Hi All, I am new to this forum as well as FPGA. We have been given a problem statement from customer that a pcb with AM29240-EH microcontroller is giving frequent failure. Some ICs in the PCB is getting obsolete. We are proposing them an FPGA based solution. Please let me know if it is possible to replicate the microcontroller logic in FPGA. Any additional information will be very helpful. Regards, Muthuraman S
I guess the demanding task could be to find a VHDL/Verilog component which is 100% compatible to the AM29240. It seems to be waaaay more complex than e.g. an AVR. Even if there was a component you would have difficulties to reach the 25MHz clock speed unless you spent a really large and really expensive FPGA for it.
Fetz wrote: > Even if there was a component you would have difficulties to reach the > 25MHz clock speed unless you spent a really large and really expensive > FPGA for it. I have started a MIPS soft core. The AM29240 has a simulary structure. I can say the speed will be not the problem. 33MHz runs a real AM29240. My fist look was what is inside. The AM29240 has no Floating point Unit. The CPU is a typical Berkeley RISC-Design. It has 192Register. This is huge number for a CPU. It is lot to do. cache, DMU-Unit, external Controler, CPU, special function register..... Before fall back to this retro CPU Redesign, try to change the CPU.
> Before fall back to this retro CPU Redesign, try to change the CPU. And this also may be a additional problem not solved by changig the CPU: >>> that a pcb ... is giving frequent failure. >>> Some ICs in the PCB is getting obsolete. We had this with a lot simpler 8085 and 8251, 8253, 8255... >>> We are proposing them an FPGA based solution. And we had to keep the original software in binary format, so we had to redesign those parts of the components used by this software into a FPGA. And that was a task with several surprises, based on the asynchronous design of the original parts. Although the parts are fairly simple it took about 3 man-year to get this design into production. You also must keep the software? If no, the only reasonable way is to redesign the board with an up-to-date CPU.
Thanks for your replies. Looks like a good conversation with good datum for making decisions.
Just an idea, if you must keep the (old) software: What about using an up-to-date CPU, storing the binary of the (old) software, and emulating the old CPU? But could be a problem, if old software has timing constraints... At some point you can perhaps discard the requirement to keep the old software and write native software for it. If you need help to do this or have additional questions, please feel free to contact me.