Que Sara, Que Sara
Whatever will be, will be.I wrote the VHDL code for the SARA+F6 chip in about an hour, so it would have been a miracle if it worked right. No surprise, simulation revealed a problem. I got so hung up on getting the timing right for indexed writes to SARA RAM that I forgot all about A7 (which determines the state of /WE.) Fortunately it was a single line of code, and after I fixed it, the simulation looked more promising.The simulation here is supposed to hit a couple of bankswitching hotspots. These seem fine, given my assumption of capacitors on A3 and A0 to delay them a tiny bit.The issue is getting the correct /CE of the RAM and ROM and getting /WE to hit at the right times. The testbench is intended to simulate this, though I could have gotten something wrong:1300: STA $1020,x ; x=$20 STA $1060 LDA $1090Given proper timing on RC circuits connected to three ports on "timer2" the /WE has a single rising edge during STA $1060, and two rising edges during STA $1020. Plus /WE stays high during LDA $1090. I might not have the address bus states totally correct but I think it's correct enough.Also, I have one free macrocell on the 22V10, so I might be able to do F4+SARA as an option. Of course, F8+SARA would work too by just commenting out some code...Next steps: design RC circuits for approx. 600ns, 1.1 us and 1.5 us delays, burn the 22v10 and build a prototype (then pray that I don't need to drag out a logic analyzer.)
9 Comments
Recommended Comments