Can an FPGA be operated reliably in a car wheel?

What are the maximum g forces an FPGA could operate under? E.g. an FPGA sitting in a car wheel, where the car is travelling at 100 km/hr experiences approx 400g due to the centripetal acceleration. (assuming a 15" diameter wheel). Can anyone point to products or applications where this is done already? (I know tyre pressure sensors are coming..but these are not FPGAs)

What would the mass of your typical FPGA package/silicon be? Say a 256 ball BGA(e.g. FT256 or similar). Force = mass x acceleration so I guess you could work out the force on the solder balls etc etc.

I know vibration probably is an issue also - but at least that can be mitigated to some extent with the mechanical design of our product ,e..g vibration damping etc. However, the centripetal force is another matter - its always there whenever the wheel is rotating. (and grows with the square of the cars velocity).

Reply to
Andrew FPGA
Loading thread data ...

Your figures seem correct, close to the outside of the wheel.

The first thing you should think about is how not to have the FPGA there. Maybe you need sensors or outputs in the wheel itself, but do you really have to process the information there? Take a look inside a VCR to see how high-bandwith signals are communicated with the spinning heads.

If you can't avoid it, keep the circuit close to the axis, so the acceleration is less.

Reply to
MikeShepherd564

Thanks for pointing that out Mike, my physics were obviously a bit lacking. Redone the calculations and I see that I now get 108 g if the FPGA is located at 50mm from the axis of rotation. I need a minimum area for the electronics so a 50 mm radius gives me that.

Still, the question remains is 100g too much for an FPGA? How much g could it take reliably? I know 100g is way too much for a human and I know that some MEMS accelerometers spec up to 10,000 g for the absolute max rating (measurement up to 250g max I have seen so far).

Regards Andrew

snipped-for-privacy@bt> >...FPGA...in a car wheel...100km/hr...approx 400g...

Reply to
Andrew FPGA

Ever since WW II, >60 years ago, electronic circuits have been used in artillery shells. I have seen a whole TV camera and transmitter housed in the tip of a

155 mm shell. Those accelerations upon firing are much higher than 400g. Plastic-encapsulated conventional commercial-grade ICs are actually extremely rugged, since everything is completely encapsulated in plastic, the silicon or the wires have nowhere to go. Modern BGA packages do not even have bonding wires... I will look for acceleration test data in our reliability reports. I am sure the G-loads are not your problem. Temperature may be... Peter Alfke, Xilinx
Reply to
Peter Alfke

Ever since WW II, >60 years ago, electronic circuits have been used in artillery shells. I have seen a whole TV camera and transmitter housed in the tip of a

155 mm shell. Those accelerations upon firing are much higher than 400g. Plastic-encapsulated conventional commercial-grade ICs are actually extremely rugged, since everything is completely encapsulated in plastic, the silicon or the wires have nowhere to go. Modern BGA packages do not even have bonding wires... I will look for acceleration test data in our reliability reports. I am sure the G-loads are not your problem. Temperature may be... Peter Alfke, Xilinx
Reply to
Peter Alfke

By getting as close to bare die as is reasonable (chip scale packaging comes to mind) you can reduce the mass to the point where forces are inconsequential. When hybrid microelectronics started with all the teeny tiny assemblies using bare die, printed resistors, and other space saving techniques, it was to house electronics in artillery shells. Those were only fired over the ocean in WW II for fear the enemy might get hold of an unexploded shell and discover the tiny electronics.

Work toward a smallest mass solution and you'll probably have an easy time of it. At least you don't have to worry about the electrons bunching up to one side of the chip!

Reply to
John_H

: Still, the question remains is 100g too much for an FPGA? How much g : could it take reliably? I know 100g is way too much for a human and I : know that some MEMS accelerometers spec up to 10,000 g for the absolute : max rating (measurement up to 250g max I have seen so far).

All I have is a gut feeling that the FPGA will take more gs than the PCB it's mounted on? Also consider the differential forces arising from fiting a flat PCB to a curved wheel - different areas are at different radial distances...

cds

Reply to
c d saunter

Let me guess - trying to do something like this perhaps....

formatting link

Reply to
Mike Harrison

I take you want to mount the FPGA perpendicular to the rotation axis ? Are you aware that Tin flows ? Under strain, tin flows slowly, over the days. You can see that tin wire moves by itself under gravity. So even if the pbc is able to take the force and the FPGA case takes the strain, the tin won't.

Rene

--
Ing.Buero R.Tschaggelar - http://www.ibrtses.com
& commercial newsgroups - http://www.talkto.net
Reply to
Rene Tschaggelar

Reply to
fpga_toys

Great post Rene ... in similar ways most other non-crystalized substances which are in fact either elastic or plastic, but we consider "solids" because they seem to hold their shape.

formatting link

Note "mild steel" as a plastic :)

I'd seriously wonder just how plastic/elastic the doped silicon is and the alum interconnects inside the die as well. at 400G's at lot of things flow that we consider stable solids.

Reply to
fpga_toys

I suppose this is because room temperature is near enough to the 231C melting point of tin for elevated-temperature creep to be relevant.

On the other hand, I see no mention of tin in the design!

Tom

Reply to
Thomas Womack

There are indeed a few other ways to contact a TQFP or BGA, but none as simple as just solder it onto a pcb.

Rene

--
Ing.Buero R.Tschaggelar - http://www.ibrtses.com
& commercial newsgroups - http://www.talkto.net
Reply to
Rene Tschaggelar

Hmm, I'm suspicious of this! :-) Rene, do you have any references to some data on this. I think most materials have to have a minimum amount of stress before the strain becomes non-reversible or, looking at Antti's wikipedia link, plastic. I wonder what that is for Tin? I think I'm skeptical because of the 'room-temperature-glass-flows-myth'. (No, it doesn't!) Thanks, Syms. p.s. I'll hang a piece of solder off the side of my desk see if it grows!

Reply to
Symon

I've done more research. I think you're claiming that Tin is viscoelastic. See

formatting link

However, on the UK's National Physical Laboratory, I found this link.

formatting link

This says Tin is an isotropic material with a tensile strength of 20-35 MPa, and a Young's modulus of 49.9 GPa. No mention of it being viscoeleastic.

What do you think?

Cheers, Syms.

Reply to
Symon

Ah, is this an issue with the new lead-free solders? I knew pure tin crept, but I wasn't aware that solders did.

Tom

Reply to
Thomas Womack

...and this paper talks about the elastic limit of Tin, so I guess it has one!

formatting link

I found this:-

formatting link

and on this page

formatting link
we find that Tin has a yield stress of 9-14 MPa. So, it's a simple calculation to work out if the tin solder is gonna be strong enough not to creep.

This experiment, "The creep of solder", from Cambridge university may be of interest.

formatting link

Anyway, my place & route has finished, so back to proper work!! :-)

HTH, Syms.

Reply to
Symon

On a sunny day (Mon, 8 May 2006 13:49:53 +0100) it happened "Symon" wrote in :

:-) Yes we know that.

Reply to
Jan Panteltje

Shame. Glass flows. You just don't "see" it flow so that makes it a myth for you? The plate glass in the Notre Dame cathedral has measurable differences in thickness from the top to the bottom as just one simple example.

Reply to
John_H

I know it is viscoelastic. Try to mount a tin (solder) wire of arbitrary length horizontal on one end. It'll bend down rather quick. The older lead-tin wire was quick in this respect, a matter of a few hours. The new leadfree takes a lot longer, possibly weeks.

Another indicator is that you never(!) put a tinned copper wire into a screw terminal. It'll lead to a failure over time (years). The solder creeps under the pressure and the contact becomes worse, the contact resistance becomes higher and thus in a high current application, the contact becomes warmer over time until the contact fails. Thus, always use untinned wire in a screw terminal.

Rene

Reply to
Rene Tschaggelar

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.