issue is the fast onset of the 80V surge which creates ringing.

It's only this ringing I have to cap. So potentially I cold do something

with a high-pass but ... no space.



--

Regards, Joerg



http://www.analogconsultants.com/

What is the ringing frequency? All you have told is that it lasts for 100s milliseconds. You obviously do not have enough high frequency loss on your input. The fix would then be a series ferrite bead with a high frequency capacitor shunt to ground to kill the Q.

 

[Joerg]

[email protected] wrote:
[email protected] wrote:
Hello Folks, Does anyone know a manufacturer that
provides a wee bit more info in datasheets for
their TVS? The issue: We have SMAJ78CA in a board
because long cables, filters and such can cause
nasty ringing. The board must tolerate several
100msec bursts of 80V amplitude in a row, but only
on rare occasions and then such a burst is a
one-time event. Long story short, these things
won't clamp well before 100V is reached in the
ringing and then some stuff with 100V abs max
ratings can go phhhhut.
Many of these flea-sized TVS have really weak average
power handling capability, usually just 10s mW, so
those 100msec burst times of yours have a good chance
of blowing the SMAJ78CA. You better start with the
repetitive rate power handling of the part before you
waste a bunch of time looking for V/I limit
guarantees.
SMA isn't exactly flea-sized. They can take a watt or
two for a few hundred msec. That's all I am asking them
to do. [...]
No they can't. The Vishay ds says repetitive power
handling of 300W at 0.01% duty which comes to 30mW.
Of course not continuously. That's why I wrote "for a few
hundred msec". As mentioned before, the duty cycle will be
extremely low, and it'll be less than 0.01%. I am not going
to sink 300W pulse into them but maybe 1-2W pulses. --
Regards, Joerg http://www.analogconsultants.com/
Okay , from Fig 5.
http://www.mouser.com/catalog/specsheets/smaj50.pdf the
transient thermal impedance is less than 10oC/W (~7) for the
100ms-1000ms pulses, so it's a non-issue. But anything
approaching continuous and Rja goes towards 120oC/W in that
flea package.
That is exactly what I meant. It is a non-issue _if_ you know
how much it will conduct at 80V, worst case maximum. And I
don't, hence my inquiry and my post here. -- Regards, Joerg
http://www.analogconsultants.com/
You could combine the precision of a zener with the peak energy
handling of the TVS to get your results- the zener sets the peak
firing voltage, TVS is low voltage not critical except that its
threshold for significant conduction exceeds steady state worst
case maximum DC on line with sufficient margin so that SCR shuts
off by holding current deprivation. Please view in a fixed-width
font such as Courier. . . . . . .
----------+-----------+------------ . |
| . | | . |
| . ___/ ___/ . // \
// \ . --- ZD --- TVS . |
| . | | . |
| . | ----- .
+---------/ \ SCR . | ----- .
| | . [390] | .
| | . | | .
| | . ----------+-----------+-------------- . .

That is a pretty good idea. Well, if I had the space for a big
enough

SCR. But there is another problem: The SCR will not let go until
the TVS

has largely come out of conduction and the decay of the 80V surges
is

going to be very slow. That would cook out the TVS.



So essentially I have to leave everything under 80V untouched or
sink

only a very small current whenever the circuit is below 80V. Just
to

avoid overheating any of the protective devices.



The main issue is the fast onset of the 80V surge which creates
ringing.

It's only this ringing I have to cap. So potentially I cold do
something

with a high-pass but ... no space.



--

Regards, Joerg



http://www.analogconsultants.com/

What is the ringing frequency? All you have told is that it lasts for
100s milliseconds. You obviously do not have enough high frequency
loss on your input. The fix would then be a series ferrite bead with
a high frequency capacitor shunt to ground to kill the Q.

Sorry if I wasn't totally clear in explaining it. Goes like this:

Unit sits at normal voltage, 24-30V or so. Now comes a surge to 80V with
a very fast ramp-up. That causes a ringing of around 10kHz, decaying
within 3-4 cycles, highly damped. Then the 80V stays on for 100msec. At
the end of the 100msec period the voltage gradually sags back down from
80V to around 24-30V and this slow sagging can easily take a second. The
latter is why a triggered SCR presents a problem because the decay from
80V is slow. It's only the ringing that's fast and only that needs to be
suppressed.

Whatever circuit is in there, it must let go when below 80V.

 

[Joerg]

Hello Folks, Does anyone know a manufacturer that provides
a wee bit more info in datasheets for their TVS? The issue:
We have SMAJ78CA in a board because long cables, filters
and such can cause nasty ringing. The board must tolerate
several 100msec bursts of 80V amplitude in a row, but only
on rare occasions and then such a burst is a one-time
event. Long story short, these things won't clamp well
before 100V is reached in the ringing and then some stuff
with 100V abs max ratings can go phhhhut.
Many of these flea-sized TVS have really weak average power
handling capability, usually just 10s mW, so those 100msec
burst times of yours have a good chance of blowing the
SMAJ78CA. You better start with the repetitive rate power
handling of the part before you waste a bunch of time looking
for V/I limit guarantees.
SMA isn't exactly flea-sized. They can take a watt or two for a
few hundred msec. That's all I am asking them to do. [...]
No they can't. The Vishay ds says repetitive power handling of
300W at 0.01% duty which comes to 30mW.
Of course not continuously. That's why I wrote "for a few hundred
msec".
As mentioned before, the duty cycle will be extremely low, and
it'll be
less than 0.01%.
I am not going to sink 300W pulse into them but maybe 1-2W pulses.
--
Regards, Joerg
http://www.analogconsultants.com/
Okay , from Fig 5.
http://www.mouser.com/catalog/specsheets/smaj50.pdf the transient
thermal impedance is less than 10oC/W (~7) for the 100ms-1000ms
pulses, so it's a non-issue. But anything approaching continuous and
Rja goes towards 120oC/W in that flea package.



That is exactly what I meant. It is a non-issue _if_ you know how much

it will conduct at 80V, worst case maximum. And I don't, hence my

inquiry and my post here.



--

Regards, Joerg



http://www.analogconsultants.com/
You could combine the precision of a zener with the peak energy handling of the TVS to get your results- the zener sets the peak firing voltage, TVS is low voltage not critical except that its threshold for significant conduction exceeds steady state worst case maximum DC on line with sufficient margin so that SCR shuts off by holding current deprivation.


Please view in a fixed-width font such as
Courier.

.
.
.
.
.
. ----------+-----------+------------
. | |
. | |
. | |
. ___/ ___/
. // \ // \
. --- ZD --- TVS
. | |
. | |
. | |
. | -----
. +---------/ \ SCR
. | -----
. | |
. [390] |
. | |
. | |
. | |
. ----------+-----------+--------------
.
.

That is a pretty good idea. Well, if I had the space for a big enough
SCR. But there is another problem: The SCR will not let go until the TVS
has largely come out of conduction and the decay of the 80V surges is
going to be very slow. That would cook out the TVS.

So essentially I have to leave everything under 80V untouched or sink
only a very small current whenever the circuit is below 80V. Just to
avoid overheating any of the protective devices.

The main issue is the fast onset of the 80V surge which creates ringing.
It's only this ringing I have to cap. So potentially I cold do something
with a high-pass but ... no space.

How about a surge-rated surface-mount series resistor and a shunt cap
or zener? That's about what's in a telecom line network. There are
lots of parts designed specifically for that application. I have some
files around here somewhere, research I did for a friend...

If I had a surge-rated zener that would be great. But it must take >10A
current spikes.

 

[Joerg]

On Sunday, September 23, 2012 10:20:19 AM UTC-4, Joerg wrote:

[email protected] wrote:


On Saturday, September 22, 2012 9:06:34 PM UTC-4, Joerg wrote:

[email protected] wrote:

On Saturday, September 22, 2012 7:59:55 PM UTC-4, Joerg wrote:

[email protected] wrote:

Hello Folks, Does anyone know a manufacturer that provides
a wee bit more info in datasheets for their TVS? The issue:
We have SMAJ78CA in a board because long cables, filters
and such can cause nasty ringing. The board must tolerate
several 100msec bursts of 80V amplitude in a row, but only
on rare occasions and then such a burst is a one-time
event. Long story short, these things won't clamp well before
100V is reached in the ringing and then some stuff
with 100V abs max ratings can go phhhhut.

Many of these flea-sized TVS have really weak average power
handling capability, usually just 10s mW, so those 100msec
burst times of yours have a good chance of blowing the
SMAJ78CA. You better start with the repetitive rate power
handling of the part before you waste a bunch of time looking
for V/I limit guarantees.

SMA isn't exactly flea-sized. They can take a watt or two for a
few hundred msec. That's all I am asking them to do. [...]

No they can't. The Vishay ds says repetitive power handling of
300W at 0.01% duty which comes to 30mW.

Of course not continuously. That's why I wrote "for a few hundred
msec".
As mentioned before, the duty cycle will be extremely low, and
it'll be
less than 0.01%.
I am not going to sink 300W pulse into them but maybe 1-2W pulses.
--
Regards, Joerg
http://www.analogconsultants.com/

Okay , from Fig 5.
http://www.mouser.com/catalog/specsheets/smaj50.pdf the transient
thermal impedance is less than 10oC/W (~7) for the 100ms-1000ms
pulses, so it's a non-issue. But anything approaching continuous and
Rja goes towards 120oC/W in that flea package.




That is exactly what I meant. It is a non-issue _if_ you know how much

it will conduct at 80V, worst case maximum. And I don't, hence my

inquiry and my post here.



--

Regards, Joerg



http://www.analogconsultants.com/

You could combine the precision of a zener with the peak energy
handling of the TVS to get your results- the zener sets the peak
firing voltage, TVS is low voltage not critical except that its
threshold for significant conduction exceeds steady state worst case
maximum DC on line with sufficient margin so that SCR shuts off by
holding current deprivation.


Please view in a fixed-width font such as
Courier.

.
.
.
.
.
. ----------+-----------+------------
. | |
. | |
. | |
. ___/ ___/
. // \ // \
. --- ZD --- TVS
. | |
. | |
. | |
. | -----
. +---------/ \ SCR
. | -----
. | |
. [390] |
. | |
. | |
. | |
. ----------+-----------+--------------
.
.

That is a pretty good idea. Well, if I had the space for a big enough
SCR. But there is another problem: The SCR will not let go until the TVS
has largely come out of conduction and the decay of the 80V surges is
going to be very slow. That would cook out the TVS.

So essentially I have to leave everything under 80V untouched or sink
only a very small current whenever the circuit is below 80V. Just to
avoid overheating any of the protective devices.

The main issue is the fast onset of the 80V surge which creates ringing.
It's only this ringing I have to cap. So potentially I cold do something
with a high-pass but ... no space.

80 volt rail or what ever.

+
|
|
+-+---+
z |
A |
| |
| |
.-. |
| | |
| | |
'-' +
| |/
+-+-| some higher voltage type smt
.-. |>
| | +
| | |
'-' |
+-----+

low side or back feed to the supply rail
(created by AACircuit v1.28.6 beta 04/19/05 www.tech-chat.de)

All that can be fitted in a small area..

I've used such circuit and it wells very well.

All I really have is the square-footage of an SMA package. The
transistor is not trivial because a >10A spike must be snuffed out. Ok,
could use a FET but still, that spike is no small feat.

 

[Nemo]

Martin Riddle wrote:

Thanks, but those don't work because of the "90V +/-30%". At -30% that
would result in kablouie and at +30% the electronics behind it would be
dead.

What about a voltage regulator?

 

[Joerg]

What about a voltage regulator?

Too large :-(

 

[Jamie]

[email protected] wrote:

[email protected] wrote:

[email protected] wrote:

[email protected] wrote:

Hello Folks, Does anyone know a manufacturer that provides

a wee bit more info in datasheets for their TVS? The issue:

We have SMAJ78CA in a board because long cables, filters

and such can cause nasty ringing. The board must tolerate

several 100msec bursts of 80V amplitude in a row, but only

on rare occasions and then such a burst is a one-time

event. Long story short, these things won't clamp well

before 100V is reached in the ringing and then some stuff

with 100V abs max ratings can go phhhhut.

Many of these flea-sized TVS have really weak average power

handling capability, usually just 10s mW, so those 100msec

burst times of yours have a good chance of blowing the

SMAJ78CA. You better start with the repetitive rate power

handling of the part before you waste a bunch of time looking

for V/I limit guarantees.

SMA isn't exactly flea-sized. They can take a watt or two for a

few hundred msec. That's all I am asking them to do. [...]

No they can't. The Vishay ds says repetitive power handling of

300W at 0.01% duty which comes to 30mW.

Of course not continuously. That's why I wrote "for a few hundred

As mentioned before, the duty cycle will be extremely low, and

it'll be

less than 0.01%.

I am not going to sink 300W pulse into them but maybe 1-2W pulses.

Regards, Joerg

Okay , from Fig 5.

http://www.mouser.com/catalog/specsheets/smaj50.pdf the transient

thermal impedance is less than 10oC/W (~7) for the 100ms-1000ms

pulses, so it's a non-issue. But anything approaching continuous and

Rja goes towards 120oC/W in that flea package.

That is exactly what I meant. It is a non-issue _if_ you know how much

it will conduct at 80V, worst case maximum. And I don't, hence my

inquiry and my post here.

Regards, Joerg

You could combine the precision of a zener with the peak energy handling of the TVS to get your results- the zener sets the peak firing voltage, TVS is low voltage not critical except that its threshold for significant conduction exceeds steady state worst case maximum DC on line with sufficient margin so that SCR shuts off by holding current deprivation.

Please view in a fixed-width font such as






. ----------+-----------+------------

. | |

. | |

. | |

. ___/ ___/

. // \ // \

. --- ZD --- TVS

. | |

. | |

. | |

. | -----

. +---------/ \ SCR

. | -----

. | |

. [390] |

. | |

. | |

. | |

. ----------+-----------+--------------

.

That is a pretty good idea. Well, if I had the space for a big enough

SCR. But there is another problem: The SCR will not let go until the TVS

has largely come out of conduction and the decay of the 80V surges is

going to be very slow. That would cook out the TVS.



So essentially I have to leave everything under 80V untouched or sink

only a very small current whenever the circuit is below 80V. Just to

avoid overheating any of the protective devices.



The main issue is the fast onset of the 80V surge which creates ringing.

It's only this ringing I have to cap. So potentially I cold do something

with a high-pass but ... no space.



--

Regards, Joerg



http://www.analogconsultants.com/

What is the ringing frequency? All you have told is that it lasts for 100s milliseconds. You obviously do not have enough high frequency loss on your input. The fix would then be a series ferrite bead with a high frequency capacitor shunt to ground to kill the Q.

or simply put a R in series to lower the Q.

Jamie

 

[[email protected]]

Too large :-(

how much current does it need pass in normal operation?
would a fet and disconnect work?

-Lasse

 

[[email protected]]

[email protected] wrote:

[email protected] wrote:
[email protected] wrote:
Hello Folks, Does anyone know a manufacturer that
provides a wee bit more info in datasheets for
their TVS? The issue: We have SMAJ78CA in a board
because long cables, filters and such can cause
nasty ringing. The board must tolerate several
100msec bursts of 80V amplitude in a row, but only
on rare occasions and then such a burst is a
one-time event. Long story short, these things
won't clamp well before 100V is reached in the
ringing and then some stuff with 100V abs max
ratings can go phhhhut.
Many of these flea-sized TVS have really weak average
power handling capability, usually just 10s mW, so
those 100msec burst times of yours have a good chance
of blowing the SMAJ78CA. You better start with the
repetitive rate power handling of the part before you
waste a bunch of time looking for V/I limit
guarantees.
SMA isn't exactly flea-sized. They can take a watt or
two for a few hundred msec. That's all I am asking them
to do. [...]
No they can't. The Vishay ds says repetitive power
handling of 300W at 0.01% duty which comes to 30mW.
Of course not continuously. That's why I wrote "for a few
hundred msec". As mentioned before, the duty cycle will be
extremely low, and it'll be less than 0.01%. I am not going
to sink 300W pulse into them but maybe 1-2W pulses. --
Regards, Joerg http://www.analogconsultants.com/
Okay , from Fig 5.
http://www.mouser.com/catalog/specsheets/smaj50.pdf the
transient thermal impedance is less than 10oC/W (~7) for the
100ms-1000ms pulses, so it's a non-issue. But anything
approaching continuous and Rja goes towards 120oC/W in that
flea package.
That is exactly what I meant. It is a non-issue _if_ you know
how much it will conduct at 80V, worst case maximum. And I
don't, hence my inquiry and my post here. -- Regards, Joerg
http://www.analogconsultants.com/
You could combine the precision of a zener with the peak energy
handling of the TVS to get your results- the zener sets the peak
firing voltage, TVS is low voltage not critical except that its
threshold for significant conduction exceeds steady state worst
case maximum DC on line with sufficient margin so that SCR shuts
off by holding current deprivation. Please view in a fixed-width
font such as Courier. . . . . . .
----------+-----------+------------ . |
| . | | . |
| . ___/ ___/ . // \
// \ . --- ZD --- TVS . |
| . | | . |
| . | ----- .
+---------/ \ SCR . | ----- .
| | . [390] | .
| | . | | .
| | . ----------+-----------+-------------- . .


That is a pretty good idea. Well, if I had the space for a big
enough

SCR. But there is another problem: The SCR will not let go until
the TVS

has largely come out of conduction and the decay of the 80V surges
is

going to be very slow. That would cook out the TVS.



So essentially I have to leave everything under 80V untouched or
sink

only a very small current whenever the circuit is below 80V. Just
to

avoid overheating any of the protective devices.



The main issue is the fast onset of the 80V surge which creates
ringing.

It's only this ringing I have to cap. So potentially I cold do
something

with a high-pass but ... no space.



--

Regards, Joerg



http://www.analogconsultants.com/

What is the ringing frequency? All you have told is that it lasts for

100s milliseconds. You obviously do not have enough high frequency

loss on your input. The fix would then be a series ferrite bead with

a high frequency capacitor shunt to ground to kill the Q.

Sorry if I wasn't totally clear in explaining it. Goes like this:



Unit sits at normal voltage, 24-30V or so. Now comes a surge to 80V with

a very fast ramp-up. That causes a ringing of around 10kHz, decaying

within 3-4 cycles, highly damped. Then the 80V stays on for 100msec. At

the end of the 100msec period the voltage gradually sags back down from

80V to around 24-30V and this slow sagging can easily take a second. The

latter is why a triggered SCR presents a problem because the decay from

80V is slow. It's only the ringing that's fast and only that needs to be

suppressed.



Whatever circuit is in there, it must let go when below 80V.



--

Regards, Joerg



http://www.analogconsultants.com/

That's not ringing, that's a crap power supply losing regulation and breaking into oscillation. What's causing it to spazz out like that? Makes no difference, this calls more for an LDO than a clamp - there is this product that will fit the bill: http://www.ti.com/lit/ds/symlink/tps7a4001.pdf

 

[Joerg]

how much current does it need pass in normal operation?
would a fet and disconnect work?

About an amp. If it can be opened fast it could be done with one of
thiose 3*3mm DFN FETs. But needs that some electronics, and there's just
no space.