Reed switch failure modes

[Mikko OH2HVJ]

We're seeing some failures (about 0.1%) of reed switches we use (Hamlin
59045-T). They seem so get stuck in closed position in some cases.
The switches are used to activate the device from outside the waterproof
case with a keyfob magnet.

The circuit has a 1M||100nF pulldown+filter and reed switch from there
to +2.8VDC.

I'm now wondering what is going on. Some ideas are:
-Too strong magnets used for activating the devices
-Welding of contacts (with 100nF 2.8V ?!)
-Moisture destroying something (supposed to be hermetic glass!)

Sometimes the switch can be opened by knocking it, sometimes not. The
reed switch is moulded in plastic, so opening it up typically causes enough
mechanical vibration to destroy the fault before getting there.

Any ideas what could be the culprit ?

--
mikko OH2HVJ

[Phil Hobbs]

A microscope will tell you if the contacts welded, even after decapping breaks them apart.

Cheers

Phil Hobbs

[Spehro Pefhany]

100nF at 2.8V is 0.4uJ nominal. Hamlin claims that below 1uJ welding
should not be a problem, others (such as NI) urge caution with cable
capacitance of hundreds of pF. Personally, I would definitely use a
series resistor.

I guess you could test the remanence theory by measuring the pull-in
voltage before and after exposure to a massive rare earth magnet.



--sp


--
Best regards,
Spehro Pefhany
Amazon link for AoE 3rd Edition: http://tinyurl.com/ntrpwu8

[piglet]

I am suprised your failure rate is so low! Reeds love to dry-weld. I
have experienced much worse rates with much smaller capacitors and now
routinely add series resistors - in the tens of ohms typically, enough
to limit peak current to 100mA or less.

Somewhere there is an HP Journal article which mentions they added 22
ohm series to reed contacts switching ATE backplanes because even the
10s to 100s of pF wiring capacitance would weld reeds!

piglet

[Spehro Pefhany]

If there is some double digit uH of stray series inductance it may be
enough to save the contacts, but it will ring away for a while with a
good capacitor. A series resistor ends all that nonsense.

[John S]

On 5/4/2016 5:33 AM, Spehro Pefhany wrote:
> On Wed, 04 May 2016 13:09:00 +0300, the renowned Mikko OH2HVJ
> <mikko.sy...@nospam.fi> wrote:
>
>> We're seeing some failures (about 0.1%) of reed switches we use (Hamlin
>> 59045-T). They seem so get stuck in closed position in some cases.
>> The switches are used to activate the device from outside the waterproof
>> case with a keyfob magnet.
>>
>> The circuit has a 1M||100nF pulldown+filter and reed switch from there
>> to +2.8VDC.
>>
>> I'm now wondering what is going on. Some ideas are:
>> -Too strong magnets used for activating the devices
>> -Welding of contacts (with 100nF 2.8V ?!)
>> -Moisture destroying something (supposed to be hermetic glass!)
>>
>> Sometimes the switch can be opened by knocking it, sometimes not. The
>> reed switch is moulded in plastic, so opening it up typically causes enough
>> mechanical vibration to destroy the fault before getting there.
>>
>> Any ideas what could be the culprit ?
>
> 100nF at 2.8V is 0.4uJ nominal. Hamlin claims that below 1uJ welding
> should not be a problem, others (such as NI) urge caution with cable
> capacitance of hundreds of pF. Personally, I would definitely use a
> series resistor.

Yes. Just 10 ohms should make it safe.

[John Larkin]

Reeds are terrible. They can weld from arcing, weld from just being
closed too long, make high-resistance contacts, or just plain forget
to open or close. Mag fields, like from other reed coils, can keep
them closed. And they are big, expensive, and have terrible
thermoelectrics and reed twang noise. Bad news all around.

Can you use a more conventional relay, or better yet and SSR?


--

John Larkin Highland Technology, Inc

lunatic fringe electronics

[John S]

How would you suggest he implement that in his described environment

[John Larkin]

Hall effect sensor maybe. There are halls that act like open-collector
switches, all in one chip, tiny and probably cheaper than a Hamlin
reed. Certainly more reliable.

[Tim Wescott]

Be sure to extract some "good" ones to compare, as well as some brand-
spanking new ones.

Might be repeated damage going on that's not showing up until the uh-oh
moment.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

[Jon Elson]

There was a big problem with CP Clare reed relays about 30 years ago, due to
contamination before the contacts were put in the glass. (A friend of mine
did the electron microprobe study, and found, I think, Sodium deposits,
which had no business being inside the switch.)

So, you have a 100nF cap directly across the contacts? What current do you
get when the contacts are closed? Could be tens of amps for a few us,
certainly enough to weld contacts, assuming minimal resistance in the path.
Do you actually need that HUGE a capacitor? Or, you could put a 100 Ohm
resistor in series with the contacts, so you eliminate the current pulse.

Jon

[ChesterW]

Another failure mode for your list is shock damage. The internal
contacts are tiny cantilevered beams, and it doesn't take a big hit to
eliminate the few mil spacing of the contacts.

You can do a rough test of relative sensitivity of specific parts by
recording the switching distance from a weak magnet using a caliper
setup. The switching distance will be inversely proportional to the
contact gap. If the problem is shock damage, you should see a range of
sensitivities and the most sensitive ones may be on their way to failing
closed. Of course you need repeatability of reed location inside the
plastic package, which you likely don't have.

Maybe you could replace the plastic packaged reeds with unpackaged
reeds. At least then you would be able to see the problem.

--

Best Regards,

ChesterW
+++
Dr Chester Wildey
Founder MRRA Inc.
Electronic and Optoelectronic Instruments
MRI Motion, fNIRS Brain Scanners, Counterfeit and Covert Marker Detection
Fort Worth, Texas, USA
www.mrrainc.com
wildey at mrrainc dot com

[Adrian Jansen]

We went from reed switches to Hall effect devices for shaft speed
measurement ( low speeds, 100 - 5000 rpm ) about 30 years ago, for
exactly the same reasons. But its true that a few ohms in series with a
shunt capacitor can vastly improve the reed life.

The one advantage of a reed is that it does not draw power in the off
state, so using it as an on-off trigger is much more power efficient
than a Hall device for battery operated systems.

[ChesterW]

Some Hall effect switches operate on low single digit uA.

ChesterW

[srober...@gmail.com]

The Bell System always ran a tiny DC sealing current to clean the contacts, this was one of the reasons why. They invented the reed, they should know why. You might want to read up on them in BSTj.

STEVE

[dagmarg...@yahoo.com]

Mikko, I've seen a paralleled 100nF edm a hole in a snap/dome switch!
I shudder to think what it would do to a reed.

Count me in the series resistor camp , definitely.

Cheers,
James Arthur

[Kevin McMurtrie]

In article <m2d1p2r...@nospam.fi>,

Welding from the capacitor, as others have said, plus the slow removal
of the field causing nearby metals to remain magnetized. Together it's
getting stuck at a much lower threshold than you'd think.

--
I will not see posts from astraweb, theremailer, dizum, or google
because they host Usenet flooders.

[Jon Elson]

Adrian Jansen wrote:


> We went from reed switches to Hall effect devices for shaft speed
> measurement ( low speeds, 100 - 5000 rpm ) about 30 years ago, for
> exactly the same reasons. But its true that a few ohms in series with a
> shunt capacitor can vastly improve the reed life.

This is a bad case, as the reed gets cycled thousands of times per minute,
in most cases, and can use up the rated life in a week or two. So, there, a
Hall sensor makes very good sense.

Jon

[Mikko OH2HVJ]

Phil Hobbs <pcdh...@gmail.com> writes:

> A microscope will tell you if the contacts welded, even after decapping breaks them apart.

True, forgot about using more than 20X Mantis!

--
mikko

[Mikko OH2HVJ]

piglet <erichp...@hotmail.com> writes:

> On 04/05/2016 11:09, Mikko OH2HVJ wrote:
>> We're seeing some failures (about 0.1%) of reed switches we use (Hamlin
>> 59045-T). They seem so get stuck in closed position in some cases.

> I am suprised your failure rate is so low! Reeds love to dry-weld. I
> have experienced much worse rates with much smaller capacitors and now
> routinely add series resistors - in the tens of ohms typically, enough
> to limit peak current to 100mA or less.

Sounds like your experience matches our case. I measured over 5A peak
currents through the switch. Probably the original designer thought that
a tiny 100nF cap could not do any harm. Neither would have I.

Next PCB revision will have series resistor, I'll have some statistical
data by the end of the year and will make a calendar note to report back
here!

--
mikko OH2HVJ

[Mikko OH2HVJ]

I also thought of this, but I think in this case the failure mode would
be open circuit. The datasheet does not specify minimum switch current
nor contact plating material.

Some Hamlin reed tube data sheets stated the contact material, but not
the moulded ones.

--
Mikko OH2HVJ

[Mikko OH2HVJ]

Unfortunately our power budget for this part is way less than a uA.
We could use a hall with 10ms on / 500ms off cycle, but currently we're
not waking up that often.

--
mikko

[Mikko OH2HVJ]

Jon Elson <jme...@wustl.edu> writes:

> So, you have a 100nF cap directly across the contacts? What current do you
> get when the contacts are closed? Could be tens of amps for a few us,
> certainly enough to weld contacts, assuming minimal resistance in the path.
> Do you actually need that HUGE a capacitor? Or, you could put a 100 Ohm
> resistor in series with the contacts, so you eliminate the current pulse.

The peak current is over 5A measured by adding a short piece of wire for
P6042 probe. That might add some inductance and the probe BW is only 50MHz, so
the peak might be a bit higher.

We have a PCB revision scheduled for August and that one will definetely
include the series resistor. In the meanwhile I'm thinking if there'd be
any problems going to a large 10nF or medium 1nF cap.

About caps. Where else do you this kind of ranges. The same PCB may have
0.4 pF for RF matching and 3.3F for power supply. Thats 1:10000000000000!
--
mikko

[Mikko OH2HVJ]

dagmarg...@yahoo.com writes:

> Mikko, I've seen a paralleled 100nF edm a hole in a snap/dome switch!
> I shudder to think what it would do to a reed.
>
> Count me in the series resistor camp , definitely.

What a disgusting mental image. Series resistor it is.

Thank you everybody for comments, this was again SED at its' best.
I'll report back when we have experience of about 10k units with series
resistor.


--
mikko

[M Philbrook]

In article <m2d1p2r...@nospam.fi>, mikko.sy...@nospam.fi
says...

I don't fully understand your layout as you describe it, "||" ?
this tells me you put the cap and r in parallel ?

The snubber circuit needs to be a series circuit across the contact
points.
In your case a ceramic cap only across the contact terminals should do.

Jamie

[Mikko OH2HVJ]

M Philbrook <jamie_...@charter.net> writes:

>> The circuit has a 1M||100nF pulldown+filter and reed switch from there
>> to +2.8VDC.

...

> I don't fully understand your layout as you describe it, "||" ?
> this tells me you put the cap and r in parallel ?

Yes, the R and C are parallel for EMC filtering and debouncing. The high
current path closes through the power supply planes and bypass caps.

--
mikko

[dca...@krl.org]

On Wednesday, May 4, 2016 at 6:23:59 PM UTC-4, ChesterW wrote:

>
> Another failure mode for your list is shock damage. The internal
> contacts are tiny cantilevered beams, and it doesn't take a big hit to
> eliminate the few mil spacing of the contacts.
>

> ChesterW
> +++
> Dr Chester Wildey
> Founder MRRA Inc.
> Electronic and Optoelectronic Instruments
> MRI Motion, fNIRS Brain Scanners, Counterfeit and Covert Marker Detection
> Fort Worth, Texas, USA
> www.mrrainc.com
> wildey at mrrainc dot com

Yes shock can do it. I have a LED flashlight that is water proof to about 30 feet. The concept is good. It uses a read switch and a moveable magnet to turn it on. So no possible leakage path . But it fell off the night stand and then would not work. I got it working, but have no faith in it as a rugged flashlight.

Dan