Re: Searching for engineer experienced in 24V vehicle electrical design.

From: GerberMultit00l (gerbermultit00l_at_yahoo.com)
Date: 03/11/05 

Date: 11 Mar 2005 15:16:57 -0800

First of all let me say thank you for the sudden overwhelming burst of
support. I'll try to address all responses if it makes sense to do so.

1.) Rich Grise Wrote
>Use a car starter solenoid for the relay. Better yet, truck. They
>handle hundreds of amps, albeit intermittently. You could use both -
>the starter solenoid closes a few tens or hundreds of millisecs before
>the main one, and stays closed for awhile when the main one opens;
>but you might not need this.

>Obviously, you are running this thing right on the hairy edge of its
>spec - you just need a beefier part."

Unfortunately this is not an option. I am confined to the space
allocated to me in our product packaging. I have an area of
approximately 26mm wide x 26mm long x 30mm high to fit a relay
solution. This device also needs to be PCB mounted. And finally, it
must be able to function in the temperature range from -40C to 72C. It
must survive without damage (non operational) from -50C to 125C.

I am using the chosen device near its rated limit but still within a
reasonable safety margin of 15% below published limit (which the
manufacture says is very conservative). I would like to see a larger
safety margin and am working on that.

My problems start to occur when system voltage increases out of the
specified limits for sustained durations of time (as will occur in any
vehicle from time to time). I tried implementing a new over voltage
protection scheme that will shut the relay off and protect the contacts
long before system voltage rises to a level that will be damaging to
the relay.

2.) Tim Shoppa wrote:
>What are your "stock relays" rated at? 0.4mm seems like a really
>small gap for even a 12V power relay."

Take a look at the relay specs for yourself here:
http://relays.tycoelectronics.com/datasheets/134J-VF7.pdf

Tim also wrote: "
>> They seem to start arcing at about 36V which is outside the normal
>>operation range of the vehicle and should be good enough for me.

>Way too close for me! Especially in an environment with extremely
>broad temperature and humidity requirements..."

True, I'm playing it close. However, these relays have been used in
rougher environments than what I am attempting and seem to survive.
I'm confident that with a little tweaking I can get this to work.

3.) Spehro Pefhany wrote:
>>Completely off the subject, but I suddenly wondered whether wire
>>wound resistors introduce significant inductance. I've always
>>thought of them as just resistors that were manufactured a different
>>way.

>Yes, some most certainly do. It can cause problems with current >sense
resistors, for example.

>Of course in this case, you've got the inductance of the resistor plus

>the inductance related to the loop area of the wiring. And
high-current >DC circuits just love to arc away anyhow.

>One way around it would be to use a hybrid switch."

I do not believe I have a big problem with inductance. I say this
because if the load did have a large amount of inductance I should see
a fairly large negative voltage spike (at least a couple hundred volt)
when turning off the load. I see a negative going voltage spike of
about 8V, not even enough to go below ground from the 28V starting
point.

Loop inductance due to wiring harness may be an issue with my test
setup. My product is inside an environmenal chamber and the power
supply is outside the chamber. The power lead going to the product is
approximately 3m and so is the ground return line. Wire gage is 8AWG.
I do see a small bump in voltage across the main power terminals of my
product (about 10V for 1us) when it makes or breaks the load. This
bump decreases as I lower load current so it must be inductance.

The in-vehicle application might be different. The power lead will be
restricted to below 2m and the ground return will be tied to the
vehicle chasis local to the product.

Now Spehro, how about this Hybrid Switch you were talking about?

4.) John Fields wrote:

>I think the most important things to do when trying to quench an arc
>are to get the contacts as far apart as possible as quickly as
>possible, and if you know that you're going to be using them in that
>kind of service to get them made with the proper contact material.

The manuacture offers a number of contact materials for this device. I
was contemplating trying out Silver Cadmium Oxide, Silver Tin Indium
Oxide, and Silver Copper Nickel. However, the manufacture sticks by
their guns saying that fine grain Silver is still the best material for
my application.

John also wrote:

>Opening up the gap between the contacts on relays you already have
>might help, but it may be that lowering the contact pressure by doing
>that will hurt you in the long run by allowing bounce to last for a
>longer time and changing the thermal characteristics of the contacts.
>With less pressure their resistance will increase slightly, causing
>them to heat up slightly, increasing their resistance a little more...

>Also, because of the reduced pressure, the contacts won't separate >as
quickly as they did before, which will give the arc a chance to live
>longer. It's probably not _that_ serious, but they were designed the
>way they were for a reason and when you start fooling around with >the
mechanicals it doesn't usually end up serendipetously.

I don't know if I agree with what you say in this instance. I am
increasing contact gap by filing down the armature stop of the relay so
the entire armature mechanism can swing away from the opposing contact
a little more. I am not affecting the mechanism the contact is mounted
to or its spring tension. You would have to see the internals of the
relay to understand what I say. I wish I could post pictures. The
manufactures images of this relay only show it with a cover, none
without.

Think of this relay design being like a door with a door stop behind it
preventing the knob from hitting the wall. A spring pulls the door
open and towards the door stop. I'm just grinding the door stop down a
little. The spring is under a certain amount of tension when the
armature is held 0.4mm from the opposing contact (stock relay). The
spring wants to pull the contact back to rest but it can't because of
the doorstop. By grinding the doorstop down I allow the door to swing
further back towards rest. The spring is still under tension, just a
little less since it is closer to rest position.

Now say someone wants to close the door. It takes a larger magnetic
force to get it started on its way back to close because of the
increase in distance (equivalent to a higher pull-in voltage
requirement). Once the door passes the 0.4mm mark on its way towards
being closed (the opposing contact) it is just like a stock unit. The
hold in force is the same whether the door was originally opened 0.4mm,
0.8mm, or 1.2mm

Drawing the armature in from a further distance will requires more
pull-in force. But, I have more than adequate pull in force now. A
stock relay pulls in at about 14.4V at 25C. I do not need to activate
the relay unless system voltage is above 25V. So, if the pull in
voltage of the relay increased to 16V because of my modification it
still does not affect me.

This discussion about weak contact pressure brings up an important
point. I might have found something that contributed to my recent
failures. I was driving the relay coil with a specific high voltage
NPN small signal transistor. I found that the relay coils resistance
was about 100 ohm lower than what was published. So, the required
drive current for this coil went up. The transistor did not have
enough DC current gain at higher temperatures to keep up. Therefore
the transistor was not turning on all the way. There was only 14V
across the coil! This was barely enough to pull it in, and in some
instances it would not. Since then I replaced this transistor with one
that has more than adequate current gain. The armature now snaps into
place at any temperature.

>I think you're going to wind up in trouble unless you get the right
>relays for the job. That is, _manufactured_ with the proper contact

>spacing, current rating and contact materials. What does your
>European competition use?

I agree. I chose this relay for one reason. It is the only one I
could find that fits my packaging requirements that can operate at the
capacity I need it to. It just so happens that the stock unit cannot
handle those rare instances where system voltage ramps up out of spec.
However, lucky for me this relay is also highly customizable. I can
have it built for what I need. I just need help in determining what is
good enough.

What do my competitors use? I could not tell you. This is a new
product to the market and we do not have competitors. That, and I
was asked to design a device for a market I am not familiar with.
Everything I learned about 24V vehicle systems so far either from a
lacking SAE specification on the subject or just hear say. So, I'm
venturing into a new frontier. Not the best situation to be in, but I
am there nevertheless.

5.) Don Murray wrote:

>Hi,
> your problem is an old one when switching high currents and the
>practical solution to this is a relay with blowout magnets at the
>contacts.These extinguish the arc quickly by forming a magnetic field
>thst repels the plasma field .hope this helps

Yep, I looked into this. Not practical for the small outline of this
relay. I asked the manufacture to try and find a solution.

6.) Bjarne Bäckström wrote:

> One solution, that I've used sometimes, is to simply put a (FET)
>transistor across the contacts. Turn the transistor on shortly before
>the contacts open, and turn it off when they are fully open. It's a
>simple solution -- and often cheaper than the alternatives.

I also considered this design approach. It may still be feasible and I
intend on using it as a last ditch effort. However, there are a few
obstacles to overcome.

Since I am switching my load with a high side drive I would need a
P-Channel FET across the contacts. You just can't find high voltage,
high current P-CH FETS with a low enough on resistance to keep internal
power dissipation down while passing up to 30A. That and I have no
room to heat sink the device. My total allocated space for relay and
FET is 26mm x 26mm x 30mm. Not that heat sinking would do any good
since my electronics are sealed inside a small glass-filled nylon
enclosure. I would need to hold the device on for 50ms to 100ms at a
time. It would get hot after a few activations. This might be
managable if ambient temperatures could be held to a reasonable level.
However, my product will be mounted under the hood of heavy trucks and
utility vehicles. Temperatures could easily approach 50C.

Then there is the issue of time. My management wants to release this
product later this spring. I need time to come to a solution and then
do adequate testing before I let it hit the street. I've passed the
point of no return with my current design approach.

7.) Fred Bloggs proposed a clever design idea. If I don't get
anywhere with my current design I'll surely try this. Heck, I might
just do it anyway for piece of mind. It performs a similar function to
the FET solution I was contemplating but will probably be cheaper to
implement.

I would have to reduce the size of my main relay in order to fit a
small secondary relay in my packaging space. I know of a few smaller
main relays that can easily carry the current levels I am working with.
 What was killing me is making and breaking the contacts at high
voltage. Fred's solution would correct that with a modest increase in
component count.

Question though, what causes the secondary relay to drop out. Doesn't
it hold itself in after the main relay (K1) drops out?

Fred Bloggs also wrote:

> The current rating of the contacts is a strong function of the force
>with which they are held closed. He most certainly has reduced this
>rating significantly by doubling the spacing- maybe halved it.
Fiddling
>with the armature travel is a dumb thing to do.

John Fields also brought this up. I tried to voice an arguement
against this above. Maybe this was just my ignorance. Mind to
ellaborate on why you say this? Just trying to learn. I went into
this project taking relay technology for granted. I never intended to
get so far in depth with it but I am very quickly finding out
otherwise.

8.) Tony Williams wrote:

>Hmm.... my solution was going to involve a timed
>MOSFET, but I don't think I'll bother now.

>Maybe add a diode from the 1-ohm to Ground though.
>This clamps any negative HV spike from the cabling
>inductance.

I tried the diode trick but found it had no effect. I believe this is
because I am not experiencing problems due to inductance. I think I'm
just over stressing the relay contacts at high voltage extremes.

9.) Tim Shoppa wrote:

>My god, the mind boggles. 28V * 28A = 784Watts. Into a resistor. >
On a truck. What the heck is this guy doing? Icemelting? Space
>heating?

Something similar to that :).

10.) Winfield Hill wrote:

>Nah. Just plop an Infineon PROFET Smart SIPMOS switch in there.
>Get raw performance, short-circuit and thermal protection, and
>an error indication, all at a low cost. For example, use a
>BTS443, BTS432, BTS442, etc.

I would love to use a solid state solution. I wanted to do this in a
bad way. Problem though. I can't keep a FET cool enough. My product
must operate up to 70C ambient. The FET would be passing up to 30A for
2.5 minutes. Even if I could use a ultra low RDSON part of say 4 mohm
it would still be dissipating 3.6W. Stack a few in parallel and I
still won't be able to remove the heat. My electronics are embedded in
a small sealed glass filled Nylon box (sealed from the elements that
is). Even if I could fit a heat sink inside my packaging space I still
don't have a way to get the hot air out and cool air in. We cannot
re-tool the housing to accomodate a heat sink at this point.

That and then there is the issue of reverse battery connection and the
internal body diode of the FET(s). I would need to add a rather large
diode in series with the load to prevent damaging reverse current flow.
 This diode would take up even more space.

To put it bluntly, I'm try to cram 20 pounds of sh__ in a 5 pound bag.

Well, that's all for now. Hopefully you all won't fall asleep reading
through all this. Sorry for dragging this on forever but I wanted to
spell out more details to clue you in on my specific situation.

Thanks a bunch

George "Gerb" Marutz II

Loading