Replies to This Discussion

Permalink Reply by Ken on September 5, 2010 at 6:02am

There is a big push in defense acquisition to acquire COTS (commercial, off the shelf) technology when and where it makes sense. Whether or not you can find a UAV program that is seeking COTS, I don't know.

But there is no way in heck that ArduPilot would be accepted in a traditional procurement.
* commercial temp grade parts; it gets hot in the desert, it gets cold in the sky
* plastic encased chips (non-hermetic); water absorbed in chips, not good during thermal cycling
* no conformal coat; humidity causes shorts; trace corrosion
* no packaging ("case"); easy to short out
* connectors too fragile
* no vibrational analysis (nor spec)
* no EMI/EMC analysis (nor spec)
* no software spec
* perhaps 1% of the needed documentation
* yada yada yada

For 12 years I worked at an avionics facility where we would validate drawing packages, do builds, do thermal, EMI & vibration testing. Every year or two someone would think they could do something for far less than what the government was paying and when they were ignored in the bid process they would complain to their senator/congressperson. We would have to stop our work and test their home made stuff. The guy would come in all gloating because he pulled the strings and was going to prove us all wrong. We would bolt his equipment to the shaker table and start the vibration profile.

As young Jr engineer I went to get the broom and dustpan, ready to sweep up the parts as they flew off his equipment. In seconds or minutes his equipment would cease operation. In minutes parts physically would separate. At the end of an hour of testing I would hand him a pile of components and pieces in a shoe box.

Even with specs as tough as we had to meet for fleet use, we still had huge numbers of units come back after a flight ... failed when cold; failed when hot; failed above 10,000 feet; wouldn't power on for second flight...

• ▶ Reply

DeveloperPermalink Reply by Mark Colwell on September 5, 2010 at 6:42am

Common electronics don't like the temperature cycling with vibration testing at ambient, cold and hot !!
I took apart many devices this way. I tested potable handheld computers for the marines back in 81. If I LED bond wire broke we had to fail the unit. (We would replace the bond wire and shake and bake again.) I liked doing these tests as results were dramatic when failures occurred. The test fixtures are massive at 50 pounds, we could shake 4 units at once. If an accelerometer breaks or comes loose the test goes crazy bad. Throwing parts all over.

• ▶ Reply

Permalink Reply by Tom Kent on September 6, 2010 at 6:42am

I agree completely, however I see an opening for something "disposable" in a category that doesn't need the same reliability as avionics that go on a manned airplane.

If the plan is to use it only once, then the fact that it fails a vibration test that simulates a vibration it feels once every 1000 flying hours isn't really relevant. If something goes wrong, just send another one.

It would be ideal to have everything work all the time, but on some level you have to sit back and look at what is cost effective. I'd definitely trade tens of millions of dollars in development costs for a 5% failure rate instead of a 3% failure rate.

• ▶ Reply

Permalink Reply by Michael Lingg on September 6, 2010 at 7:50am

You might be willing but what about manned aircraft that has to share airspace with a high failure rate unmanned aircraft? I'm not sure the FAA would allow such an equipment spec...

• ▶ Reply

Permalink Reply by Patrick Hammer on September 6, 2010 at 8:47am

Yer, thats what I was thinking - never mind any Mil Spec, it's satisfying the FAA that has developers throwing thmselves in front of trains - after all hair has been torn out, the divorce is completed, the house remortgaged, the car sold etc etc ........

I have a friend in the USA who built an experimental plane - out of carbon fibre. It was such a good plane and everyone said, get it certified and you'll make a fortune ... but what the FAA required from him at all levels (production, testing, avionics etc etc ....) to get the plane certified was scary scary. He reackoned, even if things went 100% smoothly first time round, he'd be lucky if he had change from $10 000 000!!! - of which around 70% of that cost would go into satifying FAA safety requirements.

• ▶ Reply

Permalink Reply by Florin Mingireanu on September 6, 2010 at 9:53am

Aviation and aerospace, in general, should be the domain of those courageous (but with an educated courage! which is different from stupid courage... :) ) and not of those that just know how to handle paperwork and are experts in this and that regulation, this and that rule, this and that paperwork method...

I remember that one astronaut once said (and that gentleman flew to the Moon) that had FAA been in action with the same paperwork back in 60's then they (NASA) would have never took off ground...

Just my 2 cents...

Thanks!

• ▶ Reply

Permalink Reply by Patrick Hammer on September 6, 2010 at 10:36am

Well said Florin.

• ▶ Reply

Permalink Reply by Curt Olson on September 6, 2010 at 11:36am

Imagine if airplanes had only been invented this week and UAV's had been flying since the early 1900's. The FAA would never let manned aircraft fly in the national airspace. Look at all the equipment failures of full scale aviation and the number of people onboard and on the ground that have been injured or killed. We have to follow the rules of course, but digging around and looking closely at risk tolerance and risk aversion always turns up interesting inconsistencies.

• ▶ Reply

Permalink Reply by Gerry Lichter on September 6, 2010 at 11:30am

[quote]It would be ideal to have everything work all the time, but on some level you have to sit back and look at what is cost effective. I'd definitely trade tens of millions of dollars in development costs for a 5% failure rate instead of a 3% failure rate.[/quote]

I would tend to agree with this. There's a price/performance curve, and if by political necessity you determine the sweet spot to be deep in the "diminishing returns" territory: it leaves you exposed to competition from more courageous players who are willing to sacrifice some reliability and safety to be competitive.

This may all be philosophical though because I doubt it's politically realistic to expect to be successful in tempering the expectations of the public on safety and reliability for the sake of saving money.

Money is no object when the stakeholders making the decisions are insulated from the costs of those decisions by multiple degrees of separation.

Epoxy potting is one way that circuits can be hardened to vibration but without mil-spec, or FAA approval: there's not much incentive to remove the ability to replace components. Parafin may be a good middle of the road if this actually becomes a problem as it's removable if a component should fail.

If the politics and red tape won't allow for it elsewhere: maybe the status quo of the exclusive domain of law enforcement and hobby markets is appropriate.

• ▶ Reply

Permalink Reply by Gerry Lichter on September 6, 2010 at 11:31am

This was very informative, and helped me understand the difficulty of the task at hand. Thank you.

• ▶ Reply

Permalink Reply by Matthias Badaire on September 7, 2010 at 2:33am

Actually this is very interesting .
I'd like to know if somebody has done or is developing an encasement for ArduPilot / ArduPilot Mega. Obvisouly this is not to reach Military Spec but at least to get better protection .

• ▶ Reply

Permalink Reply by Krzysztof Bosak on September 7, 2010 at 3:08am

"Is ardupilot military/commercial grade?"
It is not.
1. No regression testing.
2. No electronic compatibility testing.
3. No logging.
In short no support for quality control.
Just found 3 reasons, could have found 20.
This is an open source project for brainstorming.

• ▶ Reply

• ‹ Previous
• 1
• 2
• 3
• Next ›
• Page