Automotive - reverse voltage protection thing

Hi!

I need to design a well-protected power supply for an automotive device,
well protected from all the nasty things described in ISO 7637 and similar
norms.

It's a common way to use a transil to protect from "dump load" surge, it's
perfectly ok for me. Yet another thing to protect from is a reverse
voltage.

My first idea was to use a series diode to protect from reverse voltage. I
wanted to connect it as the first component, anode to the battery, and a
transil just after that. Seems ok, but..

Well, the diode itself need to have a high level of allowable reverse
voltage (ISO 7637 says about -150V spikes, yet I've seen some papers saying
that reality is much worse). At the same time this diode need to handle
quite high peak current, other wise it will die after first "dump load"
thing. I guess this leads to a big/expensive element.

One possibility is not to use this diode at all, assuming that the transil
itself is enough - in case of reverse polarity, it will conduct just like a
normal diode, effectively limiting the spike to it's -Uf. It would need a
fuse to protect the transil from permanently reversed power supply.

What is actually used in practice for such a protection? Looking for some
info I've found RBO40 component from ST, which uses a series diode, TVS
across output for load dump protection, and another across the diode.
Unfrotunately, this part is quite hard to get here.

Best Regards,
Przemyslaw

--

 

See...

http://www.analog-innovations.com/SED/PerfectDiodeForChargerIsolation.pdf


...Jim Thompson

 

What's the load ?

Graham

 

For most stuff, I use a series fuse on the + side followed by an appropriate
uni-polar TVS across the + and GND. In extreme overvoltage situations (bad
regulator in the alternator) the TVS will conduct and blow the fuse. For
short duration transients, it just shorts the excess and makes heat. It's
been my experience that these transients have very little power behind them,
and a 1500 Watt SMC TVS is plenty. In a reverse polarity situation, the TVS
acts as a regular diode and blows the fuse.

As the circuitry or system I'm protecting go up in value, things can get a
little more complex.

Chris

 

If you don't have a problem with "mechanical shokes" the easiest way
probably is a very old ancient tech: Use a ordinary light bulb of approbiate
voltage/power-rating. They exists in a very large design bandwidth and are
very cheap!

The circuits is simply as follows:
The light bulb is in series to input and load connected. Behind the bulb
connect a moderately powered zener-diode in parallel connection to the load.
The parasitic circuit of the light bulb is like a ohmic series resistor and
a air-coil in the range of maybe 100nH to 1uH. Just test it with a
LCR-meter.
If a over-current situation exists for a longer time of say 100msec the
ligth-bulb works with approx. the 10-times series resistance (It lights!).

Even you can use a cheap socket to let the bulb exchangeable...

You can play with the various scenarious to see that it works very good. The
circuit works evenly good if you have a battery-powered small device to
protect it against reverse-battery insert.

I've never seen a more straight-forward circuit.

Or beat it!!


Hope that helps -
Henry

www.ehydra.dyndns.info

 

Nice idea. Nice photos too.

There is a newer replacement for the lightbulb that is immume to
"mechanical shakes":

http://www.circuitprotection.com/polyswitch.asp

 

Sorry Joe, I knew them. They suffer from expanded resistance after several
events. And where is the inductance?

And no, I'm not an english native speaker

Having all the different possibilities is what electronics make interesting.

- Henry

 

Expanded resistance = increased resistance?

I use a lot of them and have had really good results. No mechanical
deformation with a short. Always reset (so far). Resistance too low
to measure accurately with a two wire ohm meter.

Perhaps you are exceeding some voltage specification?

They give the reliability as less than manual reset circuit breakers
1.7 times more likely to fail. (I'd really like to see the specs on
some of those automatic reset circuit breakers - they seem to fail
frequently)

2.6 times more likely to fail than a fuse

Fuses work when they open interrupting power, poly switches also fail
open (they stop resetting - so where's the harm?)

These things are great in my opinion. I've never seen specs on the
number of "events" one can suffer and keep resetting.

I was complaining about the little, bimetallic sealed in glass,
circuit breakers that kept failing; a friend gave me a Raychem poly
switch and they have worked beautifully.

There are "events" and then there are "EVENTS." In low voltage
circuits pulling 75% of their ratings they seem able to work fine.
Push the voltage limits of the devices and I have seen them fail - go
over the voltage limit and they smoke.

For protection in an auto or battery pack, or unregulated LV supply -
this is as good as it gets IMHO. Only downside is if you are working
at more than 50+% of the trip rating you may have to remove the load
or they take a long time to reset.

They have a new device called a Poly Zen - over current and over
voltage in one three terminal package (alas - only surface mount so
far). Great synergy: any heat being dissipated by the zener lowers
the trip point of the poly switch.

Since this is essentially intended as a self resetting fuse
replacement. . . Maybe you should exam your designs and design out
some of those events.

Inductance? a few ferrite beads would probably add more inductance
than the filament of a light bulb.

 

[...]

Yes, I meant "increased resistance".

Hm. Is it a flame against my design??

I welcome that polychem upgrades their product line from time to time.
Personally I tested them 10 years ago and was then more satisfied by a
ordinary melting fuse. Mainly because of lower resistance.

If they integrate a zener-diode this should work even better.

I don't have the time to answer all your points here, sorry.


Seeing is believing!
So just get a destroyed flourescent energy-saving lamp and disassemble it.
For that, unsolder first the small buck looking like lead. Don't know what
it exactly is in english There is wire internal soldered that must be
free or you will crack the thing by opening.
Then unlock the case it with a small screwdriver on the "circle". Open it.
If you're have a better designed (higher price perhaps) one, you will see a
light bulb inside. That can be a shaft of 2cm or such.
Next, unsolder the four wires to the flourescent lamp. Those are two
light-bulbs at the end of the tube. They light when the lamp starts to lower
the emission energy. Likely one of them is destroyed but the other
functions.
Test the one which looks working for resistance and inductance! You will see
about 20 ohms resistance and 70 to 100uH inductance! That is because the
internal wire is relatively long to form the coil.

That inductance is not saturable because it is essentially a air coil!

Light bulbs for lower voltage may have lower inductance. I don't examined it
intense.

You can get every where a light-bulb and a zener-diode for not much money.

BTW:
You can do whatever you like but I think it should be "ethical engineered".
Not ethical is, if you insert as a user a battery in your brand-new toy
reversely by accident and shortly think, damn I made a mistake! Hurry,
reverse it and see if it works again. And it fails! Good for
shareholder-value in the short-term, but not ethical designed.
That is my opinion about working!

regards -
Henry

 

We call it sarcasm. Nothing personal . . .

I'd agree with you there. Most of the time a fuse is a good choice.
The poly switches have a niche market where replacing a fuse isn't
always convenient - or like you mention making it more foolproof so
the consumer isn't screwed because he make a mistake.

If you were going to use a light bulb as a current limiter - a poly
switch probably has a lower initial resistance and a sharper turn off
point.

I don't have an inductance bridge. I have to take your word. That
seems like an awful lot of inductance for a lamp filament though . . .
I'm skeptical.

That is rapidly changing here. Incandescent lamps are being
supplanted by LEDs in lots of applications. The old bayonet bulbs and
sockets are becoming more scarce and expensive.

no argument from me

In the "old days" we used "ballast lamps" in teletype applications as
a sort of constant current source.

I noticed a web site where they are using banks of 100 watt light
bulbs in place of fixed resistors for single ended class A audio
amplifiers.

 

I will memorize this. But a second time: no usable inductance!

You impressed me ;-)
So I took al little time of work and researched it.
Bad: My LCR-meter seems to measure wrong if the resistance of the coil is
unusal high - that is the point with light bulbs!
So I must correct the mentioned 70 to 100uH - sorry!

Here is an article describing the mechanical construction of the bulb wire:
http://physicsed.buffalostate.edu/pubs/TPT/TPTDec99Filament.pdf
Look in the references!

From there for typical 60 watts 120 volts lamp:
length of wire = 53.3 cm
diameter of wire = 4.6e-5 m
length of coil = 8.3 cm
number of turns = 1130

The bulb wire consists of two coils. I describe the inner coil and think
this is a reasonable approximation for a first try to calculate the
inductance. The outer coil is a much bigger structure. So there is not much
interaction between them. Also I know that the real length is a bit
different because of the second outer coil winding. PLEASE HELP GREAT
MATHEMATICANS!!

I used the common rfsim99 program to calculate the inductance as an
air-coil:
But first calculate the diameter of one turn:
Divide the (length of coil) by (number of turns). And then the result by pi.
This gives the diameter of one turn. Not very accurate as the turns space
between.
The result is for the diameter of one turn: 0.15 mm
I think that agrees with what I see with my eyes on the bulb.
Enter this in rfsim99 Tools->Component->Inductor
Don't forget to think about the units used!

Result: 350nH

Not bad but very small. Should be enough for very fast transients.
Is there anyone capable to verify this with a high-quality LC-meter?

BTW: The resistance changes by a factor of approx 5 to 15 between cold and
hot bulb. Lamps with voltages lower 24 volts usually have one coil winding
only!

Would be interesting to make a table of common lamps here.

Hm. Here it is not. Personally I think LEDs are driven by marketing people
at the moment and cannot change the world very fast.
They are interesting for special light conditions already now but not for
general lighting. You can't beat the incandescent lamp efficiency and cost
for.

- Henry

 

They are interesting for special light conditions already now but not for

My fault: Please set "incandescent" to "flourescent".

- Henry

 

believable

 

I finally had a little time to go back and check the link you provided
on the construction of the lamps. Interesting!

I didn't know they incorporate a fuseable link to the hot contact.

And further thinking about calculating inductance from
length/diameter/turns . . . I don't know how sophisticated the
program you refer to is . . . while simple calculations for a single
layer coil are usually very accurate - they are not when the length is
significantly long with respect to the diameter.

And for all these air core inductance calculations, every turn of the
coil has to have the same axis as the ones to either side of it -
something that doesn't happen in a lamp filament even in a single
layer coil and with the double wound design - the inductance would
probably be much lower. Winding loops with loops guarantees that no
two turns has the same axis.

Ferrite beads are looking much better again.

 

What sort of peak current are you thinking of? What's the load?
High-current high-PIV diodes are not rare these days. And a little bit
of impedance (probably already there in the transil if not the wiring)
will limit the current most likely. Energy-wise the
reverse-voltage-protection diode will have much less energy dissipated
in it than the transil will dissipate in a load dump so I don't see why
it would be big or expensive in comparison.

Tim.

 

I didn't know they incorporate a fuseable link to the hot contact.
Seems that this feature is lost in the last years.

rfsim99 is a very good prog and free. If you like rf you cannot miss it!
I entered the values in another calculator and got the same result:
http://www.dl5swb.de/html/mini_ringkern-rechner.htm
(Have a english GUI also). Even good stuff!

No one in the NG sponsered a high-quality measurement. I cannot, sorry.

Surely there are mathematical sophisticated people out there but I didn't
got an answer to my posting.
So calculating it right for two coils is left over!
I don't think it will change the value my more than 30% perhaps.

As long as they don't saturate, why not. Cheap and effective.

- Henry

 

I'm no math genius. The simple calculations work great for simple
coils where the length is no more than twice the diameter - and longer
if absolute accuracy isn't necessary. And it isn't half bad when used
for conical coils or spirals.

It's been awhile, but I think the calculation for inductance in
microhenries is:

r2 n2
L = ---------------------------
9r + 10 h

L in micro henries
r radius in inches
n number of turns
h height - length of winding in inches

All other things being equal, thicker wire gives you more inductance
than thinner wire. Suggesting that inductance falls off as the coil
is stretched.

Taking that and trying to apply it double wound helix is probably nigh
impossible.

Normally the inductance increases with the square of the turns, but
that assumes that each turn is closely coupled to every other turn -
they have to aid in magnetic field strength without a lot of leakage
flux - stretch the helix and leakage increases, start shifting the
axis so only a portion of the magnetic field from one turn couples to
the next and leakage probably increases greatly.

I wouldn't presume to guess at the actual inductance of a light bulb
knowing the turns and diameter of the primary (smaller) helix.

That's where I'd want a good reactance bridge - one that can be
compensated for a high resistance. Impedance (inductance and
resistance together) is a vector function - in the case of a lamp
filament the resistance probably accounts for 99.9% of the total
resistance in leg of the bridge.

 

My formula used is already length corrected.

Yes. That implies a measurement frequency of at least 1MegaHz.
Or you cool the bulb very cold. E.g. liquid nitrogen.

I'm very busy designing a class E rf amplifier but when I can spent more
time I will try to find it out.
Interesting to see that Google knows relatively nothing about bulb
inductance.

What is the inductance of a started fluorescence lamp or the plasma in a
photographic gas-discharge lamp? ...

- Henry