Surviving automotive load dump

[J.A. Legris]

I think you are going the wrong way with this. Check the load dump spec and
you will find that it specifies a series resistance as well as the peak
voltage. In most cars you can get away with a transient suppressor zener
diode type device eg Multicomp 1.5SMCJ33A. When I was doing automotive
stuff (a few years ago now) each ECU (Electonic Control Unit) had its own
trasnient supressor capable of takling the whole load dump so in real life
they all got an easy time (there being many ECUs in //). If thsi approach
works for you then it's one fairly cheap sm device.

125V is a a very big load dump so you won't easily find integrated
regulators that will take it. The usual suspects (ST, Infineon, NS) all make
regulators that will take typical automotive "12V" inputs and stand reverese
supply, load dump, 24V etc.

Thanks for the info. I considered using one of those devices several
years ago, but they were expensive and didn't seem to cover the
worst-cases. Can you suggest some relevant publications?

 

[Winfield Hill]

Robert Adsett wrote http://www.mouser.com/search/refine.aspx?Mpc=Supertex+MOSFETs+-+Depletion+Mode+N-Channel&Mpcn=68902

Even enhancement-mode MOSFETs have a current-regulating capability.
See the output diagram in their datasheets. Depends on Vgs and Id.
I don't know if this mode is effective in every type made. But I've seen
that in every datasheet I seen.

Transconductance devices, yes, with their operating drain current
controlled by their gate voltage. But that's not a current-regulating
capability as we think of it, because you have to separately supply a
programming voltage. Depletion-mode current-regulating FETs work
without an external voltage: A self-biasing gate-drain current-sensing
resistor nicely does the job.

Maybe there is a power problem inside which could destroy the
MOSFET. I never thought it to the end. Suggestions here?

It's just a simple matter of power dissipation and thermal resistance
vs. your maximum junction temperature spec.

 

[John Devereux]

Probably a depletion-mode power MOSFET as a current-source? See
as intro: http://www.ehydra.dyndns.info/pdf/app_notes/AN-D11.pdf

Well, that's pretty cool. The one in the application note doesn't look
easy to get hold of; E.g. Digikey have never heard of them. [ snip ]

(I don't think I have ever seen a depletion mode FET - outside of "Art
of Electronics" - will have to read up on them again now!).

There are about two dozen types to chose from, made by a half-dozen
manufacturers. I've often called the LND150 (the part suggested in
the app note) "my favorite" and I've designed it into various ASCII
circuits posted here (dunno why Jim doesn't remember that).

Mouser stocks a range of Supertex depletion MOSFETs, like the
1mA LND150, plus higher-current parts (to 800mA), and even one
type (DN2540N5) in a TO-220 package.

They're super-easy to use in this application:

HV spikes in
------,
| depletion-mode
D MOSFET
G ---,
S |
| |
__|__ |
| | |
| |---+--- LV out
|_____|
| \
gnd LDO regulator

I didn't show bypass caps, protective zener, etc. The LDO regulator
runs from the MOSFET's -Vgs gate-bias voltage appropriate for the
current drawn by the load.

I don't want to dissapate much power in normal use, so I have a SMPS
downstream.

How about this:



Supertex
DN3525
Ferrite .------------.
----UUU-->|-----+|+---------| |
| | |V| | | SMPS |
z .-. === | | |
A 10k| | | | '------------'
| | | | | |
150V TVS '-' | --- |
| '------' --- |
| | |100uF |
| z | |
| A 33V | |
| | | |
------------------------------------------

 

[Jim Thompson]

Probably a depletion-mode power MOSFET as a current-source? See
as intro: http://www.ehydra.dyndns.info/pdf/app_notes/AN-D11.pdf

Well, that's pretty cool. The one in the application note doesn't look
easy to get hold of; E.g. Digikey have never heard of them. [ snip ]

(I don't think I have ever seen a depletion mode FET - outside of "Art
of Electronics" - will have to read up on them again now!).

There are about two dozen types to chose from, made by a half-dozen
manufacturers. I've often called the LND150 (the part suggested in
the app note) "my favorite" and I've designed it into various ASCII
circuits posted here (dunno why Jim doesn't remember that).

Mouser stocks a range of Supertex depletion MOSFETs, like the
1mA LND150, plus higher-current parts (to 800mA), and even one
type (DN2540N5) in a TO-220 package.

They're super-easy to use in this application:

HV spikes in
------,
| depletion-mode
D MOSFET
G ---,
S |
| |
__|__ |
| | |
| |---+--- LV out
|_____|
| \
gnd LDO regulator

I didn't show bypass caps, protective zener, etc. The LDO regulator
runs from the MOSFET's -Vgs gate-bias voltage appropriate for the
current drawn by the load.

I don't want to dissapate much power in normal use, so I have a SMPS
downstream.

How about this:



Supertex
DN3525
Ferrite .------------.
----UUU-->|-----+|+---------| |
| | |V| | | SMPS |
z .-. === | | |
A 10k| | | | '------------'
| | | | | |
150V TVS '-' | --- |
| '------' --- |
| | |100uF |
| z | |
| A 33V | |
| | | |
------------------------------------------

I'm not sure. The way I read the data sheet, the device cuts of for
VGS = -2.2V, and normal operation should maintain VGS >= 0

And what is the hold-up time for your SMPS? As someone else pointed
out, load dump can last over 400ms.

...Jim Thompson

 

[John Devereux]

I'm not sure. The way I read the data sheet, the device cuts of for
VGS = -2.2V, and normal operation should maintain VGS >= 0

I think you may be reading it wrong (or perhaps the mosfet symbol is
wrong). It is supposed to work like an emitter follower. The 33V zener
should limit the downstream voltage to ~33V.

And what is the hold-up time for your SMPS? As someone else pointed
out, load dump can last over 400ms.

The circuit should keep giving out 33V during the load dunp.

 

[Jim Thompson]

I think you may be reading it wrong (or perhaps the mosfet symbol is
wrong).

I understood that.

It is supposed to work like an emitter follower. The 33V zener
should limit the downstream voltage to ~33V.

I'll give it a whirl on PSpice.

The circuit should keep giving out 33V during the load dunp.

How much current is the SMPS drawing?

...Jim Thompson

 

[John Devereux]

I understood that.


I'll give it a whirl on PSpice.


How much current is the SMPS drawing?

<200mA, although perhaps more during startup if allowed by the supply
to it. It would have a 1A built-in current limit. (It will probably be
a 34063 or a higher frequency version NCP3063).

 

[John Devereux]

<200mA, although perhaps more during startup if allowed by the supply
to it. It would have a 1A built-in current limit. (It will probably be
a 34063 or a higher frequency version NCP3063).

I should clarify that the 200mA is the expected *load* current. The
supply current to the SMPS obviously vary with supply voltage.

 

[Henry Kiefer]

| > Even enhancement-mode MOSFETs have a current-regulating capability.
| > See the output diagram in their datasheets. Depends on Vgs and Id.
| > I don't know if this mode is effective in every type made. But I've seen
| > that in every datasheet I seen.
|
| Transconductance devices, yes, with their operating drain current
| controlled by their gate voltage. But that's not a current-regulating
| capability as we think of it, because you have to separately supply a
| programming voltage. Depletion-mode current-regulating FETs work
| without an external voltage: A self-biasing gate-drain current-sensing
| resistor nicely does the job.

Yes.

|
| > Maybe there is a power problem inside which could destroy the
| > MOSFET. I never thought it to the end. Suggestions here?
|
| It's just a simple matter of power dissipation and thermal resistance
| vs. your maximum junction temperature spec.

Some MOSFET specs specifiy a maximum Id regardless of other values.
So I'm not sure.

- Henry

 

[Jim Thompson]

<200mA, although perhaps more during startup if allowed by the supply
to it. It would have a 1A built-in current limit. (It will probably be
a 34063 or a higher frequency version NCP3063).

I'm not quite sure how to interpret the "pulsed" safe operating area
chart. I couldn't find the pulse width definition on the data sheet,
though I didn't look really hard ;-)

...Jim Thompson

 

[Winfield Hill]

Winfield Hill wrote

Some MOSFET specs specifiy a maximum Id regardless of other values.
So I'm not sure.

Same thing, except that spec attempts to describe the maximum
continuous power dissipation capability of the lead-bonding wires.

 

[Winfield Hill]

Probably a depletion-mode power MOSFET as a current-source? See
as intro: http://www.ehydra.dyndns.info/pdf/app_notes/AN-D11.pdf

Well, that's pretty cool. The one in the application note doesn't look
easy to get hold of; E.g. Digikey have never heard of them. [ snip ]

(I don't think I have ever seen a depletion mode FET - outside of "Art
of Electronics" - will have to read up on them again now!).

There are about two dozen types to chose from, made by a half-dozen
manufacturers. I've often called the LND150 (the part suggested in
the app note) "my favorite" and I've designed it into various ASCII
circuits posted here (dunno why Jim doesn't remember that).

Mouser stocks a range of Supertex depletion MOSFETs, like the
1mA LND150, plus higher-current parts (to 800mA), and even one
type (DN2540N5) in a TO-220 package.

They're super-easy to use in this application:

HV spikes in
------,
| depletion-mode
D MOSFET
G ---,
S |
| |
__|__ |
| | |
| |---+--- LV out
|_____|
| \
gnd LDO regulator

I didn't show bypass caps, protective zener, etc. The LDO regulator
runs from the MOSFET's -Vgs gate-bias voltage appropriate for the
current drawn by the load.

I don't want to dissapate much power in normal use, so I have a SMPS
downstream.

How about this:



Supertex
DN3525
Ferrite .------------.
----UUU-->|-----+|+---------| |
| | |V| | | SMPS |
z .-. === | | |
A 10k| | | | '------------'
| | | | | |
150V TVS '-' | --- |
| '------' --- |
| | |100uF |
| z | |
| A 33V | |
| | | |
------------------------------------------

Worth playing with on the bench...

 

[John Devereux]

Worth playing with on the bench...

It seems to work nicely in simulation. There is a current limit
effect too due to the characteristics of the fet. I also added a
capacitor across the zener for a slow-start. This reduces the inrush
current to "Idss" (fet current at 0V gate drive).

The current is only ~30mA during the load dump, thanks to the SMPS
taking a lot less at high input voltage.

The alternative to all this is Joe Legris' suggestion
___
o>|--|___|-o----o----o---
1N4007 3R | | |
z --- ---
A === ---
| | |
o----------o----o----o--
SA28A 100u 100n

I am finding the resistor to be the problem here. High surge
resistance seems to imply Wirewound, which are relatively expensive,
especially surface mount ones. (While searching for some, I note the
Ohmite site says that all their "high energy" ones are wound by hand!)

A 2W or 3W one is quite large too.

 

[xiaodragon]

per my experience, 27V MOV is right solution and you slo can selsect
PTC.
load dump is a commom requirement at automotive application. some OEM
sugget to use MOV, but i have seen JCI's board that use PTC.
you can use surface mounted TVS.

"John Devereux дµÀ£º
"

 

[Jamie]

per my experience, 27V MOV is right solution and you slo can selsect
PTC.
load dump is a commom requirement at automotive application. some OEM
sugget to use MOV, but i have seen JCI's board that use PTC.
you can use surface mounted TVS.

"John Devereux дµÀ£º

Hmm, last time i knew, PTC positive Thermo coefficient devices?

 

[Henry Kiefer]

| Henry Kiefer wrote:
| > Winfield Hill wrote
| >>
| >> It's just a simple matter of power dissipation and thermal resistance
| >> vs. your maximum junction temperature spec.
| >
| > Some MOSFET specs specifiy a maximum Id regardless of other values.
| > So I'm not sure.
|
| Same thing, except that spec attempts to describe the maximum
| continuous power dissipation capability of the lead-bonding wires.

Interesting. The problem is that the wire system is not specified in most datasheets.

Seems, that I must do experiments.

- Henry

 

[Robert Adsett]

| Henry Kiefer wrote:
| > Winfield Hill wrote
| >>
| >> It's just a simple matter of power dissipation and thermal resistance
| >> vs. your maximum junction temperature spec.
| >
| > Some MOSFET specs specifiy a maximum Id regardless of other values.
| > So I'm not sure.
|
| Same thing, except that spec attempts to describe the maximum
| continuous power dissipation capability of the lead-bonding wires.

Interesting. The problem is that the wire system is not specified in most datasheets.

It's usually in footnotes along the lines of "total current limited by
package"

Robert

 

[Henry Kiefer]

| In article <[email protected]>, Henry
| Kiefer says...
| > | Henry Kiefer wrote:
| > | > Winfield Hill wrote
| > | >>
| > | >> It's just a simple matter of power dissipation and thermal resistance
| > | >> vs. your maximum junction temperature spec.
| > | >
| > | > Some MOSFET specs specifiy a maximum Id regardless of other values.
| > | > So I'm not sure.
| > |
| > | Same thing, except that spec attempts to describe the maximum
| > | continuous power dissipation capability of the lead-bonding wires.
| >
| > Interesting. The problem is that the wire system is not specified in most datasheets.
|
| It's usually in footnotes along the lines of "total current limited by
| package"

Yes. Sometimes.

To place the problem a level higher:
If it is so, that MOSFETs cannot be burned if the bond wire is thick enough. Why it is then, that sometimes Microcontroller pins
destroy to internal short if overloaded?

- Henry

 

[martin griffith]

Hi,

As well as the usual spikes and brownouts, automotive systems have a
"load dump" condition, where the supply can e.g. rise to ~125 volts
for ~100ms.

What is the best way to protect against this? I need fairly low cost,
and small space (SMT). It must be a very common requirement, but all
I can find are some giant MOVs.

My best effort so far is a 4-transistor discrete "pre-regulator" to
stand-off the pulse (made from high voltage BJTs). Cheap in parts but
seems overly complicated.

Current would only be ~200mA, & must not drop more than a volt or two
during normal operation.

Any ideas?

FWIW I found the Ford EMC spec here
http://www.fordemc.com/docs/download/ES-XW7T-1A278-AC.pdf
which seems to be the "current" version.P81onwards has some neat
waveforms

This is also interesting, a comparison between VM EMC specs
http://www.autoemc.net/Papers/Test/OHaraGenericEMCStd.pdf



martin

 

[John Devereux]

FWIW I found the Ford EMC spec here
http://www.fordemc.com/docs/download/ES-XW7T-1A278-AC.pdf
which seems to be the "current" version.P81onwards has some neat
waveforms

Thanks martin, I hadn't seen these. These car manufactures standards
quote a maximum "load dump" spec of 60V (but I have seen 125V
mentioned elsewhere).