Pricing for rad-hard parts

Joerg, I have no experience..(Well, the guy down the hall used a photodiode to detect x-rays at an FEL*) But I'd guess a lot depends on the type of radiation. Outer space vs, next to some hot source.

George H. (I'm thinking big chunks of Si would be better. 'Shielding' sometimes makes radiation worse.)

*IIRCC there was a stack of lead bricks in front of the PD. The right amount of lead soaked up the electrons and other stuff but let enough of the x-rays through... I'm not sure. (?)

I used to work designing satellites 20 years ago and our golden rule was about 100 to 1000 times the price of a standard component

But It may have changed since then

Cheers

Klaus

IIRC you're usually better off with SOI parts, because the high energy even ts that caise the problems also generate most of their carriers deep in the silicon.

Cheers

Phil Hobbs

The lab I worked in screened ICs for gamma radiation. The reactor charged $1000 a day. Then there's what we charged for the testing.

I can get it done for you, if you need.

ents that caise the problems also generate most of their carriers deep in t he silicon.

OK that makes sense too. George H.

That can be tough when you need ICs. The smorgasbord of available devices shrinks from half a gazillion to a small bowl when in need of rad-hard. It can make some designs challenging even if you are McGyver.

--
Regards, Joerg 

http://www.analogconsultants.com/

Not much. I just have to know whether it's 100x or 1000x :-)

Once when designing a hi-rel circuit I was agonizing over the cost of a part. Another engineer saw me staring at the schematic. "Anything wrong?" ... "No, but this little part here costs over 6 bucks" ... roaring laughter ... "To us that is the equivalent of a penny".

--
Regards, Joerg 

http://www.analogconsultants.com/

Thanks, I have stored your post in the project folder because at some point we may need such a service.

--
Regards, Joerg 

http://www.analogconsultants.com/

When we were first doing the GameCube CPU (a modified PPC750 in SOI) design one of the guys had to design a radiation detector to disable the chip so it couldn't be used for military applications. Before we actually had to make the part the ITAR rules changed so a video game was no longer a weapon of war.

Hi Joerg, Can you share more details on your application?

What is the maximum TID? Do you have extreme temperature cycles?

When you buy a Radhard part you solve only half of the problem (TID).

Please have a look here:

The above articles may give you a starting point for your own research.

Just to give you an idea of the price... a 1kohm 0603 resistor for space application may cost 5 GBP.

Cheers, Francesco

Hi Joerg, I forgot to say, shielding only works for alpha and Beta particles.... (just read some of the comment above). Before starting your design I suggest understanding the radiation environmental (probably your worst issue are Gamma ray) alpha and Beta are usually low energy particles.

Cheers, Francesco

Then I would get shot :-)

Essentially scientific expeditions into space.

The ionization dose is still under discussion. If you go from 100krad to

300krad the number of parts shrinks substantially, if 1Mrad you might as well do it all in discretes.

Thanks, that will probably come in handy.

Ah, thanks, now we are getting somewhere. I wonder what a gate driver or a power transistor would cost. Last time I had to do this was decades ago and I don't remember prices.

This is good to know. I may have to hand off design work at some point because of a rather full plate and of a plan to retire (which some clients didn't agree with ...). I assume you guys have design experience with space-rated electronics.

--
Regards, Joerg 

http://www.analogconsultants.com/

What kind of orbit ? LEO, Polar, GTO, GSO or outer space. GTO and Molnya type orbits are particularly nasty, since the satellite will cross the Van Allen radiation belts twice each orbit or 4 times a day.

Many amateur radio AMSAT satellites are built using essentially off the shelf parts use Molnya type orbits. Some subsystems fail early in the mission, while other operates for decades. Just do not use any newly introduced components, since there are no long time radiation history.

I don't know yet. They will have to decide on the level of rad-hardness and then I'll have to see if the circuitry is even economically feasible with that, given component availability.

In my case early failure wouldn't be cool. If someone uses non rad-hard parts for this they should prefer older devices with larger geometries (node). Not any of the super high density nanometer chips.

--
Regards, Joerg 

http://www.analogconsultants.com/

Hi Joerg, I forgot to say that if the application is CubeSat, then you don't need to buy ESA qualified parts. Things to considers are :

1. mechanical stress due to 25 thermal cycle/day (select a PCB substrate with a similar mechanical thermal coefficient of the IC package)
2. ceramic capacitors get cracked easily. Use large pads if assembled by hands
3. in your analysis consider SEE not just on digital circuits also in analog circuits (e.g. BJT, MOS, the final stage of an OP AMP, diodes reversed biased etc.)

Good luck with your project.

Francesco

[...]

Good points. Thanks, Francesco.

Ceramic caps are also a concern in aircraft electronics which I am familiar with.

1. in your analysis consider SEE not just on

Definitely. Analog is often where the most risk is because there may be no possible recovery from a single event. It can trigger something that shouldn't have been triggered and the computer can't take it back.

In case anyone is interested but not familiar with SEE yet here is a good concise explanation (middle slides of the presentation):

Thanks. It feels like when I got back on a mountain bike after not having ridden for 15 years. One can get hurt badly on those and I had my share of "involuntary high-speed dismounts".

--
Regards, Joerg 

http://www.analogconsultants.com/

Aerospace Corporation does testing using the Berkeley cyclotron, firing oxygen and other ions at the part. They have a list of several thousand parts and the sensitivity of each before upset and failure. I imagine you have to pay to get the detailed info. They might tell you for free if a specific part has test data already done.

Jon

Am 25.03.2018 um 16:21 schrieb Joerg:

You do not have much influence on the pad forms and sizes. That is determined by the people who build the flight boards and their processes. I had the special luck that my boards were made on two different processes by different companies, partly by hand and partly vapor phase solder.

And a resistor / cap may be changed only once. 3 times hot abs. max. is the rule and even that requires paperwork through the hierarchy. If you expect tuning, use multiple parallel footprints.

A small number of part types is better than a small number of parts (in limits) because of the qualification overhead. They might prefer 2 * 1K in par. instead of a new 499R.

I got protests because I wanted to reserve one larger and one smaller from the nominal value for the pulling inductor of a crystal. Two extra part types. But I had to insist. +-700 Hz pulling range at 100 MHz with ultimate phase noise specs and flight varicaps never seen before.

At least they allowed me to plug the crystals on the network analyzer and choose the one of the 5 that I liked best. In the end, it worked without change. What a relief!

Yes, we have talked here already about voltage regulators that can get SEUs.

We had one CPU that had to be commercial. That required an entire protection board with sensor & switches just to be save. This orbit would be lost.

We also had ram-based FPGAs with scrubbing in the configuration ram and a lot of triple module redundancy. I wrote a nice library that looks like ieee standard logic / vector and has all the redundancy nearly completely hidden. There is an alternative tool from Xilinx but that counts as a weapon.

Cheers, Gerhard

Sometimes there is influence.

Yes, that may indeed be necessary. Also because of time schedules.

If these varicaps had to go through testing and cert I can understand their reluctance.

Dodging a stack of paperwork is always nice. I just received info about a form I have to file for other reasons with one of those ridiculous "Paperwork reduction act" notices at the end. That law has turned into a sad joke. Estimated time to learn the regulations before even filling this out is 42h. Hours!

And then this portection board needs to consist of 100 certified components.

Luckily I have managed to largely stay out of the digital world so far and I don't plan to change that :-)

--
Regards, Joerg 

http://www.analogconsultants.com/