Hacker News Reader

Comment by dmix | Hacker News Reader

> We're not going to have bits of data. We're going to have gigabits of data," said Philip Baldwin, acting director of the network services division at JPL. "I don't want 1080p for video resolution. I want 8K video."

Is this a reason why India's moon craft had such low resolution images? I checked and the rover only has 2x 1 megapixel cameras for some 3d stuff. The pictures from the lander were the usual low res generic stuff we've always gotten.

Why cant they put a 4k lense on these things? That has to have scientific value even for looking at their own rovers/landers for defects.

View on HN

Our connectivity to mars is on the order of 12kbps, presumably better for the moon.

My other question would be whether this is due to the need to radiation harden stuff? We were developing for what effectively was equivalent to an arduino in terms of processing power due to the need for radiation hardening (and other legacy / slow moving reasons).

> Our connectivity to mars is on the order of 12kbps

It depends on the spacecraft. DSN is currently talking with the Mars Reconnaissance Orbiter at 1 Mbit/s (https://eyes.nasa.gov/dsn/dsn.html, select MRO, click "more detail" in the bottom-left).

Wow, >46 minutes latency. IDK what I expected.

It's not bad seeing as Mars is 225 million to 401 million Km away depending on where it is in its orbit.

46 Light Minutes is pretty much the maximum, that's 827,427,184 Kilometers round trip which is a bit beyond the maximum distance to Mars so there's some processing delays thrown in there.

We can't really get rid of the latency but we can increase the bandwidth by throwing a bunch of higher-power communication satellites on Mars.

We can certainly push more data from Mars, but the entire point of the article is that we probably won't have the bandwidth to receive it.

Unless we can convince congress and NASA execs to increase the budget and get more dishes built pronto.

Yep. We could cover Mar's sky with something similar to SpaceX's Starlink (high availability from any location on Mars) but with large data buffers and transmission systems that can pump the goods all the back to Earth.

Three geostationary satellites should do fine for the number of data consumers on Mars.

Except for anything at the pole.

But the problem remains the link from Mars to earth, not from the surface to orbit.

s/geostationary/areostationary

Nah I don't want to. "Geo" is used generically often enough and I'd rather use a generic term than a mars-specific term.

NAME Voyager 2

RANGE 20.00 billion km

ROUND-TRIP LIGHT TIME 1.54 days

DATA RATE 160.0 b/sec

Amazing.

> whether this is due to the need to radiation harden stuff?

It's been a very long time since I read about it so I cannot provide my source but yes that is likely part of it. There's also the resilience of the components to trauma to consider.

One of the key techniques for radiation hardening is big transistor geometries. Big transistors are less susceptiple to SEU (single event upset) caused by cosmic rays [0].

Unfortunately big transistors are old transistors. Modern 7 nm geometries are much too small to be rad-hard. Rad-hardness needs big 20-year-old geometries or even older. That also implies slower clock speeds.

[0] https://en.m.wikipedia.org/wiki/Single-event_upset

Surely redundant transistors would also be a useful technique. E.g. have an array of transistors at the current geometry, all doing the same function. The chance of a single event upsetting them all becomes unlikely.

Yes, the strategy IIRC SpaceX's Dragon uses is to have 3 more modern CPUs tied together, such that they're all running the same operation and checking each other. It mitigates errors without as much of a performance penalty.

The problem with newer sensors and such is that they're not designed to handle the stress of going into space as well withstanding the greatly increased radiation exposure on Mars and the Moon. The missions to send stuff up into space is already expensive and so you'll have to reduce the chances of the gear you're sending up of failing. Unless you can afford to keep making trips to Mars until you get something there that works and then stays working long enough to do viable work.

That seems so counter productive

The experimental helicopter on Mars might start to change the equation though -- it's built with many commercial grade components, instead of fancier things, and has worked well above expectations. The CPU is just a regular Snapdragon, like in phones.

https://spectrum.ieee.org/nasa-designed-perseverance-helicop...

It's all about bandwidth. Pictures are nice but command and control are vital, so more bandwidth is given over to that.

The bandwidth depends upon the power levels, antenna size, signal strength and the noise level. See the Shannon-Hartley theorem for details.

This means from a small lander on the moon all the way back to earth the bandwidth is small. Sure put a big camera on there, but it'll take weeks to send one picture in its entirety. The Indian lander only has like 2 weeks of life total.

Sure but I would assume they would eventually have some type of relay, like lunar gateway. Allow the landers to sip power and send to the relay, probably more a series of satellites in more eccentric orbits, and let those use their bigger solar panels and higher power to transmit/recieve back to ground stations.

I mean, how would we communicate with Mars colonies when the earth and mars are at opposite sides of the sun without some series of satellites in a elliptical orbit possibly at angle perpendicular to our general orbital planes?

> Why cant they put a 4k lense on these things

Bigger focal plane means more power, more data processing/storage means more power. More power means more solar arrays/batteries. More solar arrays/batteries means more mass. More mass means larger actuators to move it around. Larger actuators means more power...all of the above means more money.

I can't give a specific reason why, but there are likely cascading effects and system impacts to the mission. Its not like larger space rated focal planes, larger data storage and high powered amplifiers to get more data down don't exist. But at some point you do just have to call it "good enough" or "meets requirements" to avoid the system spiraling out of control.

4K cameras are tiny, and you could only turn them on for a brief moment to take a picture.

That doesn't make sense. Even if the speed is low right now, I don't see why you'd gimp the hardware. You can take higher res pictures and transfer them over longer periods of time.

While in the article said "8k video", I suspect he meant 8k live streaming. Which means right now or lose it forever. You can do hybrid if you want. Local cache on hardware that records on 8k, and streams live to 1k, so bandwidth is not exceeded.

You might run out of local disk space while waiting for whatever is there to transfer. Assuming you are continuing to take more pictures/videos and want to save them until they can be transferred.

I think disk space is cheap as well.

Radiation hard electronics are not.

And you don't have to send everything high quality. But at least some shiny pictures would be nice.

I don't buy that when you can buy 32gb of storage on a card the size of your pinkie nail for under $10.

Sure but what does it cost for 32gb of space rated, radiation hardened memory?

Maybe even in triplicate for a consensus system!

I've got a quote in front of me right now for a ~1 TB data recorder for just shy of $4 million.

Also keep in mind this isn't even for the highest class parts, this is more in line for something like a NASA Explorers program level [0]. Something you would get for a JWST or an Artemis level of program would be even more $$$

https://explorers.gsfc.nasa.gov/index.html

That card would be dead (non responsive or short circuited) by the time it made it to Mars. Rough rule of thumb I have heard is add 2 zeros to the cost of every IC for space grade stuff.

You're forgetting just how inhospitable space is to consumer grade electronics. If you watch non-live videos from the space station or check out some of their pictures you'll see loads of dead or stuck pixels on the cameras. That's from radiation damaging the components and those are essentially still completely protected by Earth's magnetic field. As you venture beyond LEO the situation gets much worse and an unlucky hit could kill your $10 storage ending your $74 million dollar Moon mission.

I'm not, I assume anything like that will need to be wrapped in sheets of lead. But modern storage densities are so high there's no need to run out.

Lead wrapping thick enough to sufficiently protect against radiation damage is very, very heavy. Beyond LEO, excessive weight is still ridiculously expensive.

My point being that it's the hardening that's expensive, not the storage itself. I don't think you need that much lead to shield increasingly tiny storage media.

Chandrayaan-3's life expectancy is 2 weeks. You could probably say the same thing about every aspect.

And how reliable will that storage be when it is bombarded with all of the radiation out atmosphere is currently shielding it from

Hardware has to withstand the elements, including the radiation. On Earth we have the atmosphere to absorb a substantial amount compared to on Mars and the Moon.

A big problem with computing in space is the radiation environment. To be quite frank, it's absurdly hostile. IIRC the voyagers have very little computing power (so, few things that can malfunction) and even then they run three processors that are already radiation-hardened, and each computation is run in parallel on each processor, and the outcomes are compared among the three to detect and throw away any computations that were altered by radiation-based bit flips.

Such is space. Everything needs to be designed specially for it, especially for trips beyond LEO (where the magnetosphere is still providing you significant protection).

The limit seems to be not the cameras, but the downlink.

You can trade off resolution for time, though. Surely high resolution photos would be useful for some purposes, even if it means you only get a few of them in a day.

Seems like even that would come at the cost of other data that may be more critical (sensors, telemetry, etc).

But if the network is as a whole pushing full capacity, spreading out the load won't help because there's barely any slack to spread it into. You could get 4x resolution, but you'll only get 1/4th the amount of footage.

Let's face it, we don't really need 24fps+ footage from the moon. Nothing is changing on a timescale that would be interesting. Even if the cameras are on a rover, take high quality panoramic 360 stereoscopic images every time the rover moves to a new destination. During the course of the next move, it can be uploading that data.

But who are we kidding? A group of internet forum nerds coming up with something that the people at NASA/JPL/ESA/etc haven't already considered?

> Let's face it, we don't really need 24fps+ footage from the moon. Nothing is changing on a timescale that would be interesting.

NASA's funding problems would compound if they took an attitude this naïve during what is largely a PR mission.

I'm naive? What do you expect to see? Just take a still and apply some noise to it, and it'll look "just like" video. I don't know what you're on about, but there's no people in these images. If these were manned landings with expeditions, then great, let's have some video. But that's not what we're talking about. And you know this

There are plans to send people back to the moon.

I think they're pushing for funding to upgrade the bandwidth. More channels, more data we can get at once.

I always wondered the same thing. But my theory wasn’t bandwidth, it was power, cost and (perhaps) weight.

I had no idea that we still monitored these things from the ground.

The power/weight thing is largely solved thanks to cell phones.

But they probably can't just buy it from industry, it has to be custom made for the project, with some gov-heavy regulation no doubt.

But surely they could do far better than 1 megapixel.

So you take the miniaturization gains from cell phones, and then replace the size/weight by encasing it within a sphere of water to protect it.

do you think the wavelengths used by phones are going to work through space and the atmosphere, are they focused on short length transmissions (higher freqs). The answer appears to be yes ;-)

wavelength is just a requirement for an antenna which doesn't need to be shielded. i wasn't specifically saying use COTS phone chipsets, but the fact that the miniaturization is now possible. however, the little helicopter on Mars running Snapdragon chip (irc?) shows that COTS is possible

The helicopter has a Snapdragon, and then IIRC one of the experiments on the rover has two Intel Atoms, plus using ethernet to talk to the cameras.

You'll need a ton (well, probably several tons) of water.

https://space.stackexchange.com/questions/1336/what-thicknes...

^^ Interesting read on this very topic.

Lots of space stuff is 10-20 years behind what we use on the ground because of radiation hardening. Both because volume is so low that radiation hardened electronics have a very slow update schedule, and because older technology is inherently more radiation tolerant (higher voltages, larger process nodes, etc).

The ISS has off-the-shelf notebooks, but apart from that the uptake of regular non-hardened hardware is relatively recent and quite slow. I think the Falcon 9 has a couple non-hardened systems, and of course Ingenuity has lots of regular off-the-shelf hardware. That was one of the big reasons the predicted lifespan of Ingenuity was so low, but software mitigations (like the ability to reboot mid-flight) have proven very effective.

Reply on news.ycombinator.com