X-Virus-Scanned: clean according to Sophos on Logan.com Return-Path: Received: from elasmtp-junco.atl.sa.earthlink.net ([209.86.89.63] verified) by logan.com (CommuniGate Pro SMTP 6.0.7) with ESMTP id 6562841 for lml@lancaironline.net; Thu, 24 Oct 2013 09:57:30 -0400 Received-SPF: none receiver=logan.com; client-ip=209.86.89.63; envelope-from=colyncase@earthlink.net DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=dk20050327; d=earthlink.net; b=p742gNvlkOIkLS7CJMLRvPjjFIbTJhqS0L6q9qi52TMBbdYTdY7r87Y6XwHNbhnZ; h=Received:From:Mime-Version:Content-Type:Subject:Date:In-Reply-To:To:References:Message-Id:X-Mailer:X-ELNK-Trace:X-Originating-IP; Received: from [70.16.215.180] (helo=[192.168.1.24]) by elasmtp-junco.atl.sa.earthlink.net with esmtpa (Exim 4.67) (envelope-from ) id 1VZLPE-00015f-FO for lml@lancaironline.net; Thu, 24 Oct 2013 09:56:56 -0400 From: Colyn Case Mime-Version: 1.0 (Apple Message framework v1085) Content-Type: multipart/alternative; boundary=Apple-Mail-4-588889902 Subject: Re: [LML] Re: Hot TITs on X country leg. LIVP Date: Thu, 24 Oct 2013 09:56:55 -0400 In-Reply-To: To: "Lancair Mailing List" References: Message-Id: <50E3F923-138E-4BD8-B41A-409EF48D9348@earthlink.net> X-Mailer: Apple Mail (2.1085) X-ELNK-Trace: 63d5d3452847f8b1d6dd28457998182d7e972de0d01da9403f54b8607384f575c982d44a45f64b21350badd9bab72f9c350badd9bab72f9c350badd9bab72f9c X-Originating-IP: 70.16.215.180 --Apple-Mail-4-588889902 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=us-ascii Hi Gary, no it wasn't just you. =20 The .001% I'm alluding to is the number of GA flight hours that involve = fatalities. Statistically that is a sum of various failures including engine = failures for whatever reason. So that would imply you want any one system to be quite a bit better = than that. Your point on sample size is well taken. =20 One really big problem with aviation electronics is that the total = available market is so small. When I was working in commercial semiconductors we would do a test run = of 80,000 chips to find the weak spots in the production process. = ...then go back and tweak the physical layout and go to full production. = You can't do that when your total lifetime market is in single digit = thousands. You are part of the test run. =20 but if you want to talk about stats on ignition systems, suppose there = are 2,000 units of a given system in the field and they have all flown = 200 hours. If you talked to all 2,000 owners and only found 2 failures = EVER that would be about in line with the safety goal. On the other = hand if you've talked to only 10 of the 2,000 and 9 had failures and = even if somehow you knew they were the only failures in the world, ever, = you already have reason for concern. Colyn On Oct 24, 2013, at 9:06 AM, Gary Casey wrote: I'm sure Colyn was referring to my experience with the Lightspeed (III) = system that failed 3 times in less than 400 hours (I'm on my 4th one, = making 3 failures, not 4, but I'll stay with 4 for the purpose of = discussion). Regardless, 4 failures would have made for a per-hour = failure rate of 1%, much, much worse than his .001% that he says is the = fleet average. Okay, but the sample size is very small, making the = statistic very weak. If we add one other plane to the statistic, = another ES that I know of that has over 1,000 hours on dual Lightspeeds = with no failures, the number now is 4 failures in 2,400 hours, or an = hourly failure rate of 0.17%, 6 times "better." Still not good, but it = illustrates the weakness of coming to a conclusion from such a small = data set. Gary Colyn posted: So to get to GA accident rates we as a fleet have free of failures that could contribute to a fatality (on all systems including the pilot) = 99.999% of the hours flown. A system that fails 1% of the hours flown is about 1000 times more = dangerous than that. Something to think about when selecting components. --Apple-Mail-4-588889902 Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset=us-ascii Hi = Gary,  no it wasn't just you.  
The .001% I'm alluding to = is the number of GA flight hours that involve = fatalities.
Statistically that is a sum of various failures = including engine failures for whatever reason.
So that would = imply you want any one system to be quite a bit better than = that.

Your point on sample size is well taken. =  

One really big problem with aviation = electronics is that the total available market is so = small.
When I was working in commercial semiconductors we = would do a test run of 80,000 chips to find the weak spots in the = production process.  ...then go back and tweak the physical layout = and go to full production.   You can't do that when your total = lifetime market is in single digit thousands.  You are part of the = test run.   

but if you want to talk = about stats on ignition systems, suppose there are 2,000 units of a = given system in the field and they have all flown 200 hours.   If = you talked to all 2,000 owners and only found 2 failures EVER that would = be about in line with the safety goal.   On the other hand if = you've talked to only 10 of the 2,000 and 9 had failures and even if = somehow you knew they were the only failures in the world, ever, you = already have reason for = concern.

Colyn

On Oct = 24, 2013, at 9:06 AM, Gary Casey wrote:

I'm = sure Colyn was referring to my experience with the Lightspeed (III) = system that failed 3 times in less than 400 hours (I'm on my 4th one, = making 3 failures, not 4, but I'll stay with 4 for the purpose of = discussion).  Regardless, 4 failures would have made for a per-hour = failure rate of 1%, much, much worse than his .001% that he says is the = fleet average.  Okay, but the sample size is very small, making the = statistic very weak.  If we add one other plane to the statistic, = another ES that I know of that has over 1,000 hours on dual Lightspeeds = with no failures, the number now is 4 failures in 2,400 hours, or an = hourly failure rate of 0.17%, 6 times "better."  Still not good, = but it illustrates the weakness of coming to a conclusion from such a = small data set.
Gary


Colyn = posted:

So to get to GA accident rates we as a = fleet have free of failures that
could = contribute to a fatality (on all systems including the pilot) = 99.999%
of the hours = flown.

A system that fails 1% of the hours flown is about 1000 times = more dangerous
than = that.

Something to think about when selecting = components.

= --Apple-Mail-4-588889902--