Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Problem in designing NCO using cordic algorithm

Status
Not open for further replies.

jubin007

Newbie level 4
Joined
Mar 10, 2015
Messages
7
Helped
0
Reputation
0
Reaction score
0
Trophy points
1
Activity points
60
I am pursuing M.Tech in VLSI design and have designed a pipelined Cordic based NCO but i am having glitches in my output. This is a part of my m.tech thesis and i will have to complete very soon.
It will be of great help if someone kindly helps me sort out my problem. i have attached the program and simulation result. in the wave, you will see that there is a repetitive glitch coming from my program. If you help me sort it out, it will be really helpful.
 

Attachments

  • needurgenthelpindesignofcordicbasednco.zip
    30.7 KB · Views: 4

ads-ee

Super Moderator
Staff member
Joined
Sep 10, 2013
Messages
7,820
Helped
1,811
Reputation
3,632
Reaction score
1,772
Trophy points
1,393
Location
USA
Activity points
59,030
I am pursuing M.Tech in VLSI design and have designed a pipelined Cordic based NCO but i am having glitches in my output. This is a part of my m.tech thesis and i will have to complete very soon.
It will be of great help if someone kindly helps me sort out my problem. i have attached the program and simulation result. in the wave, you will see that there is a repetitive glitch coming from my program. If you help me sort it out, it will be really helpful.

It appears you misunderstand how a Verilog for loop works...for loops are unrolled at compile time to generate multiple indexed copies of hardware.

You have written a combination of a software style for loop (as in a temporal loop) and a hardware for loop (spatial replication of the X, Y, and Z registers). This coding style doesn't work as the for loop will be unrolled and I think the only assigns that exists when simulation starts are: assign X_shr = X[17]>>>i;, assign Y_shr = Y[17]>>>i;, and assign Z_sign = Z[17][31];.

If you code your X, Y, and Z outputs with the RHS of your assigns the code will synthesize and simulate correctly as it will represent a proper set of pipelined registers.

Code Verilog - [expand]
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
// stage 1 to N-1//
  genvar i;
  generate
    for (i=0; i< 18; i=i+1)
    begin: abc
      wire Z_sign; //sign = di
      wire signed [31:0]X_shr; // shift right X
      wire signed [31:0]Y_shr; // shift right Y
      assign X_shr= X[i]>>>i;
      assign Y_shr= Y[i]>>>i;
      assign Z_sign= Z[i][31];
      always @(posedge clk)
        begin
// must use the [i] versions directly to index the correct Z, Y, and X values
      X[i+1]<= Z[i][31] ? X[i]+ (Y[i]>>>i) : X[i]- (Y[i]>>>i);
      Y[i+1]<= Z[i][31] ? Y[i]- (X[i]>>>i) : Y[i]+ (X[i]>>>i);
      Z[i+1]<= Z[i][31] ? Z[i]+ atan_table[i] : Z[i]- atan_table[i];
// can't use the continuous assign values as they will not reflect the intention
// of using the [i] versions for each of the parallel pieces of logic. I'm not
// entirely sure, which of the Z[i], X[i], and Y[i] was used for the itterations,
// but the assign was causing it to use only one of them instead of the correct
// [i] version.
//      X[i+1]<= Z_sign ? X[i]+ Y_shr : X[i]- Y_shr;
//      Y[i+1]<= Z_sign ? Y[i]- X_shr : Y[i]+ X_shr;
//      Z[i+1]<= Z_sign ? Z[i]+ atan_table[i] : Z[i]- atan_table[i];
    end
  end
  endgenerate


Here is the output with the change:
Capture.JPG

You can remove all the assigns and the declarations. You also can get rid of the generate-endgenerate if you place the for inside the always block changing the genvar i; to an integer i; declaration. Makes for nice compact clean looking code.
 
Last edited:

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Top