Reliability and NAVSEA Derating

[amptramp]

Derating is a design policy that limits the amount of stress a component is subjected to in order to improve reliability.  For example, if you have a power supply giving out 15 volts and you use a filter capacitor rated at 25 volts, you are running at 15/25 or 60% derating.

The NAVSEA Derating guide is here:

http://www.navsea.navy.mil/nswc/crane/sd18/default.aspx

To use the document, check down the sidebar to the component type, click on it and read the table provided.  From memory many years back, NAVSEA was slightly more stringent than NASA in some derating areas but neither are outrageous.  The document format is a bit less formal than the MIL-SPEC's, MIL-HDBK's and MIL-STD's that are still available on the net.

If you want to calculate reliability of components, you can use MIL-HDBK-217F which is at:

http://www.sre.org/pubs/Mil-Hdbk-217F.pdf

There is a 217G version out now that may cover more modern devices, but I haven't looked at it yet.  You calculate failure rate per million hours and add the failure rates of all the components to get a total then take the reciprocal which becomes your Mean Time Between Failures or MTBF.  There are usually contractual requirements for the MTBF to be above a certain level.

Stompboxes have to survive being tossed into a van and driven in any temperature conditions and they have to work when they get there - you don't want to get the reputation of building devices that almost work or only work in studio conditions.  The Fuzz Face and Cry Baby became popular because iconic players used them.  If they were released as new devices today, their limited temperature range and sensitivity to power supply issues would put people off.  A little bit of military design discipline never hurt any project and contrary to what you might think, it doesn't cost much more to be reliable - you just have to make the effort to prove your design follows the derating rules and you can calculate reliability if you wish.  If you offer a warranty with your stompbox, the reliability calculation provides input to what your warranty costs will be.  You can generate a spreadsheet of failure rate per component and cost of each component plus warranty labour and build a solid business analysis of actual costs.

If this is seen to be a popular idea, there are some design spreadsheets I have for tolerance and stress that show up potential problems quickly.  If I can figure out how to attach a spreadsheet and there is interest, it will provide a view of how modern (F35 era, which is where I used them) designs can be audited.  Filling out the form forces you to think and even the best designers usually had several issues to resolve.  Any interest?

[PRR]

> NAVSEA was slightly more stringent than NASA in some derating

Navy puts a million-ton cruiser to sea and wants it to work for 30 years.

When NASA lifts a thing 100 miles they can't afford spare weight, and 30 _days_ is long for many missions. Reasonable to design a little shy.

NASA did have additional qualifications. Vacuum and no-cooling for one, but also wanted deep reliability records for the specific part on the specific production facility. Navy has guys/gals who change resistors on-board. NASA often has nobody up there; if crewed, they may be busy with other problems (duct-taping the toilet again).

> A little bit of military design discipline never hurt

IMHO and in the MTBF docs, *Solder Joints* fall to the top of the list of trouble-makers. In much stomp-work, all other parts combined have a computed (not real) MTBF of several centuries, because stresses are so low. *Inspected* soldering easily cuts into that, and UN-inspected soldering (typical DIY and small-run work) can be much worse (or just as good, if you obsess it).

[amptramp]

This is where you get a good soldering quality manual:

http://jptronics.org/Collins/Quality/collins_quality.1.pdf

Collins radio is one of the top quality companies in the communications industry and was one of the leaders in amateur radio as well as military radio.  Here is the rest of the document:

http://jptronics.org/Collins/Quality/collins_quality.2.pdf

I have looked at the NASA NHB 5300.4 which is here:

http://nepp.nasa.gov/docuploads/06AA01BA-FC7E-4094-AE829CE371A7B05D/NASA-STD-8739.3.pdf

I have some quibbles with the NASA standard.  Solder is heavy, so their emphasis is on using the minimum amount that is acceptable.  This can result in dry joints for no useful reason in terrestrial use.  The other problem is that if you solder a through-hole component, you apply the heat from one side and the solder from the other which means the pad will oxidize before the solder reaches it.  Our component reliability man called me over one time to look at a board with flatpack IC's soldered to it and asked what I thought.  I said it looked like dry contacts.  He said, "Use the jeweller's loupe and take a look.  It is the best soldering you will ever see."  I tool a look and it was postcard-perfect.  The solder fillet went exactly halfway up the thickness of the gullwing leads.  He was right.  Forty years later, I have never seen anything even come close to it.  Spacecraft may use trailing edge technology because of the need to have an established reliability for everything, but I miss the exquisite workmanship.

The Attitude Control Electronics Assembly for the Hermes spacecraft was supposed to weigh 12.000 pounds and it was estimated that 0.700 pounds would be solder.  (BTW the finished unit came in at 12.017 pounds.  The complaint went from the president of our company (which was the buyer) to the president of the vendor company.  They were well under their weight budget until they added the foam potting, which they tried for the first time on the flight unit.  Instead of the 2.4 pounds per cubic foot that they should have got, it came out at 6.4 pounds per cubic foot because they never tried it until they built the flight unit.  Idiots.

[PRR]

> using the minimum amount that is acceptable.  This can result in dry joints for no useful reason

You use the minimum amount because extra solder hides your sins. You need to see that the solder has "wet" both surfaces; with a blob you can never know.

Like your half-way up joints. You can see by the contact-angle of solder on wing that the solder wet the wing. If the solder ran to the sharp angle at the top of the wing, you couldn't be sure.

Also solder is expensive, though usually not the budget-breaker.

[bloxstompboxes]

Also solder is expensive, though usually not the budget-breaker.

Yes it is. Here at work, we get it in bars like gold and in spools as well. I don't want to know how much the bars for the waves and select solder macines costs.