Today's Message Index:
----------------------
1. 05:27 AM - Re: Re: Brad's reply on P-lead switch functionality (Harley)
2. 05:27 AM - Re: Re: Brad's reply on P-lead switch functionality (David & Elaine Lamphere)
3. 06:38 AM - Toggle Switches with Fast-On Tabs (Robert L. Nuckolls, III)
4. 07:02 AM - Re: Re: Brad's reply on P-lead switch functionality (Robert L. Nuckolls, III)
5. 09:14 AM - Vern's crimp tool performance . . . (Robert L. Nuckolls, III)
6. 10:05 AM - Re: Vern's crimp tool performance . . . (Vernon Little)
7. 10:47 AM - Re: Toggle Switches with Fast-On Tabs (Vernon Little)
8. 11:23 AM - Re: E-Mag P-Mag Safety info (Speedy11@aol.com)
9. 12:07 PM - Re: Re: Brad's reply on P-lead switch functionality (S. Ramirez)
10. 02:39 PM - Re: E-Mag P-Mag Safety update . . . (Robert L. Nuckolls, III)
Message 1
| INDEX | Back to Main INDEX |
| NEXT | Skip to NEXT Message |
| LIST | Reply to LIST Regarding this Message |
| SENDER | Reply to SENDER Regarding this Message |
|
| Subject: | Re: re: Brad's reply on P-lead switch functionality |
Dale Rogers wrote:
>
> Robert L. Nuckolls, III wrote:
>> <nuckolls.bob@cox.net>
>>
>> Received a reply from Brad on functionality of the p-lead switch
>> in Emag products:
>>
>> "Grounding the p-lead 1) sends a status signal to the processor
>> (telling it
>> to stop firing), and 2) disables the driver chips (so they can't fire)."
>
> Umm, that ~sounds~ good - but what does it mean? In the 30+
> years that I've been working with computer hardware, I've
> never run across that expression for stopping the processor.
>
> In situations were a runaway process could result in damage,
> the normal method for halting a processor isn't via a status
> semaphore, but by halting the CPU clock pulse stream -
> either via grounding the output of the clock, or removing
> power from the clock crystal. With no clock pulses, the CPU
> cannot execute instructions, period. It's the only way to be
> completely certain that the CPU will in fact stop.
>
> Dale R.
Morning, Dale...
The way I read it, is that the P-lead "signal" tells the cpu to STOP
FIRING (the spark plugs), not to stop. I assume that the processor
continues running...the P-lead "signal" also disables the driver chips,
which I assume are what actually direct the higher voltage to fire the
spark plugs. These drivers are the devices that actually stop functioning.
So, the processor keeps running, but is told to stop firing the driver
chips, and in case it doesn't, the driver chips are also disabled as a
back up, I would assume.
Looking back on what I just wrote, a lot of assumptions, huh? <G>
Harley Dixon
Message 2
| INDEX | Back to Main INDEX |
| PREVIOUS | Skip to PREVIOUS Message |
| NEXT | Skip to NEXT Message |
| LIST | Reply to LIST Regarding this Message |
| SENDER | Reply to SENDER Regarding this Message |
|
| Subject: | Re: re: Brad's reply on P-lead switch functionality |
Why would you intepret that statement to mean "stop the processor" ??
The way I read that statement "status signal sent to the processor (telling
it to stop firing)" meant the program that was running would no longer
execute the "ignition fire" routine.. It could be done with a simple "if"
statement, etc,... depends on the programming language used...
Made sense to me.
Dave L.
----- Original Message -----
From: "Dale Rogers" <dale.r@cox.net>
Sent: Thursday, September 18, 2008 10:44 PM
Subject: Re: AeroElectric-List: re: Brad's reply on P-lead switch
functionality
>
> Robert L. Nuckolls, III wrote:
>> <nuckolls.bob@cox.net>
>>
>> Received a reply from Brad on functionality of the p-lead switch
>> in Emag products:
>>
>> "Grounding the p-lead 1) sends a status signal to the processor (telling
>> it
>> to stop firing), and 2) disables the driver chips (so they can't fire)."
>
> Umm, that ~sounds~ good - but what does it mean? In the 30+
> years that I've been working with computer hardware, I've
> never run across that expression for stopping the processor.
>
> In situations were a runaway process could result in damage,
> the normal method for halting a processor isn't via a status
> semaphore, but by halting the CPU clock pulse stream -
> either via grounding the output of the clock, or removing
> power from the clock crystal. With no clock pulses, the CPU
> cannot execute instructions, period. It's the only way to be
> completely certain that the CPU will in fact stop.
>
> Dale R.
>
Message 3
| INDEX | Back to Main INDEX |
| PREVIOUS | Skip to PREVIOUS Message |
| NEXT | Skip to NEXT Message |
| LIST | Reply to LIST Regarding this Message |
| SENDER | Reply to SENDER Regarding this Message |
|
| Subject: | Toggle Switches with Fast-On Tabs |
At 08:18 PM 9/18/2008 -0400, you wrote:
><Tom@costanzaandassociates.com>
>
>Vern,
>
>Don't you think that if it was an application problem that we'd be hearing
>from people on a few orders of magnitude more than we have? There must be
>several hundred people using these switches in these circuits. No??
More like thousands. The sum total of years those switches were sold
first by AEC and then B&C is about 15.
------------------------------------
<ftyoder@yoderbuilt.com>
Is their a manufacturing date or any number on the switches that would let
you identify a bad batch of switches? Sounds like some switches come with
loose rivets and some don't have loose rivets. Is that correct?
The manufacture should be able to tell you if loose rivets are acceptable.
Loose rivets are not acceptable. If you refer to the sketch
at:
http://www.aeroelectric.com/Pictures/Switches/Toggle_Switch_with_Fast-On_Tabs.jpg
The hollow rivets at (3) and (8) are what hold the switch internal
parts together -AND- provide a conductive path from part to part.
As soon as the rivet's retaining force goes down, resistance
goes up, heating goes up, and the device starts down the path
to failure.
---------------------------------------
I understand that your switches were "properly" supported. Just wondering
if the wire "support" needs to be more substantial than usual if there are
large gauge wires involved, such as would be expected for landing light
circuits? My thinking is that the larger the wire, the more momentum it has
when vibrating due to it's mass.
Secondly, maybe I've missed something here, but concerning the "loose"
rivets. It is thought that the rivets are loose because of overheating, or
vibration (with a heavy wire attached), or poor manufacturing process or
what?
Bevan, you may be on to something here . . . I'm embarrassed
to have overlooked it up to now . . .
Over the past week, I've received 5 corpses of dearly departed
switches. All had loose rivets on the current carrying tabs. The
OFF side tab was solidly retained by their rivets. Here's an
exemplar photo set:
These two pictures show the moving ON and OFF-side contacts.
Noticeably free of signs of electrical stress.
http://www.aeroelectric.com/Pictures/Switches/Carling_Failures/BRA_01.jpg
http://www.aeroelectric.com/Pictures/Switches/Carling_Failures/BRA_02.jpg
This picture shows the "saddle sores" that one would expect
in a switch where the teeter-totter shaft was properly undersized . . .
(Ref
http://www.aeroelectric.com/articles/Anatomy_of_a_Switch_Failure/Anatomy_of_a_Switch_Failure.html)
to sit in the saddle.
http://www.aeroelectric.com/Pictures/Switches/Carling_Failures/BRA_03.jpg
---------------
The spring-rate of the coil spring inside the toggle was checked
against a new switch and found to be the same within measurement
tolerances. It takes right at a pound of force to push the plastic
actuator post flush with the end of the hollow toggle shaft . . .
http://www.aeroelectric.com/Pictures/Switches/Carling_Failures/BRA_06.jpg
---------------
This switch has loose rivets at the ON-side tab (left) and
the center tab. OFF-side rivet was tight. Note signs of heating
induced corrosion on the left side rivet.
http://www.aeroelectric.com/Pictures/Switches/Carling_Failures/BRA_04.jpg
http://www.aeroelectric.com/Pictures/Switches/Carling_Failures/BRA_09.jpg
----------------
Here's a familiar picture . . .
http://www.aeroelectric.com/Pictures/Switches/Carling_Failures/BRA_08.JPG
We've seen this on-side vs. off-side bending of the teeter-
totter before. Notice the darker copper color on the right. This
was the ON-side contact that was running warmer but the OFF-side
was more severely deformed from flat.
----------------------
Here's a really cool picture . . .
http://www.aeroelectric.com/Pictures/Switches/Carling_Failures/BRA_05.jpg
This switch was obviously powered at the center terminal. The
teeter-totter was sufficiently distorted to cause teasing arc
marks on BOTH the on and off extremes. I say "teasing" because
this had to be going on for some time to make such strong marks
before the failure progressed to the point where it was popping
circuit protection.
-------------------------------
This picture shows perfectly good On and Off-side contacts
in addition to the bright areas in the bottom of the
saddle pocket. The ON-side contact is on the right with
barely detectable signs of elevated temperature operation.
http://www.aeroelectric.com/Pictures/Switches/Carling_Failures/BRA_10.jpg
-----------------------------------
This picture is shows the teeter-totter as removed and before
the lubrication was wiped off.
http://www.aeroelectric.com/Pictures/Switches/Carling_Failures/BRA_11.jpg
----------------------------------
This picture shows the loose rivet at the ON-side tab
but the tab itself is not showing signs of strong
heating.
http://www.aeroelectric.com/Pictures/Switches/Carling_Failures/BRA_12.jpg
-----------------------------------
Here's the OFF-side tab. Tight rivet, bright clean
brass.
http://www.aeroelectric.com/Pictures/Switches/Carling_Failures/BRA_14.jpg
-----------------------------------
The common feature of all the carcasses I received was
distortion of the teeter-totter from flat to a bowed
condition of various degrees. The teeter-totter metal
had lost temper and bent with heating and operational
hammering being the predominant stresses.
Stationary and moving contacts of all failed switches
showed only slight if any visible effects of heating.
All switches had loose rivets at the connection tabs.
(note that my sketched cross-section has the center
rivet (8) reversed. The head-formed-on-assembly is on
the inside of the switch, not the outside).
When dissecting a chain of failure events, one tries
to deduce the first event . . . it's not unlike looking
for the point of origin in a burned building. In this
case, we're attempting to deduce the weakest (hence
most likely to be overstressed) feature and then
see if the damage patterns radiate out from that
feature.
In the case of the analysis published earlier this
year, there were striking patterns laid down on the
teeter-totter pivot that could easily be interpreted
as a point of origin for chain of failures. However,
Bevan's question about mechanical stresses to the
tabs due to wire weight and bundling was an "eureka
moment" of sorts.
Consider the cross section sketch and photos of
the riveted tabs and know that the weakest mechanical
feature of these switches is the point where the hollow
rivets are formed over to achieve retention forces
that hold the parts together PLUS a gas-tight connection
between the rivet and it's companion pieces and parts.
While the failed switches were mostly concentrated
in the ship's higher current systems, we've seen
failures in the low current systems too (battery
master). Nonetheless, Vern has experienced a rash
of failures that spanned the full range of current
carrying tasks.
Consider the effects of a wire (heavier in the
landing light and strobe systems) hanging off the
back of the switch. Consider the effects of a wire
bundle that is supported by the switch terminals
with bundle-to-switch pigtails that are relatively
short. There's a strong moment arm from the end of
the tab that allows vibration of attached mass to
put tension on the rolled over edges of the
driven-head of the rivets.
This makes more sense as a proximate root cause of
the constellation of failures we've studied. Once
gas-tightness of the rivet head is compromised,
then corrosion goes up, resistance goes up, heating
goes up, corrosion accelerates, etc. etc.
I think I recall writing some words to the effect
that pigtails that come off the backs of switches
should have some substantial length before they
drop into a wire bundle. The goal is two-fold:
(1) service loop length to allow removal and
replacement of the switch without disturbing other
switches or the bundle and (2) a stress reliever
that prevents mass of the wire bundle from adding
to the vibrational stresses on the switch's tabs.
I think I suggested a 2" service loop that offered
free-slack in the leadwire between bundle and
switch terminal.
In the $high$ switches, the effects of mis-installation
are reduced by the manner in which wire attach terminals
are retained in the switch housing. See:
http://www.aeroelectric.com/Pictures/Switches/Toggle_Switch_with_Mold-Captured_Terminals.jpg
Vern, take a look at your installation for the purpose
of assessing probability that vibration in the
wire bundle mass is being conducted to the rivet
joints due to a short-coupled installation of the
attach wires.
If this new hypothesis proves plausible, then it
may well explain the rash of failures noted in
a product with an exceedingly rich history in
the marketplace. This has been nagging at me since
this thread began. Lots of mud was thrown against
the wall about AC vs. DC, failure to observe ratings,
high inrush due to nature of the loads, etc.
Yet we were still dealing with a legacy product
manufactured in the millions and used with
success in aircraft since the 60s. There
had to be a common thread that tied all of what
we observed together in a rational assemblage
of simple-ideas.
Current working hypothesis: Loss of gas-tight
integrity at the rolled head of the rivets anchors
a chain of failures that manifests itself with
signs of heating, distortion of the teeter-totter,
internal shorting of teeter-totter to the frame and
loss of continuity through the switch.
I think we may be getting closer . . .
Bob . . .
Message 4
| INDEX | Back to Main INDEX |
| PREVIOUS | Skip to PREVIOUS Message |
| NEXT | Skip to NEXT Message |
| LIST | Reply to LIST Regarding this Message |
| SENDER | Reply to SENDER Regarding this Message |
|
| Subject: | Re: re: Brad's reply on P-lead switch functionality |
At 07:44 PM 9/18/2008 -0700, you wrote:
>
>Robert L. Nuckolls, III wrote:
>><nuckolls.bob@cox.net>
>>
>>Received a reply from Brad on functionality of the p-lead switch
>>in Emag products:
>>
>>"Grounding the p-lead 1) sends a status signal to the processor (telling it
>>to stop firing), and 2) disables the driver chips (so they can't fire)."
>
>Umm, that ~sounds~ good - but what does it mean? In the 30+
>years that I've been working with computer hardware, I've
>never run across that expression for stopping the processor.
>
>In situations were a runaway process could result in damage,
>the normal method for halting a processor isn't via a status
>semaphore, but by halting the CPU clock pulse stream -
>either via grounding the output of the clock, or removing
>power from the clock crystal. With no clock pulses, the CPU
>cannot execute instructions, period. It's the only way to be
>completely certain that the CPU will in fact stop.
Brad's description of the p-lead signal functionality
may be interpreted as follows:
First, it sets a discrete input to the processor that
causes the software routines to stop triggering the
coil for the purpose of generating a spark. The
processor doesn't "halt in place" it's expected to
recognize an operational command and to honor that
command until it goes away.
Second, the p-lead is tied to the 'drivers' between
logic level (processor) and power level (spark coil)
such that no communication between them is possible.
I.e., even if the processor has wandered off into the
weeds, the physical connection between logic and
output is broken.
I do this in all of my processor or logic based
smart actuator designs. There's a "logic world" that
runs at 5 volts in itty-bitty chunks (read fragile)
of silicon and provide the "smart" side of the
actuator's design. Then there's the power side . . .
usually a brushless DC motor with ratings from 0.1
to several horsepower . . . it runs on 28VDC at
lots of amps.
To get "smarts" to communicate with "power" you need
a combination of level shifters and drivers that
translate from the 5 volt milliamps world to the
28 volt amps world. I always bring the operating
power for these drivers or level shifters out to
interface with the ship's flight management systems.
Since these systems are already certified with level
B or level A software, I let THEM do failure monitoring
on my product while providing with a brick-wall-shutdown
for causing my operation to cease.
This feature relieves my software and hardware from
both the rigors of high-risk software certification
AND the need to do failure monitoring. I can push
those tasks off onto hardware and software that is
already taking on that responsibility for other
systems . . . so adding my system to it is not
burdensome. But it works only if you have this
brick-wall-shutdown feature not unlike that which
Emag has incorporated into their product.
Bob . . .
Message 5
| INDEX | Back to Main INDEX |
| PREVIOUS | Skip to PREVIOUS Message |
| NEXT | Skip to NEXT Message |
| LIST | Reply to LIST Regarding this Message |
| SENDER | Reply to SENDER Regarding this Message |
|
| Subject: | Vern's crimp tool performance . . . |
Vern,
I was able to run a resistance measurement on your newly
installed terminals and got readings on the order of 200
micro-ohms . . . right in step with the values I get from
a PIDG terminal installed with an AMP T-head tool.
I was unable to measure the crimps on the terminals off the
failed switch . . . not enough lead-length to get a grip
on. But I did cross section the crimp and got the following
photomicrograph . . .
http://aeroelectric.com/Pictures/Misc/VL_Crimp_Tool_1.jpg
The terminal from the failed switch exhibits no voids in
the wire grip. Of course, you used this tool on other terminals
NOT attached to switches and you've not seen a failure there
so the results of my little look-see were predictable.
Bob . . .
----------------------------------------)
( . . . a long habit of not thinking )
( a thing wrong, gives it a superficial )
( appearance of being right . . . )
( )
( -Thomas Paine 1776- )
----------------------------------------
Message 6
| INDEX | Back to Main INDEX |
| PREVIOUS | Skip to PREVIOUS Message |
| NEXT | Skip to NEXT Message |
| LIST | Reply to LIST Regarding this Message |
| SENDER | Reply to SENDER Regarding this Message |
|
| Subject: | Vern's crimp tool performance . . . |
Thanks, Bob. I think you've safely eliminated bad crimps as a contributing
factor. Much apprectiated.
Vern
> -----Original Message-----
> From: owner-aeroelectric-list-server@matronics.com
> [mailto:owner-aeroelectric-list-server@matronics.com] On
> Behalf Of Robert L. Nuckolls, III
> Sent: September 19, 2008 9:12 AM
> To: aeroelectric-list@matronics.com
> Subject: AeroElectric-List: Vern's crimp tool performance . . .
>
>
>
> --> <nuckolls.bob@cox.net>
>
> Vern,
>
> I was able to run a resistance measurement on your newly
> installed terminals and got readings on the order of 200
> micro-ohms . . . right in step with the values I get from a
> PIDG terminal installed with an AMP T-head tool.
>
> I was unable to measure the crimps on the terminals off the
> failed switch . . . not enough lead-length to get a grip on.
> But I did cross section the crimp and got the following
> photomicrograph . . .
>
> http://aeroelectric.com/Pictures/Misc/VL_Crimp_Tool_1.jpg
>
> The terminal from the failed switch exhibits no voids in
> the wire grip. Of course, you used this tool on other
> terminals NOT attached to switches and you've not seen a
> failure there so the results of my little look-see were predictable.
>
>
> Bob . . .
>
> ----------------------------------------)
> ( . . . a long habit of not thinking )
> ( a thing wrong, gives it a superficial )
> ( appearance of being right . . . )
> ( )
> ( -Thomas Paine 1776- )
> ----------------------------------------
>
>
>
>
>
>
>
Message 7
| INDEX | Back to Main INDEX |
| PREVIOUS | Skip to PREVIOUS Message |
| NEXT | Skip to NEXT Message |
| LIST | Reply to LIST Regarding this Message |
| SENDER | Reply to SENDER Regarding this Message |
|
| Subject: | Toggle Switches with Fast-On Tabs |
Bob, thanks for your detailed response and analysis.
I'll take some photos of the wiring next week to see if I have overlooked a
potential source of vibration and unsupported wiring. I'll also see if I
get photos from my friend's panel showing the same.
It may be that vibration per se is not the issue, but in combination with
suspect rivets and large currents, we have the recipe for problems. I do
know that I've received new Carling switches with suspect rivets. One
suggestion that I got was to solder the tabs to the rivets, but I am worried
that may screw up the temper of the materials. Of note: we have at least
one non-Carling switch failure as well.
One thing that I didn't pay too much attention to was stress on the
terminals from the wires. Perhaps the terminals are putting torque or
bending preloads on the tabs that are stressing the tab/rivet joint. In
fact, additional post-installation tie-wraps in the bundles tend to pull
wires off of their natural alignment and cause these stresses.
When I'm crawling under the panel with my camera, I will check the wire
grooming.
My bias on this (looking at my photos and my friend's) is that wire grooming
is at most a secondary contributing factor. Given the wide variation in
wiring techniques in the OBAM community, if this was the cause, we'd have
serious trouble.
Maybe I'll blame it on mud-dauber wasps, that's popular!
V
> -----Original Message-----
> From: owner-aeroelectric-list-server@matronics.com
> [mailto:owner-aeroelectric-list-server@matronics.com] On
> Behalf Of Robert L. Nuckolls, III
> Sent: September 19, 2008 6:38 AM
> To: aeroelectric-list@matronics.com
> Subject: AeroElectric-List: Toggle Switches with Fast-On Tabs
>
>
>
> --> <nuckolls.bob@cox.net>
>
> At 08:18 PM 9/18/2008 -0400, you wrote:
> ><Tom@costanzaandassociates.com>
> >
> >Vern,
> >
> >Don't you think that if it was an application problem that we'd be
> >hearing from people on a few orders of magnitude more than we have?
> >There must be several hundred people using these switches in these
> >circuits. No??
>
> More like thousands. The sum total of years those switches
> were sold
> first by AEC and then B&C is about 15.
> ------------------------------------
> <ftyoder@yoderbuilt.com>
>
> Is their a manufacturing date or any number on the switches
> that would let
> you identify a bad batch of switches? Sounds like some
> switches come with
> loose rivets and some don't have loose rivets. Is that correct?
>
> The manufacture should be able to tell you if loose rivets
> are acceptable.
>
> Loose rivets are not acceptable. If you refer to the sketch
> at:
>
> http://www.aeroelectric.com/Pictures/Switches/Toggle_Switch_wi
> th_Fast-On_Tabs.jpg
>
> The hollow rivets at (3) and (8) are what hold the switch internal
> parts together -AND- provide a conductive path from part to part.
> As soon as the rivet's retaining force goes down, resistance
> goes up, heating goes up, and the device starts down the path
> to failure.
> ---------------------------------------
> <fvalarm@rapidnet.net>
>
> I understand that your switches were "properly" supported.
> Just wondering
> if the wire "support" needs to be more substantial than usual
> if there are
> large gauge wires involved, such as would be expected for
> landing light
> circuits? My thinking is that the larger the wire, the more
> momentum it has
> when vibrating due to it's mass.
>
> Secondly, maybe I've missed something here, but concerning the "loose"
> rivets. It is thought that the rivets are loose because of
> overheating, or
> vibration (with a heavy wire attached), or poor manufacturing
> process or
> what?
>
> Bevan, you may be on to something here . . . I'm embarrassed
> to have overlooked it up to now . . .
>
> Over the past week, I've received 5 corpses of dearly departed
> switches. All had loose rivets on the current carrying tabs. The
> OFF side tab was solidly retained by their rivets. Here's an
> exemplar photo set:
>
> These two pictures show the moving ON and OFF-side contacts.
> Noticeably free of signs of electrical stress.
>
> http://www.aeroelectric.com/Pictures/Switches/Carling_Failures
> /BRA_01.jpg
>
> http://www.aeroelectric.com/Pictures/Switches/Carling_Failures
> /BRA_02.jpg
>
> This picture shows the "saddle sores" that one would expect
> in a switch where the teeter-totter shaft was properly
> undersized . . .
>
> (Ref
> http://www.aeroelectric.com/articles/Anatomy_of_a_Switch_Failu
> re/Anatomy_of_a_Switch_Failure.html)
>
> to sit in the saddle.
>
> http://www.aeroelectric.com/Pictures/Switches/Carling_Failures
> /BRA_03.jpg
> ---------------
>
> The spring-rate of the coil spring inside the toggle was checked
> against a new switch and found to be the same within measurement
> tolerances. It takes right at a pound of force to push the plastic
> actuator post flush with the end of the hollow toggle shaft . . .
>
> http://www.aeroelectric.com/Pictures/Switches/Carling_Failures
> /BRA_06.jpg
>
> ---------------
>
> This switch has loose rivets at the ON-side tab (left) and
> the center tab. OFF-side rivet was tight. Note signs of heating
> induced corrosion on the left side rivet.
>
> http://www.aeroelectric.com/Pictures/Switches/Carling_Failures
> /BRA_04.jpg
>
> http://www.aeroelectric.com/Pictures/Switches/Carling_Failures
> /BRA_09.jpg
>
> ----------------
>
> Here's a familiar picture . . .
>
> http://www.aeroelectric.com/Pictures/Switches/Carling_Failures
> /BRA_08.JPG
>
> We've seen this on-side vs. off-side bending of the teeter-
> totter before. Notice the darker copper color on the right. This
> was the ON-side contact that was running warmer but the OFF-side
> was more severely deformed from flat.
>
> ----------------------
>
> Here's a really cool picture . . .
>
> http://www.aeroelectric.com/Pictures/Switches/Carling_Failures
> /BRA_05.jpg
>
> This switch was obviously powered at the center terminal. The
> teeter-totter was sufficiently distorted to cause teasing arc
> marks on BOTH the on and off extremes. I say "teasing" because
> this had to be going on for some time to make such strong marks
> before the failure progressed to the point where it was popping
> circuit protection.
>
> -------------------------------
>
> This picture shows perfectly good On and Off-side contacts
> in addition to the bright areas in the bottom of the
> saddle pocket. The ON-side contact is on the right with
> barely detectable signs of elevated temperature operation.
>
> http://www.aeroelectric.com/Pictures/Switches/Carling_Failures
> /BRA_10.jpg
>
>
> -----------------------------------
>
> This picture is shows the teeter-totter as removed and before
> the lubrication was wiped off.
>
> http://www.aeroelectric.com/Pictures/Switches/Carling_Failures
> /BRA_11.jpg
>
>
> ----------------------------------
>
> This picture shows the loose rivet at the ON-side tab
> but the tab itself is not showing signs of strong
> heating.
>
> http://www.aeroelectric.com/Pictures/Switches/Carling_Failures
> /BRA_12.jpg
>
> -----------------------------------
>
> Here's the OFF-side tab. Tight rivet, bright clean
> brass.
>
> http://www.aeroelectric.com/Pictures/Switches/Carling_Failures
> /BRA_14.jpg
>
> -----------------------------------
>
> The common feature of all the carcasses I received was
> distortion of the teeter-totter from flat to a bowed
> condition of various degrees. The teeter-totter metal
> had lost temper and bent with heating and operational
> hammering being the predominant stresses.
>
> Stationary and moving contacts of all failed switches
> showed only slight if any visible effects of heating.
>
> All switches had loose rivets at the connection tabs.
> (note that my sketched cross-section has the center
> rivet (8) reversed. The head-formed-on-assembly is on
> the inside of the switch, not the outside).
>
> When dissecting a chain of failure events, one tries
> to deduce the first event . . . it's not unlike looking
> for the point of origin in a burned building. In this
> case, we're attempting to deduce the weakest (hence
> most likely to be overstressed) feature and then
> see if the damage patterns radiate out from that
> feature.
>
> In the case of the analysis published earlier this
> year, there were striking patterns laid down on the
> teeter-totter pivot that could easily be interpreted
> as a point of origin for chain of failures. However,
> Bevan's question about mechanical stresses to the
> tabs due to wire weight and bundling was an "eureka
> moment" of sorts.
>
> Consider the cross section sketch and photos of
> the riveted tabs and know that the weakest mechanical
> feature of these switches is the point where the hollow
> rivets are formed over to achieve retention forces
> that hold the parts together PLUS a gas-tight connection
> between the rivet and it's companion pieces and parts.
>
> While the failed switches were mostly concentrated
> in the ship's higher current systems, we've seen
> failures in the low current systems too (battery
> master). Nonetheless, Vern has experienced a rash
> of failures that spanned the full range of current
> carrying tasks.
>
> Consider the effects of a wire (heavier in the
> landing light and strobe systems) hanging off the
> back of the switch. Consider the effects of a wire
> bundle that is supported by the switch terminals
> with bundle-to-switch pigtails that are relatively
> short. There's a strong moment arm from the end of
> the tab that allows vibration of attached mass to
> put tension on the rolled over edges of the
> driven-head of the rivets.
>
> This makes more sense as a proximate root cause of
> the constellation of failures we've studied. Once
> gas-tightness of the rivet head is compromised,
> then corrosion goes up, resistance goes up, heating
> goes up, corrosion accelerates, etc. etc.
>
> I think I recall writing some words to the effect
> that pigtails that come off the backs of switches
> should have some substantial length before they
> drop into a wire bundle. The goal is two-fold:
> (1) service loop length to allow removal and
> replacement of the switch without disturbing other
> switches or the bundle and (2) a stress reliever
> that prevents mass of the wire bundle from adding
> to the vibrational stresses on the switch's tabs.
> I think I suggested a 2" service loop that offered
> free-slack in the leadwire between bundle and
> switch terminal.
>
> In the $high$ switches, the effects of mis-installation
> are reduced by the manner in which wire attach terminals
> are retained in the switch housing. See:
>
> http://www.aeroelectric.com/Pictures/Switches/Toggle_Switch_wi
> th_Mold-Captured_Terminals.jpg
>
> Vern, take a look at your installation for the purpose
> of assessing probability that vibration in the
> wire bundle mass is being conducted to the rivet
> joints due to a short-coupled installation of the
> attach wires.
>
> If this new hypothesis proves plausible, then it
> may well explain the rash of failures noted in
> a product with an exceedingly rich history in
> the marketplace. This has been nagging at me since
> this thread began. Lots of mud was thrown against
> the wall about AC vs. DC, failure to observe ratings,
> high inrush due to nature of the loads, etc.
>
> Yet we were still dealing with a legacy product
> manufactured in the millions and used with
> success in aircraft since the 60s. There
> had to be a common thread that tied all of what
> we observed together in a rational assemblage
> of simple-ideas.
>
> Current working hypothesis: Loss of gas-tight
> integrity at the rolled head of the rivets anchors
> a chain of failures that manifests itself with
> signs of heating, distortion of the teeter-totter,
> internal shorting of teeter-totter to the frame and
> loss of continuity through the switch.
>
> I think we may be getting closer . . .
>
> Bob . . .
>
>
>
>
>
>
>
>
Message 8
| INDEX | Back to Main INDEX |
| PREVIOUS | Skip to PREVIOUS Message |
| NEXT | Skip to NEXT Message |
| LIST | Reply to LIST Regarding this Message |
| SENDER | Reply to SENDER Regarding this Message |
|
| Subject: | Re: E-Mag P-Mag Safety info |
Bob,
Thanks for acting as our expert go-between with EMagAir. I've been tempted
to call them, but I'm sure they would prefer to avoid another phone call or
email of possible. Your contact with them answered questions for all of us
while simultaneously reducing the duplicative responses for Brad.
Stan Sutterfield
Call Emag and if push comes to shove, run a Magneto/Emag combination
for awhile. Bottom line is that the sky is not falling.
**************Looking for simple solutions to your real-life financial
challenges? Check out WalletPop for the latest news and information, tips and
calculators. (http://www.walletpop.com/?NCID=emlcntuswall00000001)
Message 9
| INDEX | Back to Main INDEX |
| PREVIOUS | Skip to PREVIOUS Message |
| NEXT | Skip to NEXT Message |
| LIST | Reply to LIST Regarding this Message |
| SENDER | Reply to SENDER Regarding this Message |
|
| Subject: | re: Brad's reply on P-lead switch functionality |
-----Original Message-----
>> <nuckolls.bob@cox.net>
>>
>> Received a reply from Brad on functionality of the p-lead switch
>> in Emag products:
>>
>> "Grounding the p-lead 1) sends a status signal to the processor
>> (telling it
>> to stop firing), and 2) disables the driver chips (so they can't fire)."
> Umm, that ~sounds~ good - but what does it mean? In the 30+
> years that I've been working with computer hardware, I've
> never run across that expression for stopping the processor.
> In situations were a runaway process could result in damage,
> the normal method for halting a processor isn't via a status
> semaphore, but by halting the CPU clock pulse stream -
> either via grounding the output of the clock, or removing
> power from the clock crystal. With no clock pulses, the CPU
> cannot execute instructions, period. It's the only way to be
> completely certain that the CPU will in fact stop.
> Dale R.
Dale,
By now you've seen Bob's reply about how the P-lead is a discrete input to
the processor to cause the software routines to stop an output from
triggering the actual spark drivers, AND the P-lead is also a hard-wired
input to the spark drivers. Thus the processor isn't expected to halt in
place; it is expected to recognize the input and honor it. If it doesn't
honor it and is lost in the weeds, the P-lead will command the spark drivers
to halt spark generation anyway. As Bob put insinuated, this relieves
Emagair from the rigors or high-risk software certification and the need to
do failure monitoring by pushing the tasks off onto hardware. This is good
design practice, regardless.
Now, I do agree with you that if something goes wrong big time and you want
to kill the processor to avoid further harm, it is good to kill the clock,
but one other thing must be done and that is to asynchronously reset all
hardware. Asynchronously resetting hardware assures that all flip flops
will reach a known state regardless of whether there is a clock or not, and
all outputs should be a function of these states, either directly or through
combinatorial circuitry. There are several ways to achieve a quiet clock,
but grounding (crow-barring) a clock or any output, for that matter, is not
good design practice, as it affects circuit reliability later on.
Killing the clock is true only for today's synchronous CPUs. I was aware of
two companies in the 1990s that were involved in designing asynchronous CPUs
for the benefit that asynchronocity brings, but they went under as far as I
know; however, some other company or government entity may have bought their
technology and made product somewhere without my knowledge. Such product,
of course, uses no clock. I cannot foresee asynchronous CPUs being with us
in the near future, so I shouldn't even have brought them up! I just wanted
to CMA.
Simon
Message 10
| INDEX | Back to Main INDEX |
| PREVIOUS | Skip to PREVIOUS Message |
| NEXT | Skip to NEXT Message |
| LIST | Reply to LIST Regarding this Message |
| SENDER | Reply to SENDER Regarding this Message |
|
| Subject: | Re: E-Mag P-Mag Safety update . . . |
At 02:22 PM 9/19/2008 -0400, you wrote:
>Bob,
>Thanks for acting as our expert go-between with EMagAir. I've been
>tempted to call them, but I'm sure they would prefer to avoid another
>phone call or email of possible. Your contact with them answered
>questions for all of us while simultaneously reducing the duplicative
>responses for Brad.
>Stan Sutterfield
>
>Call Emag and if push comes to shove, run a Magneto/Emag combination
>for awhile. Bottom line is that the sky is not falling.
I've been having some private conversations with interested
parties hoping to put a team effort together to evaluate
Emag's proposed fix for wandering magnets . . . and assuage
market-place concerns about these serious failures modes.
It seems that my proposal is not in conformance with the
business model of some invitees.
In the mean time, know that if you are in possession of
a device described in Emag's service bulletin, you're
well advised to take them up on the offer of upgrade.
I've got an alterative plan to explore for verification
of design integrity.
Bob . . .
----------------------------------------)
( . . . a long habit of not thinking )
( a thing wrong, gives it a superficial )
( appearance of being right . . . )
( )
( -Thomas Paine 1776- )
----------------------------------------
Other Matronics Email List Services
These Email List Services are sponsored solely by Matronics and through the generous Contributions of its members.
-- Please support this service by making your Contribution today! --
|