The good news from the press conference is that the low-power fault state is basically the same thing as deep sleep, which they use all the time.

I don't know two things: first, how long it has until it loses its mission clock, which complicates the situation, and second, how the stuck-on IDD heater that deep sleep was added to get around will affect the recovery -- since that's a 0.5A load that will start as soon as the battery controller comes up.

ICYMI : Here is a link to the YouTube recording of the 'Dust Storm' teleconference: https://youtube.com/watch?v=fIKxdRFx2Wo#

The extent of this dust storm seems to be emulating the 1971 Mariner global dust storm. That storm lasted over three months so we could be in for a long wait as the atmospheric transfer between poles slows down and the dust settles.

Another potential alternative is the 1977 dust storms observed by Viking. There was an early season storm that followed the Acidalia storm track (started in Mare Acidalium, crossed south across Chryse and Thaumasia, and exploded in the southern hemisphere) that cleared by mid-summer, followed by an even more intense storm in late summer originating from the Hellas Basin.

If it gets no power from the solar panels, is there a ball-park figure on how long the pre-existing battery charge could keep the clock going? Days or weeks?

[EDIT: Based on figures here:
https://trs.jpl.nasa.gov/bitstream/handle/2...400/05-3884.pdf
and some assumptions, my own rough estimate is 8 days if zero power from the solar panels.]

Because a lot of questions being asked here were answered in the press briefing, I'm going to break my usual practice and link to my own writing, in this case a single-page version of all my live tweets of the press briefing.

John Callas said "If rover is generating less than 22Wh, then it won't have enough power to maintain clock". I'm not sure how to parse this. AFAIK, the mission clock is powered directly from the batteries during sleep and will presumably drain them down to some minimum voltage cutoff. 22Wh per sol would be a little under 1 watt of power, which is a heck of a high power draw for a simple clock.

At any rate, I think it's safe to assume that over the next few days there will be essentially no power generated.

The main thought I keeping coming back to is that Oppy has been out of contact for much longer stretches of time during conjunctions (though obviously this is a very different circumstance in other respects!) We know the rover won't suffer some cold-related issue as Spirit did; it's just a matter of crossing our fingers and waiting it out...

https://trs.jpl.nasa.gov/handle/2014/43244 "The effects of clock drift on the Mars Exploration Rovers" is an interesting paper about the MER clock architecture. It didn't really have anything germane to the issue of losing time reference but it has a lot of detail about how the mission clock works.

During conjunction the rover has still been powered, of course. The most worrisome thing about this is whether there's some issue associated with losing the mission clock (go back through all the Spirit status reports after loss of comm for lots of discussion about various permutations there). We can assume that Spirit just got too cold, but there's no proof of that I'm aware of. And then there's the possibility that the panels will be so dusty after the storm clears that they won't produce a useful amount of power (although I think that's probably unlikely.)

He said that below 22Wh they would likely have a clock fault, so the fact that 22 was the last number reported from the rover made him think it very likely that they had already triggered that fault.

If the dust will settle uniformly, the panels will be just as dusty as the sky will have been during the storm. So, a cleaning event will be required after or near the end of the dusty sky phase.

Not strictly true, at least if you're referring to how dusty the local skies were. The very high opacity the rover last saw is quite localized. The planet-encircling dust event, by its nature, distributes dust geographically. When local lifting stops, Meridiani could see rapid clearing that is not dust settling on panels. Then, a long slow period of sedimentation will get things dusty, but I imagine the Meridiani winds will not let too much dust accumulate (Mars may imagine differently, of course).

New MARCI weather report:

http://www.msss.com/msss_images/latest_weather.html

It includes a link to this "Storm Watch" website:

https://mars.nasa.gov/weather/storm-watch-2018/

To add a layman's context to the amount of sunlight blocked from the dust storm, if I was standing next to Opportunity, how dark would it be to my human eyes? (Bright as a full moon on Earth? for example)

There are a lot of variables to consider there Steve and the human eye has remarkable adaptability to light intensity. To put this in context the difference in intensity between a really bright sunlit day in a high albedo environment and a moonlit night is around 9 or 10 million to 1. The eye doesn't mind whether it gets its lumens by the bucket load or with a tea spoon provided it gets enough. But since the impact of tau on solar intensity is a negative exponential function with a tau of 10.8+ you probably wouldn't be able to see your hand in front of your face.

OK, so according to the source linked below, a sunny day on earth is about 100,000 lux, while a moonlit night is about 0.25 lux.

At Mars distance the sun will be about 1/4 brightness compared to Earth, so say 25000 lux under clear skies. A tau of 10 means a reduction of about 50,000 fold in the light intensity, so I would estimate about 0.5 lux. This is comparable to a moonlit night on Earth.

https://answers.yahoo.com/question/index?qi...mp;guccounter=1

If I understand it correctly, tau measurements are performed by considering only the sun or another small apparent source of light.
https://agupubs.onlinelibrary.wiley.com/doi...29/1998JE900017
Scattered light seems to be neglected. This approach tells something about the total incoming light, if tau is small. But for large values of tau, most of the light is scattered. So, it doesn't need to be much darker than it would be, if the whole sky would be as bright as the moon on Earth.

More useful, in this case, would probably be the comparison of about 700 Wh with 22 Wh, which is about a factor of 34 darker than under clear sunshine. Reduce this factor a bit by the dust layer on the panels. It may still be bright enough to read newspapers easily, like on a cloudy or very cloudy day on Earth.

Exactly! The Sun itself was fainter than a full moon on Earth on Sunday. But in each photon-dust interaction the extinction (which tau measures) is <10% absorption and >90% scattering to sky light. The scattering is ~85% downward, 15% upward. There would be 10-11 extinction events (more like 14-15 at the time Opportunity was awake, with the Sun not vertical). Modeling literally the round numbers I just used, I get 98% loss of light at the surface (i.e., sky light >10,000 times direct sun light); you got 97% (and I believe the panels more than made up numbers).

What did it look like? Other than the decidedly ruddy to brownish tint, it looked like a somewhat overcast day. The sort of mid-afternoon when you can just barely see (or not quite see) the Sun through clouds that stretch from horizon to horizon. When the world would be black through eclipse glasses, but easily visible without. Normal human vision would adapt well--it's not night, moonless or moonlit, or even sunset. Solar panels, however, are not known for being so dark-adaptable. So the drama is real, even if people seem to want to describe it in terms that would be as dramatic for people as the actual situation is dramatic for the rover.

This weekend, I expect tau of several hundred. Not on Mars, but at home, where we expect thunderstorms.

PS, scattered light is not 'neglected'. In photometrically measuring opacity it is hunted and eliminated with extreme prejudice.

As a data point, I have solar panels on my house. On a typical day in March they generate about 30 kW-h. During the cloudiest, lowest-production day this March, I got about 3 kW-h, about 10x worse.

700/22 is 32x less production, so this would be a really cloudy day, at least by SoCal standards. (Solar distance is already scaled out so don't complain about that.)

Another example that people may have witnessed is being under heavy smoke from forest fires. The sun may be just barely visible as a dull orange disk (so a huge tau), but the landscape will be as bright as some level of cloudy day due to light scattered from the sky (smoke).

I don't know how smoke particle size compares with Martian dust, which would affect the details of the scattering (how much into what angles vs frequency), so the details may be different.

Still it's worth pointing out that so far 22 Whrs/sol is an upper limit...

On the 22 WH just to keep the clock alive, maybe that number is accounting for the stuck IDD heater?

From the paper - "The Mission Clock in the MER rovers uses a custom hybrid crystal oscillator from Q-Tech Corporation. " which however is proprietary, at least at the time the paper was published. The skew of +-10s over 10 sols with temperature is pretty reasonable for a typical +-50 ppm crystal oscillator. (~900,000s for 10 sols, so about +- 10ppm with 10s drift). To get better PPMs clocks in your phones use temperature compensation, ie they adjust the voltage based on temp, with <2 ppm or less commercially available. Of course, that costs power too (~mA), and isn't space rated. A COTS RTC (real-time clock, think wrist-watch in a tiny package) without temperature compensation can be < 1uA though, again not space rated.

just as a reminder how Mars looked in 2001 fully in dust - shots with the MGS orbiter camera:

https://photojournal.jpl.nasa.gov/catalog/pia03170

Regarding tau and the efficiency of solar panels, another issue is that Oppy's solar panels are of triple-junction type. This allows a wider spectral coverage and improved efficiency (each junction works at a different part of the spectrum). But this increased efficiency comes with a drawback as the junctions are in series. During a dust storm, the light is likely skewed toward red. Hence, not much blue is left and the blue junction becomes the limiting factor.

I don't see how, that heater is turned off when the battery controller is turned off, that being the whole point of deep sleep.

As siravan noted the spectrum at the surface would be skewed towards the red. The spectral response of Opportunities solar cells extends into the near infrared and it would not surprise if this is where the remnant charge is coming from. While comparing Earth to Mars is a bit chalk and cheese the attached link shows the effect on visible light of a dust storm in my old stamping ground. https://www.youtube.com/watch?v=BrlD22HwPvI

The descriptions of current (human) visibility on Mars as a cloudy day may be a trifle optimistic, even given the range of visibility of the human eye:

According to the paper cited here:

the stuck on heater causes a drain of 0.5 amps, but it is actually really only on when the temperature drops sufficiently. The paper estimated a cost of 180 WH per day at the time the anomaly occurred.

Assuming 24 volt battery power, 1 watt corresponds to around 40 milliamps, which doesn't seem too unreasonable for the mission clock (plus alarm mechanism).

Not sure what is meant. The chart describes a factor of ~30 below full sun (ok, a factor of 60 below given Mars-Sun distance) as 'near windows'. Moreover, overcast days on Earth literally have optical depth >10 quite frequently (well, frequency is location dependent, but I grew up in Seattle).

If blue is the limiting factor for the triple junction cells (per siravan), the light providing array energy cannot be all IR; frankly, there should be more red light available than IR (per unit wavelength) given the Sun's spectrum and the albedo of dust in the red.

Normally you don't use linear regulation from battery voltage, because most of the power is used in the regulator. 40 mA at 5V is only 200 mW (which is still a lot for a simple clock -- for example, a Chronodot RTC http://docs.macetech.com/doku.php/chronodot_v2.0 -- admittedly not rad-hard -- uses 840 nanoamps in standby timekeeping!). But it's not impossible that in this case they did use linear regulation because it's simpler and losing all solar power wasn't really a credible fault for the short mission design life.

At any rate, if the clock is lost, we may see a lot of X-band "sweep & beep" commanding once there's any expectation that the rover is getting enough power during the day to communicate. On Spirit they started this about 4 months after the loss of comm. See https://mars.nasa.gov/mer/mission/status_spiritAll_2010.html for some descriptions of this.

I did try to estimate the illumination based on the solar panel output, and I was puzzled by why that was coming out so differently from the estimate based on tau. What bothers me about the 22 Watt-hr quoted for the solar panel production is that seems way too much to just run a clock. Are we certain that number is accurate?

Also, if there was a break in the storm the rover might have gotten half an hour of sun which would charge the battery a little. We don't know how much the panels were producing at the time the tau was measured.

On a different topic, the 8 RHUs are providing 192 Watt-hours of energy in the form of heat. If I remember correctly, sunny day production from the panels is about 600 Watt-hours of electricity. So a big chunk of the energy budget of this "solar" powered rover is actually from nuclear sources, even when it is sunny. That surprised me when I worked out the numbers.

Another point is that if blue limits the power, and the sky is very red, then estimates of the overall brightness on the ground based on the 22/700 ratio will be underestimates, due to the extra red light not contributing to array power. Probably not a huge factor, though.

No, and it was a bit of an off-the-cuff answer, but barring some additional information from the project it's all we have to go on, and it's not implausible given what details of the design we do know. For example, some types of Q-Tech oscillators really do draw 40 mA just for the oscillator, and then for a complete clock there has to be a counter and maybe some other stuff -- the fault protection paper references a "mission clock FPGA".

I also don't fully understand how the scheduling of transmission and reception periods happens if the mission clock is lost. This seemed to introduce a lot of complexity into the attempted recovery process for Spirit. Presumably without the mission clock there's no way for the rover to figure when 11 LST is (in theory it could do with from solar power production but that would vary a lot based on tau, tilt, dust on the panels, etc.); AFAIK this scenario isn't described in the fault protection paper.

The rover transmits roughly around midday on Mars, so the orbiters listen then. Once the signal is received they can send commands to the rover, which listens periodically, eventually getting back into sync with Earth.

The only thing I got from the press conference is that when the rover detects sunlight it sets a timer so that it can wake up periodically to attempt to phone home. I think it is a 4 hour interval, so presumably the DSN will listen for 4 hours during daylight to see if it hears from the rover.

While the GaInP top cell is the limiting (short circuit) control its spectral response extends out to 750nm with a high response out to 600nm. So low levels of solar power will continue to be generated when much of the visible spectrum is suppressed. This link provides some information on MER solar panel performance https://ntrs.nasa.gov/archive/nasa/casi.ntr...20070010752.pdf
Absorption due to dust is highest towards the blue, decreasing to level out around 650 nm. Scattering increases with wavelength. So with the optical depths over the course of the day with tau of 10.8 it is possible that the small charging occurred around midday and if the tau has increased, has terminated.

As noted before, if it was that easy, then how do you explain the need for the "sweep&beep" campaign done for months with Spirit?

Yes, but when the rover wakes up to phone home/listen, the IDD current draw could be a factor then if the temp is low enough to turn it on uncommanded.

I guess the question is: how does the rover decide if it's midday? One idea would be exceeding some minimum power level. More elaborate would be to take several power level samples over at least a day and fit a sinusoid. If it's something like the former it may get stuck thinking it's not midday until tau drops sufficiently.

I expressed the same concern back in post #51, yes.

Perhaps, but this kind of elaborate solution that requires state to be recorded and used from sol to sol is usually not used in deep fault responses, because it's really hard to test all the possible permutations.

Not to mention the fact that flash memory is unavailable, so only measurements from a single Sol could be considered, and the CPU would have to stay on throughout the Sol.

In principle, the rover might be able to autonomously reset its clock if there was enough energy to take images of the sun position but (as the previous post points out) such complex algorithms would be unlikely in fault protection. Waking up every 4 hours is a simple and seemingly adequate solution.

The biggest danger I see from the dust storm is a poor dust factor going into the winter season leading to desperate measures. If I recall, the issue with Spirit was precipitated by over-winter issues that upset the original plan of driving along Home Plate. Hopefully we will get cleaning after the current dust storm but we may also get a second dust storm in the current summer season.

How severe is this dust storm compared with others weathered by Opportunity? Estimating the total solar radiation in heavy dust conditions is an interesting exercise and the scattered light will be the main contributor when the direct radiation diminishes. I'm trying to model all this with my simulated weather imagery package. A key factor would be how much light gets absorbed for each scattering event (a single scattering albedo of about .90 in green light and .97 in the IR see figure 8). A tau of around 10 would only cut the light down about 50% if absorption wasn't a factor. I can note a Titan analog with a tau of around 8 and estimates of 10% of the visible band light getting through when absorption is factored in. Deimos' formulation in post #28 is reasonable. As mentioned more light in the IR will help for the solar panels. Serpens' YouTube video is pretty impressive.

The worst previous was back in 2007: "Due to extensive dust storms in Mars' southern hemisphere causing record atmospheric opacity levels, Opportunity is currently experiencing its lowest power levels to date. The tau measurement as of sol 1225 is 4.12, resulting in a mere 280 watt-hours of array energy. A tau measurement of 5.0 would result in approximately 150 watt-hours."

https://mars.nasa.gov/mer/mission/status_op...07.html#sol1382

Keep in mind that we have two competing measurements, the tau determined by analysis from Pancam images, and the actual solar production. From an engineering perspective, only the second one is of direct interest. I'm sure they have a model to go from the first to the second, but once you have an actual number for the second you should use it. And of course for any model, you not only have to know the irradiance but also how much dust is on the panels.

It doesn't. Assuming a loss of clock fault - it simply boots up when 2 amps are on the array. The IDD heater is not a significant concern when you're at 2 amps.

While the flash is not available, the rover also has 11MB of EEPROM thst could be used to store state information. I think, for example, that this is where Earth position as a function of SCLK is kept. Without time reference, no HGA comm will work until this can be updated.

This may help (or not). https://www.swsc-journal.org/articles/swsc/...swsc150027.html
The maximum tau assessed by Opportunity in 2007 was 5.5. The maximum assessed this time around before she went dark was 10.8 and this may have increased. This doubling of the tau, ignoring the airmass variable means that direct insolation would be 0.005 that enjoyed by opportunity at the height of the 2007 storm.

EDIT. An update by A.J.S. Rayl. http://www.planetary.org/explore/space-top...torm-sleep.html
An interesting extract: “The dust here is thicker than anything I have ever encountered, going back to Viking missions,” said MER Deputy Principal Investigator Ray Arvidson ........It’s dark, like the end of twilight dark.”
Paolo. described it as basically "the difference between a full sunshine day and a full moon night kind of state". Even a minor increase in tau over the 10.8 value and all ambient light would go.

From the latest MRO weather report:

http://www.planetary.org/explore/space-top...mer-update.html has some discussion with John Callas about the consequences of the mission clock fault on Spirit.

It might be interesting if someone from TPS sat down with him to discuss the recovery of Opportunity in light of that experience. (Edit: I note that there's a little bit of that in http://www.planetary.org/explore/space-top...torm-sleep.html linked upthread.)

"The Martian dust storm has grown in size and is now officially a "planet-encircling" (or "global") dust event."
https://www.jpl.nasa.gov/news/news.php?feature=7164

Christopher Go is the ace of [superior planet] astrophotography and his latest Mars picture (as of now; his page updates in case anyone is reading this weeks from now) gives an outstanding idea of how devastating this storm is:

http://astro.christone.net/mars/

Looking down on what should be extremely high-contrasty terrain and seeing that blankness might communicate the intensity more clearly (no pun intended) than stats re: sunlight.