4*6 bits register does not retain it's value EPM7064

[MSmit101]

Hi all,

I have something very strange happening. I have programmed a Altera 7000 Series EPM7064 device. In it I created 4*6 bits register. Simulation is all good. Reset, writing and reading is all good.

Only in real life, checked with my Logic Analyser, I see the write but a microsecond later the value of the register does not appear on the output pins. Instead only zeroes appear.

Here the output of the simulator:

At the bottom is the input data, all 3 lower bits high. In the middle the output of the register is written because WR and EN is low. And after writing it stays visible on the output pins [2][1][0].

Here the output of the programmed device with the logic analyser:

The data D5/D6/D7 briefly appears at D2/D3/D4 and then goes back to zero and it stays that way.

Here the Verilog code:

module scc_rom_mapper (wr_n,sltsl_n,reset_n,a15_a13_a12,data,address_upper);

input [2:0] a15_a13_a12;

input [5:0] data;

input wr_n,sltsl_n,reset_n;

output reg [5:0] address_upper = 0;

// 4*6 bits storage

reg [5:0] mem0 = 6'b000000;

reg [5:0] mem1 = 6'b000001;

reg [5:0] mem2 = 6'b000010;

reg [5:0] mem3 = 6'b000011;

wire WE;

assign WE = !reset_n || (!sltsl_n && !wr_n);

// write memory

//////////////////

always @(posedge WE) begin

if (!reset_n) begin

mem0 <= 6'b000000;

mem1 <= 6'b000001;

mem2 <= 6'b000010;

mem3 <= 6'b000011;

end

if (!sltsl_n) begin

// store only when upper 4k is selected

case(a15_a13_a12)

3'b001: mem0 <= data;

3'b011: mem1 <= data;

3'b101: mem2 <= data;

3'b111: mem3 <= data;

default: ;

endcase

end

end

// read memory

////////////////////////

always @(*) begin

// always select correct bank when reading or writing

case(a15_a13_a12[2:1])

2'b00: address_upper <= mem0;

2'b01: address_upper <= mem1;

2'b10: address_upper <= mem2;

2'b11: address_upper <= mem3;

endcase

end

endmodule

[sstrell]

It's not clear which signals are which in your simulation or your scope picture. Can you clarify?

And your coding for the "clock", WE, is strange. Do sltsl_n or wr_n toggle high/low for a write?

Also, you need an "else" on your check of sltsl_n in order to correctly code a register:

if (!reset_n) begin

.

.

.

else if (!sltsl_n)

Finally, it might be cleaner to code a 4x6 array instead of separate registers like this:

reg [5:0] mem [0:3];

#iwork4intel

[MSmit101]

In my design for an old retro computer project. I allow WRite when the SLot is SELected. I tried earlier an always (*) block with if/else statements but this wouldn't work. Hence the "clock" type of setup.

BTW. I did solve the issue at hand, being that the address_upper lines are cleared and not showing mem.

Somehow the reg[x] mem [x] initialisation was done at every evaluation. After putting this in an initial begin/end block this was solved.

So problem gone. ;-)

[SyafieqS]

Hi Mario,

What I understand the control signal in your case is sltsl_n(sel) and wr_n(wr) should be toggling. The register is written when sltsl_n and wr_n is high on same clock edge. It return the current data when sltsl_n is high and wr_n is low. Using if else statement for this process as shown below is snippet sample.

 always @ (posedge clk) begin

  if (!rstn)

   register <= 0;

  else begin

   if (sel & wr)     // when both signal is high

    register <= wdata;  // non-blocking statement wdata assigned to register

   else

    register <= register;

  end

 end