Electronics on space probes

[m1omg]

While I know something about the electronics/onboard computers of newer probes (like that MESSENGER uses 4 single-board radiation resistant RAD6000 computer based on IBM's POWER1 CPU clocked 25 Mhz, and saves instrument and picture data on 2 solid state recorders 1 GB each),but I am very curious about what did they used on probes and satelites in the 50s,60s,70s and 80s.All that I know is that Voyager probes used digital tape recorders to save data and that even probes from the middle to late 60s used digital cameras but how it was possible to save and transmit these data to Earth or even send command to the probes with 60s technology?I know that there were computers in the 60s but I think that the smallest usable digital computers back then were as large as a refrignator and they had only about 1-4 KBs of memory, certainly not enough for photographs.Also I am interested about how are these probes commanded from Earth.By command, graphical interface, automatized processed or what?Sorry maybe it's a stupid question but I really have NO idea.

thanks in advance for answers

[Larry Jacks]

Some of the earliest space probes used film cameras. They then processed the film and transmitted it in a manner similar to a fax machine. This was long before digital cameras existed.

Another technique used on some later vehicles was called "slow scan TV". Basically, a TV camera would take a single image and transmit it, often very slowly. It was state of the art for quite some time. IIRC, the Voyager landers used SSTV techniques.

Transistors were invented in 1947 or 48 but it took a while for them to be widely used. I'd have to do some investigation but it wouldn't surprise me if some very early satellites actually used vacuum tubes. Tubes are larger and use more power than transistors but they were proven technology and much more radiation resistant.

In the early days, space probes and satellites were much similier than today's and required much more ground support. The first military satellite that I know of to have a microprocessor on board was the DSCS-III. The first one was launched on October 30, 1982. Before that, ground commands were loaded into registers and simplier circuits such as timers were used to execute the commands.

If you want to learn more about computers used in spacecraft, here are a couple good links:

Computers in Spaceflight: The NASA Experience
Part I : Manned Spacecraft Computers
Part II : Computers On Board Unmanned Spacecraft

[m1omg]

What Voyager landers?As far as I know there were no Voyager landers, just 2 flyby probes and a cancelled Mars probe program in the planning stage.
And I think that vacuum tubes on spacecraft = pure impossibility because you must save ENERGY for the instruments, correct me if I am wrong pls.
But I found this,yes,Voyager flyby probes used digital storage but Vidicon SSTV cameras ;
http://en.wikipedia.org/wiki/Voyager...cecraft_design
And thx for links, they are interesting.

[Van Rijn]

I expect he was referring to the Viking lander scanning cameras. From there:

The Viking Lander cameras were facsimile scanning systems. They had a mirror that rotated around a horizontal axis to scan about 100 degrees in elevation. They also rotated around a vertical axis to scan nearly 360 degrees of azimuth. Note that a scan line refers a vertical line of data collected at a given camera azimuth.

Larry, I've had a quick read of the second article (don't have time for the first one right now). Excellent article! Lots of interesting information there. Thanks for that link.

[01101001]

What Voyager landers?

Perhaps the Mars Voyager Landers?

They were canceled.

Mars Mission Launch Sequence

Cancelled Missions which have been studied, scheduled and partially funded:
1970s
Voyager Mars Orbiters and Landers (Nasa). Two orbiters and landers to be launched on one single Saturn V.

[Van Rijn]

And I think that vacuum tubes on spacecraft = pure impossibility because you must save ENERGY for the instruments.

Energy use is based on design requirements. If nothing else, a good number of spacecraft had vidicon tubes. I'm pretty sure the Soviets did more than that, but I don't have references right now.

[m1omg]

Also, why are the radiation hardened processors (or miniaturised single board computers) so astronomically more expansive than their common counterparts?The state of art RAD6000 and RAD750 are basically just radiation resistant versions of POWER1 (a simplified but much higher clocked version PowerPC 601 was used in the mid90s Apple computers) and PowerPC 750 aka G3 used in popular iMac models from late 90s and early 2000s, respectively, with added memory.I understand that the radiation hardened chip would probably cost much but not several hundred thosands to million times as much, it is really THAT hard to make a chip resistant to radiation?!I've read on wikipedia that quality control and manifacturing in small quantities contrubutes to the price, but I would imagine military having a big demand for these kinds of CPUs (in case of a nuclear explosion data will remain intact outside the blast radius but inside of radiation effects and it can withstand an extreme temperature range compared to an ordinary chip, so it can even survive explosions and catastrophical overheatings or freezing in cold climate so the critical data can be recovered even from wreck) so I don't understand why it is manifactures only in very small amounts and it is 100000x's more expansive than ordinary chip thx in advance for explanation.

[drainbread]

Also, why are the radiation hardened processors (or miniaturised single board computers) so astronomically more expansive than their common counterparts?The state of art RAD6000 and RAD750 are basically just radiation resistant versions of POWER1 (a simplified but much higher clocked version PowerPC 601 was used in the mid90s Apple computers) and PowerPC 750 aka G3 used in popular iMac models from late 90s and early 2000s, respectively, with added memory.I understand that the radiation hardened chip would probably cost much but not several hundred thosands to million times as much, it is really THAT hard to make a chip resistant to radiation?!I've read on wikipedia that quality control and manifacturing in small quantities contrubutes to the price, but I would imagine military having a big demand for these kinds of CPUs (in case of a nuclear explosion data will remain intact outside the blast radius but inside of radiation effects and it can withstand an extreme temperature range compared to an ordinary chip, so it can even survive explosions and catastrophical overheatings or freezing in cold climate so the critical data can be recovered even from wreck) so I don't understand why it is manifactures only in very small amounts and it is 100000x's more expansive than ordinary chip thx in advance for explanation.

What you have to understand is that the parts go through an extended engineering stage and I'm talking about everything, not just the processor.

All parts must be able to survive launch, have radiation shielding, be able to radiate enough heat in a vacuum to function and have enough computing power to do the job.

The computing power is the easy one to solve, you don't need a powerful processor due to low level programming languages but such a low power processor would be pretty much useless for anything else...

Forget about military applications.

[Larry Jacks]

What Voyager landers?As far as I know there were no Voyager landers, just 2 flyby probes and a cancelled Mars probe program in the planning stage.

That's what is known as a brainfart. I have a terrible memory for names and simply got it wrong.

As for vacuum tubes being used in space, I know that Traveling Wave Tube Amplifiers were being used on communications satellites well into the 1990s and perhaps to this day. Vidicon tubes were used for TV broadcasts from space for a long time. It's also possible that some very early satellites like Sputnik 1 used vacuum tubes instead of transistors. This article suggests I'm right.

The radios on board Sputnik are described as D-200 units and were designed by a member of Korolev’s design team named V. I. Lappo. The meaning of the D-200 designation is unclear and our research thus far has failed to produce a schematic of this transmitter, but Tikhonravov, in a presentation before the 24th International Astronautical Congress in 1973, characterized the transmitters as “vacuum valve-type” with a power of 1 watt. Figure 3 shows the transmitter unit mounted adjacent to the antenna connections in the front casing half. One transmitter operated on a frequency of 20.005 MHz (megacycles in 1957) and the other on 40.002 MHz. The choice of these frequencies not only allowed reception by amateurs using existing equipment but also enabled a receiver set at exactly 20 or 40 MHz to produce an audio tone plus or minus the Doppler shift without ever going through zero Hz. This insured that the telemetry was audible throughout an entire pass without additional tuning of the receiver.

You have to remember the state of the art in electronics in the late 1950s. Vacuum tubes had been in use for decades. They were well understood and highly developed. Transistors were still pretty new. Many of the transistors available at that time didn't work well at high frequencies like those used in satellites. Even a high percentage of the computers being used at the time still depended on vacuum tubes.

[m1omg]

What you have to understand is that the parts go through an extended engineering stage and I'm talking about everything, not just the processor.

All parts must be able to survive launch, have radiation shielding, be able to radiate enough heat in a vacuum to function and have enough computing power to do the job.

The computing power is the easy one to solve, you don't need a powerful processor due to low level programming languages but such a low power processor would be pretty much useless for anything else...

Forget about military applications.

forget?these cpus were state of art around decade before and software fir them was mostly programmed in assemblers, you need fancy cpus only for bloated "os" like windows and fancy 3d games, but some artists still record their albums on original Amiga 500s with 8 mhz 16-bit cpus and you canh ndo image editing in them so i dont see a reason for whhy they shouldnt be used in military aplicatiions, even vacuum tubes analog computers and 4-bit integrated circuits were used for aircraft simulations etv. so i dont see a problem, my "modern" two core mac is actually slower in tasks like text editing... due to bloated "new" os that has functionns that you'll use maybe once a lifetime than my good ol'
386 DX 40 Mhz from 1992 with DOS that was starting programs immediatelly when I pressed enter in norton commander and it booted for 6 seconds

the most insulting myth from the money hoarding pc industry is that you need at least 800 mhz cpu for editing text and anything lower is unusable while i was playing Diablo on 486 66 mhz and text can be done on some 8-bit micro or PC/XT

amiga had great multimedial and graphic capabilities in 8-megahertz and i am pretty sure that military dont need sophisticated guis or 4-channel stereo sound and 4096 color displays, all of what was already included in amiga and pc started approaching it oonly in mid 90s

even eniac was build for military and that computer was several hundred million times less powerful than POWER1 25 Mhz
and military needs components that would survive nuclear co9nflict, even the granddady of the internet ARPAnet was concieved as a form of network of computers that would be able to survive nuclear conflict

that something is "obsolete" by current popular pc "standarts" (while you can do most of your tasks on a 200 mhz machine including lo res video if you don't play the "lastest and greatest" mindless fps shooters) doesn't mean that it really is

i have read somewhere here that astronaust on the shuttle are actually still using old pentium 166 laptops for most of their computer work and it works perfectly

you can simplisticaly edit video on a 25 mhz mk68040 mac performa and POWER cpus are much much more powerful

[Swift]

Also, why are the radiation hardened processors (or miniaturised single board computers) so astronomically more expansive than their common counterparts?
....
I've read on wikipedia that quality control and manifacturing in small quantities contrubutes to the price, but I would imagine military having a big demand for these kinds of CPUs

I think the quantities involved is exactly the reason. The profits in computer chips are in the quantitites. What is a "big" demand for the military and NASA - a thousand chips? When you look at civilian demand, not just for computers, but cell phones, GPS, all kinds of smart devices, you are talking 10s of miilions of CPUs.

There also are higher manufacturing costs. For many years I worked for a company that grows quartz crystals. Quartz is used for its piezoelectric properties and is the "timing" part of the electronics - clocks, freqency control, etc. We sold quartz for radiation hard applications (you have to make everything rad hard, not just the CPU). The quartz that we sold for this had to be particularly high purity crystals and was post-growth processed to remove other impurities. It ended up increasing our costs by, IIRC, a factor of 5 or more.

[m1omg]

I think the quantities involved is exactly the reason. The profits in computer chips are in the quantitites. What is a "big" demand for the military and NASA - a thousand chips? When you look at civilian demand, not just for computers, but cell phones, GPS, all kinds of smart devices, you are talking 10s of miilions of CPUs.

There also are higher manufacturing costs. For many years I worked for a company that grows quartz crystals. Quartz is used for its piezoelectric properties and is the "timing" part of the electronics - clocks, freqency control, etc. We sold quartz for radiation hard applications (you have to make everything rad hard, not just the CPU). The quartz that we sold for this had to be particularly high purity crystals and was post-growth processed to remove other impurities. It ended up increasing our costs by, IIRC, a factor of 5 or more.

Well, I think they also would be of use in airplanes , it would increase safety much and increase demand = lower price.

5x price is another thing compared to 100000x price increase

[Larry Jacks]

Even if you used radiation hardened chips in airplanes, the market would be very small compared to general civilian usage. First of all, most airplanes don't need radiation hardened chips. My Piper Cherokee surely doesn't and neither would most of the general aviation fleet. That leaves military aircraft (probably not trainers), airliners, and perhaps business jets. Worldwide, you're talking about several thousand planes (perhaps 10,000-20,000 to estimate on the high side). That's smaller than the cell phone market at a typical state university.

I'd venture that most of the military planes that need radiation hardened chips already have them. Few airliners and business jets have them because they aren't considered necessary or cost effective. The degree of safety increase - if any - would be small compared to the price increase. That money could be better spent on other things to improve safety, such as retrofitting older airliners with Enhanced Ground Proximity Warning Systems (EGPWS) and things like that. Offhand, I don't know of a single airline accident that was traced to radiation effects but over the years there has been a lot of accidents where perfectly good planes impact terrain. This is called "Controlled Flight Into Terrain" (CFIT) and over the years this has been one of the leading causes of airline accidents. IIRC, no EGPWS equipped plane has had a CFIT accident.

[Swift]

Well, I think they also would be of use in airplanes , it would increase safety much and increase demand = lower price.

5x price is another thing compared to 100000x price increase

As Larry said, I can't imagine why you'd want rad hard electronics in an airplane.

I won't pretend to understand the whole supply chain economics here. That 5x I quoted is just for the raw material for one component. If, for example, each step in the process multiplies the cost by 5x, and then add to that the fact that you don't have the economics of scale, well, its going to very quickly add up.

[Larry Jacks]

When a microprocessor manufacturer like Intel spends several hundred million dollars (or more) developing a new processor chip, they fully expect to sell millions of the chips. The amortized R&D cost per chip might only be a few dollars. On the other hand, it can cost about the same to develop a radiation hardened microprocessor (even a simple one) with a potential market of perhaps a few hundred or a few thousand units. The R&D cost per chip is huge. There simply isn't a large market for hardened chips.

Even for the market that exists, it can be fragmented by competing chips. The microprocessor used in the DSCS-III satellite that I mentioned above was a hardened version of the Digital Equipment Corporation PDP-11 16-bit CPU. There were about a dozen DSCS-IIIs built. Each satellite had two of the chips on board. It's possible the chip was used in some other satellites but probably not very many. It wouldn't surprise me in the least if fewer than 100 of those chips were ever sold. While the basic R&D for the PDP-11 happened in the 1970s, the cost of making it radiation hardened was borne by the small number of users. Even if it only cost $10 million or so to develop a radiation version of the chip, the per chip cost would be enormous.

[m1omg]

When a microprocessor manufacturer like Intel spends several hundred million dollars (or more) developing a new processor chip, they fully expect to sell millions of the chips. The amortized R&D cost per chip might only be a few dollars. On the other hand, it can cost about the same to develop a radiation hardened microprocessor (even a simple one) with a potential market of perhaps a few hundred or a few thousand units. The R&D cost per chip is huge. There simply isn't a large market for hardened chips.

Even for the market that exists, it can be fragmented by competing chips. The microprocessor used in the DSCS-III satellite that I mentioned above was a hardened version of the Digital Equipment Corporation PDP-11 16-bit CPU. There were about a dozen DSCS-IIIs built. Each satellite had two of the chips on board. It's possible the chip was used in some other satellites but probably not very many. It wouldn't surprise me in the least if fewer than 100 of those chips were ever sold. While the basic R&D for the PDP-11 happened in the 1970s, the cost of making it radiation hardened was borne by the small number of users. Even if it only cost $10 million or so to develop a radiation version of the chip, the per chip cost would be enormous.

Oh,thank you, now I understand

[billslugg]

I believe a RAD6000 runs somewhere about $300,000. A big part of the cost is in the documentation. If a component fails they want to be able to go back and find out what day it was made, who made it, who painted the numbers on it, what the assay results were from the batch that the cadmium plating came from, what mine the copper originated, what the mean and std dev of component values from that batch was. They will get a 10 foot shelf of documentation books with the thing.

[samkent]

What about putting lesser chips/boards in a rad hardened case?

[Larry Jacks]

Here's an article that discusses the challenges of using commercial off-the-shelf components in a high radiation environment. Given the economic issues, engineers are trying to find ways to use commerical products. It isn't easy.

One of the biggest problems is that as the chips become more dense due to the evolution of sub-micron technology, they become more vulnerable to random bit-flips (called Single Event Upsets) due to highly penetrating gamma and cosmic rays. Alpha and Beta particles are pretty easy to shield against but cosmic and gamma rays aren't.

The degree of radiation hardening necessary depends on where the space system will be operating. It takes more shielding to operate out at geosynch than in LEO because GEO is beyond the Van Allen radiation belts. Likewise, operating around Jupiter and perhaps the other gas giants has an even higher radiation environment. Radiation shielding adds weight to the vehicle that it can ill affort. You might use tricks like mounting the electronics behind the propellant tanks to provide a measure of shielding but that can cause problems, too. It's a difficult challenge.

A good discussion of hardening challenges is at the bottom of this article.

[novaderrik]

how powerful is the processor on the Hubble? isn't it something like 66mhz- which would have been beyond state of the art when it was launched.

[01101001]

how powerful is the processor on the Hubble? isn't it something like 66mhz

New York Times: On Second Outing, Astronauts Continue Upgrade of Telescope (December 1999)

While the new computer is a big improvement for Hubble, having 20 times the speed and 6 times the memory of the old one, it is hardly state-of-the-art by current personal computer standards.

Hubble's new brain is powered by an Intel 486 DX2 chip running at a speed of 25 megahertz, a processor long obsolete in desktop computers now driven by newer Pentium chips.

[Van Rijn]

Also, from here:

http://sm3a.gsfc.nasa.gov/frontline.html

Notes from the Hubble Project team, January 4, 1999

With the installation of the 486 Advanced Computer during SM-3A, the DF-224 flight software project has come to an end. Although it was a 1970s-vintage computer, the DF-224 served us very well through the years. With the addition of a 386 coprocessor and additional memory during the First Servicing Mission in 1993, the DF-224 provided flight software services to each of the other HST subsystems right up to its replacement on December 23 of last year.

Edited to add:

Odd, I think they have a date error on that post, or they were doing a bit of time travel, since the SM-3A mission happened in December '99.

[m1omg]

New York Times: On Second Outing, Astronauts Continue Upgrade of Telescope (December 1999)

I think that the original chip was the embedded variety of 386SX, 386EX.
386SX(EX) is usually clocked at 16 Mhz, I think, I was working on a computer with 386SX and it had clock speed of 16 Mhz.

The difference between 386(later designed DX to recognise it from the cut-down 386SX) and 386SX is lower clock speed (386DX is avilable at speed up to 40 Mhz from AMD) and the data bus is only 16-bit on 386SX (the ordinary 386 or 386DX has 32-bit data bus), making it a lot slower and lower cost, it was introduced after the ordinary 386.

[ASEI]

Do the single event upsets permanently damage the electronics, or just flip random memory bits? Maybe you could "harden" the processors by having N redundant processors taking identical inputs and a majority poll on what the output "should" have been. Maybe that would even afford some damage tolerance if the processor's performance begins to accumulate damage.

[joema]

...I am very curious about what did they used on probes and satellites in the 50s,60s,70s and 80s....probes from the middle to late 60s used digital cameras but how it was possible to save and transmit these data to Earth or even send command to the probes with 60s technology?...I am interested about how are these probes commanded from Earth.By command, graphical interface, automatized processed or what?...

Except for Larry's answer, most of the thread diverged from your initial question.

To reiterate, you don't need microprocessors, integrated circuits or digital cameras to control a space probe or transmit images.

How to get the image: Slow scan TV (or similar) can function using only analog circuitry. You don't need a digital imager or a frame buffer. Using a vidicon tube, you just transmit back the data as the image is scanned, one line at a time: http://en.wikipedia.org/wiki/Vidicon
http://en.wikipedia.org/wiki/Slow_scan_TV

How to get the data back: Analog telemetry has existed for over 70 years. In WWII, the German V2 rocket used a multichannel telemetry system called "Messina" to transmit flight data during tests.

Also during WWII, remotely-controlled, TV-guided glide bombs were developed, all with analog circuitry. http://en.wikipedia.org/wiki/GB-4 So the basic technology of two-way telemetry (control and data) existed long before the first space probe was launched.

How to control it: early space problems (e.g, Ranger) used simple sequencers. In effect a "washing machine timer". At pre-determined intervals, commands would execute -- take picture, transmit data, etc.

Later probes used more advanced versions of this. Using registers and counters made from discrete components (individual transistors, not ICs), the probe would execute certain commands at specific times. Modified commands could be uploaded to the probe. These consisted of a few numbers, typically toggled into a console using rotary switches in base 8 (octal) and transmitted to the probe. http://history.nasa.gov/computers/Ch5-2.html

I believe Mariner IV (which returned the first close picture of Mars in 1965) had no on-board computer of any type. Rather it used a simple sequencer. However instead of blindly executing x operation at y time, these were variables which could be modified from earth by telemetric command.

I think Mariner VI in 1969 was the first space probe to have a digital computer. The microprocessor didn't exist, Integrated Circuits were in their infancy. I'm sure it was built using discrete components. The memory capacity was 128 words of 22 bits each, probably magnetic core: http://en.wikipedia.org/wiki/Core_memory

Mariner VI also used an on-board analog tape recorder to store TV images for later transmission to earth. This was needed since the transmission rate varied from 66 bits/sec up to 16,200 bits/sec. Even at the highest rate, it would take a long time to send a single image.

[Swift]

Do the single event upsets permanently damage the electronics, or just flip random memory bits?

I believe both can happen.

[billslugg]

Except for Larry's answer, most of the thread diverged from your initial question.

On 12-Mar-08 at 7:29 PM, m1omg appended to his original post the question:
"so, why are the radiation hardened processors (or miniaturised single board computers) so astronomically more expansive than their common counterparts?"
The divergence was at the request of the original poster.