X-Virus-Scanned: clean according to Sophos on Logan.com Return-Path: Sender: To: lml@lancaironline.net Date: Wed, 14 Oct 2009 22:22:29 -0400 Message-ID: X-Original-Return-Path: Received: from web57513.mail.re1.yahoo.com ([66.196.100.80] verified) by logan.com (CommuniGate Pro SMTP 5.2.16) with SMTP id 3885481 for lml@lancaironline.net; Wed, 14 Oct 2009 09:22:19 -0400 Received-SPF: none receiver=logan.com; client-ip=66.196.100.80; envelope-from=casey.gary@yahoo.com Received: (qmail 29251 invoked by uid 60001); 14 Oct 2009 13:21:44 -0000 DomainKey-Signature:a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; h=Message-ID:X-YMail-OSG:Received:X-Mailer:Date:From:Subject:To:MIME-Version:Content-Type; b=mqUYl2BEKWO+zRynt9R/I6IP31gDLa3eyyfUV+8DSADe8KnRuVv88sOJXtmxF5sAXtKdXwXv2t3LVmEdlB9mTSiWWHd0PpbL2cNbzGq68i/qGxokvD60/bHgrlMXi3oksykTuvYSpnCvSjyUGfVx3iviLd1YJd9GTDpuEnZt27o=; X-Original-Message-ID: <909421.19140.qm@web57513.mail.re1.yahoo.com> X-YMail-OSG: KgdpVeEVM1n4r9wxBERw.M0SOjXBlSB4HlYx20YaIeS2SSiI1R3vnYKiOEPqqAKc7eSKTatVxe7Rr2NF.Yo_MjNNmyCrcqLy..XNezHoOzkKp6lGX4PNPfB0SXyCpgseutyjVlJLZy7SWRZyalJPBtRXtsN30pzwHYo5sNzgC.rvUHl6l5vX7EE9H4j5cFkYVvrF9zIl_OmwghcnVXSxMKOSeIve06xuyBko1pWlHMCX7hD55wUXeJ9xs.rs5EwTFJWW3F4MGpLc7uAuAJWuDWVNAYb0MzVTHg8ENLh96bouV785FlSflVP3O.kWk0Zzukb4S0W.bnSV2ts- Received: from [97.122.189.63] by web57513.mail.re1.yahoo.com via HTTP; Wed, 14 Oct 2009 06:21:43 PDT X-Mailer: YahooMailRC/182.10 YahooMailWebService/0.7.347.3 X-Original-Date: Wed, 14 Oct 2009 06:21:43 -0700 (PDT) From: Gary Casey Subject: Re: Transient suppression on relays and Snapjacks X-Original-To: Lancair Mailing List MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="0-1506190906-1255526503=:19140" --0-1506190906-1255526503=:19140 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable I'm sure Eric makes a fine product that will work as he advertises, but I h= ave to take a very slight exception to some of his statements. True, diode= s can fail open or shorted. Generally in this application, failing open is= relatively benign since the system will still function as intended - there= will be no system fault and the only problem will be a loss of protection = of the switch against arcing. Failing shorted would seem to be a serious f= ault, but most(see exception below) diodes, zener or not, won't survive the= resulting high current and will consequently destroy themselves and go ope= n. The result is an open circuit whether the diode junction itself fails o= pen or shorted. A standard diode (we would call them "free-wheeling diodes= ") used in this application only sees current during the suppression mode a= nd absorbs about 0.7 volts for maybe 0.1 sec at 0.5 amps peak, decaying to = zero. The total energy is very small (about 0.02 watt-seconds), as most of the inductive energy is dissipated in the resistance of the coil. = A zener, on the other hand, is there to absorb the inductive energy itself= . It might dissipate 50 volts for 0.01 seconds at the same average current= for about 1watt-second of energy. the instantaneous power is much, much h= igher - .35 watts for the diode and 25 watts for the zener. Point is that = the zener is "working" much harder, although for a much shorter period of t= ime, and it is designed to be able to do that. Point of all this is that I= don't buy the statement that "their electrical failures are virtually impo= ssible...". I deliberately left off the qualifier "when used as specified"= because that could be said about any of the other suppression methods as w= ell.=0A=0AEither form of diode suppression results in a diode continuously = "bucking" battery voltage when the switch is turned on. A standard free-wh= eeling diode (something like a 1N4007) has a voltage rating of maybe 1000 v= olts, so it will not conduct under most any voltage transient seen in an ai= rcraft, which could be up to 150 volts. On the other hand, a zener is desi= gned to conduct in the reverse direction at a given voltage. These are lik= e most back-to-back zeners in that they are bidirectional, having the same = voltage rating in each direction. The "Snap Jack" supplied, I presume, for= a 14-volt system is the P6KE18CA which, if I read the data sheet correctly= , will start to conduct at 18 volts (1mA) and will be conducting 24 Amps at= 25 Volts. Sorry, but I would not use a device with that little overhead. = Instead I would pick one with a rating of at least 35 and maybe more like = 60 volts. Remember, this device is connected directly to the battery when = the switch is turned on, so it will try to absorb any voltage transient in the= system. Essentially it is protecting the whole electrical system. I'm su= re that's why Eric picks a device with such a high power rating, far and aw= ay higher than is required to absorb the transient from a single solenoid c= oil(the rated peak power dissipation is 600 watts for 1msec, 100 watts for = 10msec).=0A=0ACertainly overrating the device will reduce the probability o= f failure, but it could make the failure more serious. It could, if shorte= d, conduct enough current to to either melt the switch wire or open any pro= tection device upstream (CB). In that case the system would be rendered in= operative. The tradeoff then is an unlikely serious failure mode vs a more= likely benign failure mode. In the pass car business we strive for the un= likely failure mode, but the typical aircraft design practice is to accept = a higher potential failure RATE in favor of the benign failure MODE (system= keeps working). A complete electrical failure in a car, even though it st= ops the engine dead, is far less serious than a complete electrical failure= in an aircraft even though the engine keeps running.=0A=0ASorry about the = long post and, again, I'm sure that Eric sells a reliable, high quality pro= duct. For me, I'll use no suppression at all, thank you. What would be my= failure mode? The master might not shut off if the switch contacts weld. = Then I would have to disconnect the battery or let the battery go dead.=0A= Gary=0A=0A=0A=0A________________________________=0AFred,=0A =0AI am also in= terested in transient suppression. I build it into most of the devices I se= ll, because I can't depend on the customers' attention to this detail. As a= result, the more of my parts you have, the more transient suppression you = have as well.=0A =0AMy comments on the methods of failure--=0A=E2=80=9CThe = following list....:=0A=09* Capacitors (shorted), especially electrolytic ca= pacitors. The paste electrolyte tends to lose moisture with age, leading to= failure. Thin dielectric layers may be punctured by overvoltage transients= .=0AOne comment--I have avoided the use of electrolytics as far as possible= in all my design work for the reasons stated. Often a REALLY BIG electroly= tic gets used where a smaller tantalum plus a series inductor plus a small = parallel ceramic and a zener should be used. This is sloppy engineering.=0A= =09* Diodes open (rectifying diodes) or shorted (Zener diodes).=0ATrue as a= Rule of Thumb...but it depends on parts and circumstances.=0A=09* Inductor= and transformer windings open or shorted to conductive core. Failures rela= ted to overheating (insulation breakdown) are easily detected by smell.=0AT= rue. Let's try to minimize transformers and relay coils.=0A=09* Resistors o= pen, almost never shorted. Usually this is due to overcurrent heating, alth= ough it is less frequently caused by overvoltage transient (arc-over) or ph= ysical damage (vibration or impact). Resistors may also change resistance v= alue if overheated!=0ATrue. Design note: The almost-obsolete carbon-comp re= sistors perform well in voltage-transient and overvoltage conditions. =0AMy= SnapJacks are Bidirectional Zeners and as such never see current except du= ring transient suppression. It is hard to judge their failure modes since t= heir electrical failures are virtually impossible when used as specified. M= echanical failure from vibration would be possible, and then they would fai= l open. I have never seen or heard of a failure short of a lightning strike= .=0AThe issue of zeners, bidirectional zeners --or even diodes in general--= failing short always brings up the idea of adding a fuse. This fuse could = be far less reliable than the diode. A better solution would be to add a se= ries resistor to limit the fault current, but it would be problematic as to= whether or not this improves the function of the suppression diode.=0ABy t= he way, the question of lighning strike suppression needs addressing. Light= ning is attracted differently on composite or metal-skinned aircraft, but i= n general, adding arc suppression to the "corners" and the main busses of t= he aircraft seems to be a cost effective thing to do.=0AYou're welcome to r= epublish this on any list.=0ARegards,=0AEric M. Jones=0A=0A=0A --0-1506190906-1255526503=:19140 Content-Type: text/html; charset=utf-8 Content-Transfer-Encoding: quoted-printable
=0A
I'm sure Eric makes a fine product = that will work as he advertises, but I have to take a very slight exception= to some of his statements.  True, diodes can fail open or shorted. &n= bsp;Generally in this application, failing open is relatively benign since = the system will still function as intended - there will be no system fault = and the only problem will be a loss of protection of the switch against arc= ing.  Failing shorted would seem to be a serious fault, but most(see e= xception below) diodes, zener or not, won't survive the resulting high curr= ent and will consequently destroy themselves and go open.  The result = is an open circuit whether the diode junction itself fails open or shorted.=  A standard diode (we would call them "free-wheeling diodes") used in= this application only sees current during the suppression mode and absorbs= about 0.7 volts for maybe 0.1 sec at 0.5 amps peak, decaying to zero.  The total ener= gy is very small (about 0.02 watt-seconds), as most of the inductive energy= is dissipated in the resistance of the coil.  A zener, on the other h= and, is there to absorb the inductive energy itself.  It might dissipa= te 50 volts for 0.01 seconds at the same average current for about 1watt-se= cond of energy.  the instantaneous power is much, much higher - .35 wa= tts for the diode and 25 watts for the zener.  Point is that the zener= is "working" much harder, although for a much shorter period of time, and = it is designed to be able to do that.  Point of all this is that I don= 't buy the statement that "their electrical failures are virtually impossib= le...".  I deliberately left off the qualifier "when used as specified= " because that could be said about any of the other suppression methods as = well.

Either form of diode suppression results in a diode continuously "bucking" battery voltage when the switch = is turned on.  A standard free-wheeling diode (something like a 1N4007= ) has a voltage rating of maybe 1000 volts, so it will not conduct under mo= st any voltage transient seen in an aircraft, which could be up to 150 volt= s.  On the other hand, a zener is designed to conduct in the reverse d= irection at a given voltage.  These are like most back-to-back zeners = in that they are bidirectional, having the same voltage rating in each dire= ction.  The "Snap Jack" supplied, I presume, for a 14-volt system is t= he P6KE18CA which, if I read the data sheet correctly, will start to conduc= t at 18 volts (1mA) and will be conducting 24 Amps at 25 Volts.  Sorry= , but I would not use a device with that little overhead.  Instead I w= ould pick one with a rating of at least 35 and maybe more like 60 volts. &n= bsp;Remember, this device is connected directly to the battery when the switch is turned on, so it will try to absorb any voltage transient in= the system.  Essentially it is protecting the whole electrical system= .  I'm sure that's why Eric picks a device with such a high power rati= ng, far and away higher than is required to absorb the transient from a sin= gle solenoid coil(the rated peak power dissipation is 600 watts for 1msec, = 100 watts for 10msec).

Certainly overrating the de= vice will reduce the probability of failure, but it could make the failure = more serious.  It could, if shorted, conduct enough current to to eith= er melt the switch wire or open any protection device upstream (CB).  = In that case the system would be rendered inoperative.  The tradeoff t= hen is an unlikely serious failure mode vs a more likely benign failure mod= e.  In the pass car business we strive for the unlikely failure mode, = but the typical aircraft design practice is to accept a higher potential failure RATE in favor of the benign failure MODE (system keeps w= orking).  A complete electrical failure in a car, even though it stops= the engine dead, is far less serious than a complete electrical failure in= an aircraft even though the engine keeps running.

Sorry about the long post and, again, I'm sure that Eric sells a reliable,= high quality product.  For me, I'll use no suppression at all, thank = you.  What would be my failure mode?  The master might not shut o= ff if the switch contacts weld.  Then I would have to disconnect the b= attery or let the battery go dead.
Gary

Fred,

 

I am also interest= ed in transient suppression. I build it into most of the devices I sell, be= cause I can't depend on the customers' attention to this detail. As a result, the more o= f my parts you have, the more transient suppression you have as well.<= /span>

 

=E2=80=9CThe following list....:

  • Capa= citors (shorted), especially electrolytic capacitors. The = paste electrolyte tends to lose moisture with age, leading to failure. Thin dielectric layers may be punctured by = overvoltage transients.

One comment--I have avoided the use of electr= olytics as far as possible in all my design work for the reasons stated. Of= ten a REALLY BIG electrolytic gets used where a smaller tantalum plus = a series inductor plus a small parallel ceramic and a zener should be used. This is= sloppy engineering.

  • Diodes open (rectifying diodes) or shorted (Zener diodes).

True as a Rule of Thumb..= .but it depends on parts and circumstances.

  • Inductor and transformer win= dings open or shorted to conductive core. Failures related to overheating (insulation br= eakdown) are easily detected by smell.

True. Let's try to minimize transformers and relay coils.

  • Resistors open, almost never shorted. Usually this is due to overcur= rent heating, although it is less frequently caused by overvoltage transien= t (arc-over) or physical damage (vibration or impact). Resistors may also c= hange resistance value if overheated!

True. Design note: The almost-obs= olete carbon-comp resistors perform well in voltage-transient and over= voltage conditions. 

My SnapJa= cks are Bidirectional Zeners and as such never see current except during tr= ansient suppression. It is hard to judge their failure modes since their el= ectrical failures are virtually impossible when used as specified. Mec= hanical failure from vibration would be possible, and then they would = fail open. I have never seen or heard of a failure short of a lightning strike.

The issue= of zeners, bidirectional zeners --or even diodes in general-- failing shor= t always brings up the idea of adding a fuse. This fuse could be far l= ess reliable than the diode. A better solution would be to add a series res= istor to limit the fault current, but it would be problematic as to whether= or not this improves the function of the suppression diode.

<= span style=3D"font-size: 12pt; ">By the way, the question of lighning strik= e suppression needs addressing. Lightning is attracted differently on compos= ite or metal-skinned aircraft, but in general, adding arc suppression to th= e "corners" and the main busses of the aircraft seems to be a cost effectiv= e thing to do.

You're welcom= e to republish this on any list.

Regards,
Eric M. Jones

=0A=0A --0-1506190906-1255526503=:19140--