A tale of Quality from the Space Age

Things that are not directly connected with Factorio.
User avatar
BlueTemplar
Smart Inserter
Smart Inserter
Posts: 3007
Joined: Fri Jun 08, 2018 2:16 pm
Contact:

A tale of Quality from the Space Age

Post by BlueTemplar »

From a discussion about :
Voyager 1 Breaks Its Silence With NASA via a Radio Transmitter Not Used Since 1981

In particular a specific comment :
https://news.ycombinator.com/item?id=42167188
yodon wrote:
Nothing lasts forever, and if you don't figure out when it's going to fail, it's going to be sooner rather than later.
You might be surprised about the reality of the situation.

I had a professor who worked on the design and fabrication of the Apollo Guidance Computers, which likely was a somewhat similar process to the one being discussed here. It's been quite a few years since his lecture on it, but the process went something like this:

They started with an analysis of the predicted lifetime/reliability of every chip type/component available to potentially include in the design.

The design was constrained to only use components with the top x% of predicted life.

Then they surveyed each manufacturer of each of those component types to find the manufacturer with the highest lifetime components for each of the highest lifetime component types.

Then they surveyed the manufacturing batches of that manufacturer, to identify the batches with the highest lifetimes from that manufacturer.

Then they selected components from the highest lifetime batches of from the highest lifetime manufacturers of the highest lifetime components.

Using those components, they assembled a series of guidance computers, in batches.

They tested those batches, pushing units from each batch to failure.

They then selected the highest quality manufacturing batch as the production units.

When he gave this talk, decades after the Apollo era, NASA had been continuing to run lifetime failure analyses on other units from the production batch, to try to understand the ultimate failure rate for theoretical purposes.

Several decades after the Apollo program ended, they had still never seen any failure events in these systems, and shortly before the time of his lecture, I believe NASA had finally shut off the failure testing of these systems, as they were so remote from then "modern" technology (this was decades ago, hence the quotes around "modern").

This is what happens when you have the best minds committed to designing systems that don't fail. Yes, the systems probably will fail before the heat death of the universe. No, we don't have any idea when that failure time will be. Yes, it's likely to be a very long time in the future.

(And, of course, this is typed from memory about a lecture decades ago on events happening decades before that. This being HN, someone here probably worked on those systems, in which case hopefully they can add color and fix any defects in the narrative above).
BobDiggity (mod-scenario-pack)
Post Reply

Return to “Off topic”