OT Rosetta arrives at comet

[Wh1sper]

Uses a stack and RPN arithmetic logic like the programmable HP calculators of that period. 

My long standing favorite Language!
(ago)

[miguelvp]

I'm pretty sure compilers always generate RPN code when they encounter formulas, it wouldn't make sense to do it a different way.

[jaxbird]

I find it incredibly impressive to have a space craft doing a 10 year journey to orbit and possibly land on a comet, while the comet is travelling at incredible speeds around our sun!...

Truly amazing in my book, I'd rather we spent a much larger budged on science and space exploration that currently. It's the future, it's worth the money!

[bwat]

I'm pretty sure compilers always generate RPN code when they encounter formulas, it wouldn't make sense to do it a different way.

This isn't really applicable in this case. The microprocessor in question has a stack architecture - it's essentially a FORTH microprocessor - it's ALU instructions don't take specific operands as the operands are on the data stack. The machine basically has a type of FORTH as it's assembly mnemonics. This is from Koopman's book I linked to earlier and it has the instruction set:  http://users.ece.cmu.edu/~koopman/stack_computers/sec4_5.html.

As for compilers in general, there's no need to generate RPN code if you've got a typical accumulator architecture or a general purpose register architecture. Say you had a typical RISC machine with 31/32 registers, then you would do this:

Code: [Select]

adi r3, r0, 99      ; r0 contains 0, so r3 <= 99
adi r4, r0, 98      ; r0 contains 0, so r4 <= 98
add r5, r3, r4      ; r5 <= 99+98
and not something equivalent to this which needs a data stack and potentially memory accesses if the stack is not on the chip:

Code: [Select]

LIT 99
LIT 98
PLUS

[EDIT: I've realised that when you write RPN you're meaning the order of evaluation, i.e. evaluate the operands then perform the operation. When I read RPN I think stack plus zero operand instructions. Actually, it's syntax not semantics. Anyway, if I'm right in my assumption about what your interpretation is, then you're correct as the arithmetic operation will have its operands evaluated before the operation takes place - even in a non-strict functional programming language.]

[miguelvp]

I was thinking more of floating point processors, where you do fld to load the values and then do the operation. All the ones I've seen are stack based.

[Alexei.Polkhanov]

I look at few pages of that book - only ones shared by Google and what surprised me, somewhat is that book describes use of components in BGA (Ball Grid Array) packages. I was under impression that BGA would be too risky and not reliable enough for space applications comparable to one of those gold lead ceramic QFP packages. Don't want to get red circle of death on your $1B space toy.

[T3sl4co1l]

It's my understanding, BGA is more reliable per-pin than QFPs.  And you can get a hell of a lot more pins down...

They also use leaded solder, not the stuff prone to cracking.  Though the higher strength of RoHS processes tends to lead to higher reliability anyway; it seems tin whiskers were much ado about nothing after all.

Tim

[Alexei.Polkhanov]

Well it is strange. Why spend ll effort to make components in "military packages". I mean those pinkish ceramic QFPs with gold leads, extended temperature ranges etc. and then use BGAs. I am not an expert in designing hardware for space applications but there is definitely huge temp variations, absence of air/gas that allow for heat transfer, vibration during launch that can knock your teeth out and then on top of that radiation. And yet we have BGA-related problems in room environments.

I have heard (or watched in some video) that sometimes BGA component sort of under-filled with epoxy. It have to be very thin runny epoxy to be able to squeeze it in between soldered balls. I have seen it done on one board from cable modem. Is this something I can buy from DigiKey? Just curious.

[KJDS]

Every BGA will have been xrayed from a few different angles. It becomes very obvious if a joint isn't good.

[T3sl4co1l]

Well it is strange. Why spend ll effort to make components in "military packages". I mean those pinkish ceramic QFPs with gold leads, extended temperature ranges etc. and then use BGAs. I am not an expert in designing hardware for space applications but there is definitely huge temp variations, absence of air/gas that allow for heat transfer, vibration during launch that can knock your teeth out and then on top of that radiation. And yet we have BGA-related problems in room environments.

A shittily engineered product is a shittily engineered product... what do you expect?

Examples of consumer use:
Plop down a footprint either from the manufacturer's suggestion, or IPC guidelines.
Solder in standard reflow.
Ship product.

Whereas aerospace may go back all the way to the footprint to optimize geometry for their process; the board material TCE can be matched to that of the device (I've seen it in IPC books but never IRL: there are PCB materials with rigid metal supporting cores to control CTE and such -- this isn't your average aluminum PCB), and the solder alloy, flux and reflow profile can all be optimized for best reliability.  Test data is generated from accelerated life and vibration testing, presumably with a margin beyond the (already severe) stresses of launch or whatever.

Tim

[TerraHertz]

Meanwhile, more pics:

http://www.esa.int/Our_Activities/Space_Science/Rosetta
http://www.esa.int/Our_Activities/Space_Science/Rosetta/Highlights/Postcards_from_Rosetta     Latest images

My favorite so far:


Considering the comet is about 4Km in some dimension, and that view is kind of looking 'down' towards the center of gravity, it gives me that aching-fingertips climbing risk-exposure feeling. At least if you fell there, you'd have a long time to wonder how much your impact was going to hurt. Also, you'd get a number of preview impacts on the way down.

Thanks for the link to the Forth-CPU book.

[firewalker]

One day to go.

Rosetta will deploy Philae on November 12 at 12:35 AM PST. Confirmation of the landing will arrive at ground stations around 8:03 AM PST.

Alexander.

[German_EE]

16:00 UTC or 17:00 CET for those of us in Europe (Mission Control at ESOC are about thirty minutes walk away from me).

[JoeN]

The lander (Philae) uses the radiation-hardened Harris RTX2010 microprocessor made by Intersil.
http://www.researchgate.net/publication/234265387_Embedded_Computer_System_on_the_Rosetta_Spacecraft

NRND!

http://www.intersil.com/en/products/space-and-harsh-environment/rad-hard-digital/rh-microprocessors-and-peripherals/HS-RTX2010RH.html

 

[Kjelt]

Considering the comet is about 4Km in some dimension, and that view is kind of looking 'down' towards the center of gravity, it gives me that aching-fingertips climbing risk-exposure feeling. At least if you fell there, you'd have a long time to wonder how much your impact was going to hurt. Also, you'd get a number of preview impacts on the way down.

So what is the gravity on that thing? Will it attract enough to keep on standing on it or if you jump you will float into space?

[Frost]

So what is the gravity on that thing?

The gavity 1:100.000 compared with the earth.
The weight of the lander Philae is 100kg here on earth
but it only 1g when Philae lands on the comet.

Will it attract enough to keep on standing on it or if you jump you will float into space?

No, it depends on the unknown underground if its to dense
the lander might jump back into space.
But they use a gas jet to force Philae to stay on the ground
as long as they need to apply the harpoons and the self-drilling
anchors.

[EEVblog]

Aussie systems engineer gives it a 50% chance of touchdown success
http://www.abc.net.au/science/articles/2014/11/10/4123194.htm

[DJohn]

I was thinking more of floating point processors, where you do fld to load the values and then do the operation. All the ones I've seen are stack based.

Sorry, this is an old post, but I've only just seen it.  The only stack-based floating point I know is the old x87.  It still survives in modern x86/x64 processors, but compilers will try to use the register-based SSE instructions instead (for single precision at least, which is all I usually care about).

Floating point in everything else I've used (MIPS, PPC, ARM, 68K, vector processors from various Playstations, GPUs) has been register based.

[Kjelt]

Aussie systems engineer gives it a 50% chance of touchdown success

In the newspaper today it said 25% chance on failure esp. possible tumbling of the Philae lander, that would be 75% chance on success.
Well in 8 hours we will know, keep our fingers crossed.
Impression of how it should look best case

[Homer J Simpson]

Live landing coverage.

http://www.space.com/17933-nasa-television-webcasts-live-space-tv.html

Amazing stuff.

[Kjelt]

or here:
http://rosetta.esa.int/
still 15 minutes to wait till we know if it succeeded or not

[KJDS]

cheers in the control room

[Kjelt]

Yeah  they look happy still waiting official words

[Kjelt]

they are receiving telemetry data, if it would have bounced off this would not be the case (what I can understand from the discussion).

[bookaboo]

Thanks for link, all looking good.
"Sitting on surface, talking to us"