Hi everyone From the point of view of EMI (or any type of interference), are FPGA devices so relyable as ASIC or industrial micro? I was thinking in case like this: Due to any interference, a small part of the FPGA is deconfigured and affect in any manner to the global working. Is thi possible Thank
If you are talking flash based FPGA's such as Altera MaxII parts then no, however, if you are using the larger Cyclone/Stratix I&II parts which boot from external flash then yes, it is possible and has been known to occur in severe EMI environments. However the Altera parts have provision in the embedded chip hardware to continuously check the contents of the internal configuration via a CRC which is loaded on boot. If the continuous background check fails an immediate auto reboot from external flash can be initiated.
Don't know about Xilinx parts, I expect, as always, they are perfect in every detail - in theory.
The concern is with the metal interconnect on the die, as there is nothing we can do for the metal wires on the pcb.
For wiring on the pcb, voltages of 1 volt might disturb an IO signal (on any chip).
For wiring inside the FPGA, a voltage of a few hundred millivolts couls also upset the operation of the chip (change a logic value of a signal).
If the longest internal wire for the largest device is 2 centimeters long, what is the potential that is imposed on it by your EMI/RFI environment?
E&M Field theory equations are used to solve this.
Look at as one turn transformer with one turn being partial, 2cm long, located the distance that is to be simulated away from your other turn which creates the magnetic field....two square loops coupling can be used to estimate, dividing the results by 4 for just the partial turn...
It is likely the wiring on the pcb is the first to fail, as these wires are much longer than even the longest wires on die. But, if you have considered the pcb wiring, and kept that very short, then you must also consider the on die wiring.
Placing the pcb orthogonal to the fields will go a very long way in mitigation for both the pcb, and the die (as the wiring is in the same plane for both).
Places people are concerned about this include: train motor controls, MRI scanners, and any other environment where very large magnetic field transients will occur.
We've had a discussion today about quite the same question at the university. We concluded, that FPGAs are not less safer than an industrial computer with the same housing. Of course not as safe as an ASIC device, but still, it is a very good alternative for industrial computers/PLCs/etc. Good luck