NASA’s Mars helicopter was much more revolutionary than we knew

It was an expensive PR stunt, mostly.

What value is all that compute power on Mars? We can do computing here.

With a one-way propagation delay of from 8 minutes to 21 minutes, depending on current orbital positions. This may make flight control ... interesting.

The big demonstration was that ordinary COTS stuff can work at all on Mars, so maybe we don't need to full space-qualified everywhere.

Joe Gwinn

Well, they did spend $80M on its development. Hardly COTS, I'd say. OK, it has a few COTS components. Don't they all?

Jeroen Belleman

Yes, but the chips are plain old COTS, just like in your cellphone. This was not supposed to be workable in space, or on Mars. And yet ...

A parallel happened when the US DoD discovered that commercial epoxy IC packages yielded more reliable components the full hermetic ceramic packages. DoD paid for lots of component development back in the day when only DoD systems could afford ICs, but it was the automotive environment that drove reliability under harsh conditions. For a nickel, not a few hundred dollars.

Joe Gwinn

Joe Gwinn

Generally true for high temperature 180'C designs too. The epoxy packages have a little 'give' when hot, the ceramics don't.

I'm surprised that radiation wasn't a problem though. I would have thought that older, larger geometries would be better.

I've used COTS components in radiation environments for years. Not everything is radiation resistant, but over the years, you learn a bit about what works and what doesn't.

Jeroen Belleman

I believe it's not the size that makes a device rad-hard. Mostly, old designs would tolerate larger spreads in device parameters. They would survive radiation-induced parameter shifts as well.

Some old designs contained lateral PNPs that were only barely good enough. Those would fail early under irradiation. Low power chips do not survive for very long either. All-NPN designs with mA standing currents survive kGy doses just fine. YMMV.

Jeroen Belleman

That is indeed likely. As a rule, we tried not to rely on digital memory staying put. There was just a minimum of local storage, rewritten from a remote location every second or so. We did not use local hardware redundancy.

On the other hand, we did use Altera 7xxx CPLDs, some of which have by now accumulated about half a kGy without malfunction.

Just to set the scene, FAST TTL and LM337 regulators would stop working with more than 40 Gy or so. LSTTL and LM317s just keep going. I've used ECL in places, not because it was fast, but rather because it survived >10kGy. I used some LF351 opamps that just keep working, even after having taken well over 50 kGy! Probably they are no longer within the dataseet specs, but they are still good enough for what I'm doing with them. (DC working point feedback in a wideband amplifier.)

Jeroen Belleman