1
\$\begingroup\$

Im trying to capture DDR data from ADS5463 (TI ADC). As the datasheet suggested I need to delay the clock and sample the data with DRY clock. Im using Lattice FPGA LFE3-35EA.. and using Lattice High Speed I/O Interface with IPexpress: enter image description here

using Diamond I creted this suggested Interface:

enter image description here enter image description here

Afterwards I used this IP as is.. made my top module and just connected the pins, I made very few changes on the leds and inputs/outputs.

I connected the ADC to the FPGA using this pins:

USE PRIMARY NET "clk" ;
LOCATE COMP "clk" SITE "L4" ;
//FREQUENCY NET "clk" 200.000000 MHz ;

LOCATE COMP "rstn" SITE "C21" ;
LOCATE COMP "datain[0]" SITE "Y3" ;
LOCATE COMP "datain[1]" SITE "W2" ;
LOCATE COMP "datain[2]" SITE "T4" ;
LOCATE COMP "datain[3]" SITE "U1" ;
LOCATE COMP "datain[4]" SITE "P1" ;
LOCATE COMP "datain[5]" SITE "N2" ;
LOCATE COMP "datain[6]" SITE "L3" ;
LOCATE COMP "datain[7]" SITE "J2" ;
LOCATE COMP "datain[8]" SITE "G1" ;
LOCATE COMP "datain[9]" SITE "E1" ;
LOCATE COMP "datain[10]" SITE "D2" ;
LOCATE COMP "datain[11]" SITE "B1" ;

LOCATE COMP "led_clk" SITE "F19" ;

I supplied a sampling frequency of 400M (so the DRY clock is 200M as I config) and some data sine wave in diffrent frequencies:

1M:

enter image description here

enter image description here

2M: enter image description here

On one hand i capture the right frequency, but i get very noisy results while I know it shouldnt happen. I should get clean Sine wave with clean FFT.

Im adding the matlab code also:

%% Read data from files
data1 = get_data('C:\Users\---\--.txt', '%*10s%13s%[^\n\r]', 11, inf);
data2 = get_data('C:\---\--.txt', '%*23s%14s%[^\n\r]', 11, inf);

source1 = zeros;
for k1 = 1:length(data1)
    val = convertStringsToChars(data1(k1));
    val = fliplr(val);    
    val = bin2dec(val);
    source1(k1) = val;
end

source2 = zeros;
for k2 = 1:length(data2)
    val = convertStringsToChars(data2(k2));
    val = fliplr(val);    
    val = bin2dec(val);
    source2(k2) = val;
end



source = zeros;
index=1;
for k2 = 1:length(source1)
    source(index) = source1(k2);
    index = index+1;
    source(index) = source2(k2);
    index = index+1;
end


% Plot the data

figure ('Name','test', 'units','normalized','outerposition',[0 0 1 1])

range = 1:1:length(source);
subplot(2,1,1);
plot(range,source); 
axis([0 max(range) 0 4096])
title('Data - Time scale')
xlabel('Samples [n]')
ylabel('Amplitude')
grid

% FFT

Fs = 400e6;            % Sampling frequency                    
T = 1/Fs;             % Sampling period       
L = max(range);             % Length of signal
t = (0:L-1)*T;        % Time vector

Y = fft(source);
Y(1)=0;

P2 = abs(Y/L);
P1 = P2(1:L/2+1);
P1(2:end-1) = 2*P1(2:end-1);
f = Fs*(0:(L/2))/L;

subplot(2,1,2);
plot(f,P1)
axis([0 100e6 0 100])
title('Data - Spectrum')
xlabel('f (Hz)')
ylabel('|P1(f)|')

Why is that happening, does someone have an idea? Im using the Interface as Lattice suggested and Im getting bad results. Anyone who is working with Lattice FPGA i will be glad for some help.

Edit:

I adding the 1MHz data coming out from the FPGA:

 ps    posedge_data  negedge_data
 delta                          rstn
  0 +0 101111000001  101111000001  1  
  1 +0 101111000001  101111000001  1  
  2 +0 101111000001  101111000001  1  
  3 +0 101111000001  001111000001  1  
  4 +0 001111000001  101111000001  1  
  5 +0 101111000001  101111000001  1  
  6 +0 101111000001  101111000001  1  
  7 +0 001111000001  011111000001  1  
  8 +0 010111000001  010111000001  1  
  9 +0 010111000001  010111000001  1  
 10 +0 010111000001  010111000001  1  
 11 +0 110111000001  110111000001  1  
 12 +0 110111000001  110111000001  1  
 13 +0 110111000001  000111000001  1  
 14 +0 000111000001  010011000001  1  
 15 +0 000111000001  011011000001  1  
 16 +0 011011000001  111011000001  1  
 17 +0 101011000001  011011000001  1  
 18 +0 011011000001  011011000001  1  
 19 +0 010011000001  010011000001  1  
 20 +0 010011000001  000011000001  1  
 21 +0 000011000001  100101000001  1  
 22 +0 000011000001  101101000001  1  
 23 +0 101101000001  001101000001  1  
 24 +0 001101000001  001101000001  1  
 25 +0 001101000001  001101000001  1  
 26 +0 100101000001  100101000001  1  
 27 +0 000101000001  000001000001  1  
 28 +0 001001000001  001001000001  1  
 29 +0 001001000001  101001000001  1  
 30 +0 101001000001  101001000001  1  
 31 +0 101001000001  111110000001  1  
 32 +0 101001000001  111110000001  1  
 33 +0 111110000001  101110000001  1  
 34 +0 001110000001  011110000001  1  
 35 +0 010110000001  010110000001  1  
 36 +0 000110000001  010010000001  1  
 37 +0 011010000001  011010000001  1  
 38 +0 001010000001  011010000001  1  
 39 +0 001010000001  010010000001  1  
 40 +0 010010000001  110100000001  1  
 41 +0 111100000001  111100000001  1  
 42 +0 101100000001  011100000001  1  
 43 +0 010100000001  010100000001  1  
 44 +0 000100000001  010000000001  1  
 45 +0 000100000001  011000000001  1  
 46 +0 011000000001  010000000001  1  
 47 +0 010000000001  110111111110  1  
 48 +0 000000111111  111111111110  1  
 49 +0 111111111110  011111111110  1  
 50 +0 011111111110  011111111110  1  
 51 +0 110111111110  100111111110  1  
 52 +0 100111111110  010011111110  1  
 53 +0 011011111110  011011111110  1  
 54 +0 011011111110  010011111110  1  
 55 +0 010011111110  110101111110  1  
 56 +0 100011111110  111101111110  1  
 57 +0 011101111110  011101111110  1  
 58 +0 011101111110  110101111110  1  
 59 +0 110101111110  110101111110  1  
 60 +0 100101111110  111001111110  1  
 61 +0 111001111110  111001111110  1  
 62 +0 111001111110  111001111110  1  
 63 +0 110001111110  100001111110  1  
 64 +0 000001111110  010110111110  1  
 65 +0 011110111110  011110111110  1  
 66 +0 101110111110  111110111110  1  
 67 +0 001110111110  010110111110  1  
 68 +0 010110111110  010010111110  1  
 69 +0 000110111110  011010111110  1  
 70 +0 011010111110  011010111110  1  
 71 +0 001010111110  011010111110  1  
 72 +0 110010111110  110010111110  1  
 73 +0 010010111110  010100111110  1  
 74 +0 011100111110  011100111110  1  
 75 +0 011100111110  001100111110  1  
 76 +0 001100111110  011100111110  1  
 77 +0 001100111110  010100111110  1  
 78 +0 010100111110  010100111110  1  
 79 +0 010100111110  110100111110  1  
 80 +0 111000111110  111000111110  1  
 81 +0 101000111110  101000111110  1  
 82 +0 101000111110  001000111110  1  
 83 +0 001000111110  011000111110  1  
 84 +0 010000111110  110000111110  1  
 85 +0 110000111110  110000111110  1  
 86 +0 110000111110  110000111110  1  
 87 +0 100000111110  000000111110  1  
 88 +0 000000111110  110111011110  1  
 89 +0 111111011110  111111011110  1  
 90 +0 111111011110  111111011110  1  
 91 +0 111111011110  111111011110  1  
 92 +0 111111011110  111111011110  1  
 93 +0 101111011110  001111011110  1  
 94 +0 001111011110  001111011110  1  
 95 +0 001111011110  001111011110  1  
 96 +0 001111011110  001111011110  1  
 97 +0 001111011110  001111011110  1  
 98 +0 001111011110  001111011110  1  
 99 +0 001111011110  011111011110  1   

Im adding also the 2MHz data points:

 ps    posedge_data  negedge_data
 delta                          rstn
  0 +0 001100111110  001100111110  1  
  1 +0 000100111110  100000111110  1  
  2 +0 100100111110  101000111110  1  
  3 +0 101000111110  111000111110  1  
  4 +0 110000111110  100000111110  1  
  5 +0 100000111110  110111011110  1  
  6 +0 111111011110  011111011110  1  
  7 +0 011111011110  001111011110  1  
  8 +0 001111011110  011111011110  1  
  9 +0 010111011110  000111011110  1  
 10 +0 000111011110  000111011110  1  
 11 +0 000111011110  110011011110  1  
 12 +0 000111011110  111111011110  1  
 13 +0 111011011110  100111011110  1  
 14 +0 100111011110  100111011110  1  
 15 +0 100111011110  100111011110  1  
 16 +0 100111011110  100111011110  1  
 17 +0 100111011110  110111011110  1  
 18 +0 100111011110  110111011110  1  
 19 +0 110111011110  101000111110  1  
 20 +0 011111111110  000000111110  1  
 21 +0 000000111110  010000111110  1  
 22 +0 000000111110  100000111110  1  
 23 +0 110000111110  101000111110  1  
 24 +0 101000111110  111100111110  1  
 25 +0 100100111110  110100111110  1  
 26 +0 110100111110  101010111110  1  
 27 +0 011100111110  110010111110  1  
 28 +0 100010111110  110010111110  1  
 29 +0 111010111110  111110111110  1  
 30 +0 110110111110  110110111110  1  
 31 +0 110110111110  110001111110  1  
 32 +0 110001111110  010001111110  1  
 33 +0 011001111110  011101111110  1  
 34 +0 110101111110  110101111110  1  
 35 +0 111101111110  011011111110  1  
 36 +0 010011111110  110011111110  1  
 37 +0 111011111110  111111111110  1  
 38 +0 110111111110  110111111110  1  
 39 +0 111111100001  111000000001  1  
 40 +0 111000000001  011100000001  1  
 41 +0 011000000001  010100000001  1  
 42 +0 010100000001  011010000001  1  
 43 +0 011100000001  110010000001  1  
 44 +0 110010000001  111110000001  1  
 45 +0 111010000001  110110000001  1  
 46 +0 110110000001  111001000001  1  
 47 +0 011110000001  110001000001  1  
 48 +0 110001000001  111001000001  1  
 49 +0 111001000001  111101000001  1  
 50 +0 110101000001  110101000001  1  
 51 +0 011101000001  001011000001  1  
 52 +0 111101000001  100011000001  1  
 53 +0 100011000001  101011000001  1  
 54 +0 101011000001  101111000001  1  
 55 +0 101011000001  110111000001  1  
 56 +0 100111000001  100111000001  1  
 57 +0 110111000001  111111000001  1  
 58 +0 101111000001  111111000001  1  
 59 +0 111111000001  111111000001  1  
 60 +0 111111000001  111111000001  1  
 61 +0 111111100001  101000100001  1  
 62 +0 100000100001  100000100001  1  
 63 +0 100000100001  100000100001  1  
 64 +0 000000100001  100000100001  1  
 65 +0 100000100001  100000100001  1  
 66 +0 100000000001  110111000001  1  
 67 +0 111111000001  111111000001  1  
 68 +0 011111000001  001111000001  1  
 69 +0 001111000001  011111000001  1  
 70 +0 001111000001  000111000001  1  
 71 +0 010111000001  000011000001  1  
 72 +0 000111000001  001011000001  1  
 73 +0 001011000001  001011000001  1  
 74 +0 100011000001  110101000001  1  
 75 +0 111101000001  111101000001  1  
 76 +0 111101000001  110101000001  1  
 77 +0 110101000001  100001000001  1  
 78 +0 111001000001  101001000001  1  
 79 +0 001001000001  100110000001  1  
 80 +0 100001000001  101110000001  1  
 81 +0 001110000001  000010000001  1  
 82 +0 000110000001  010010000001  1  
 83 +0 010010000001  010100000001  1  
 84 +0 011100000001  011100000001  1  
 85 +0 001100000001  011000000001  1  
 86 +0 011000000001  010111111110  1  
 87 +0 010000111111  001111111110  1  
 88 +0 011111111110  101011111110  1  
 89 +0 101111111110  101011111110  1  
 90 +0 101011111110  101101111110  1  
 91 +0 101011111110  101101111110  1  
 92 +0 001101111110  101001111110  1  
 93 +0 101101111110  111001111110  1  
 94 +0 101001111110  111110111110  1  
 95 +0 111110111110  111110111110  1  
 96 +0 011110111110  110010111110  1  
 97 +0 110110111110  111010111110  1  
 98 +0 111010111110  110010111110  1  
 99 +0 010010111110  010100111110  1  

I added 100 points out of 8192 captured points. In addition I will add picture of the signal generators which I used: enter image description here

all the signals here I tried to analyze is -20dBm in diffrent frequencies.

\$\endgroup\$
10
  • \$\begingroup\$ What's your noise floor with no signal? Can you show us a few cycles of the captured waveform at each frequency? \$\endgroup\$ – Bruce Abbott Feb 27 '20 at 20:36
  • \$\begingroup\$ @BruceAbbott Test equipment won't be avaiable to me for a few days (it is the weekend here) I ran a few tests, and the noise floor is around -50dBm, while I expect to see results at around -70dBm.I see an almost flat plain at -50dBm and downwards \$\endgroup\$ – Michael Astahov Feb 27 '20 at 21:43
  • \$\begingroup\$ "Test equipment won't be available to me for a few days" - But you still have the data, right? Looking at the 1MHz wave it looks like many bits are flipping to 1 (or incorrectly detected as 1) at certain code transitions. A closer look at those codes might provide a clue to why this is happening. \$\endgroup\$ – Bruce Abbott Feb 28 '20 at 5:29
  • \$\begingroup\$ @BruceAbbott I edited the post, unfortunately I have only the 100MHz data file (I overwrite the exported data all the time), can you see there something? maybe something with the OVR signal (I exceeding the 2.2V?) \$\endgroup\$ – Michael Astahov Feb 28 '20 at 7:26
  • 1
    \$\begingroup\$ keyword: aliasing. For detailed explanation see e,g, e2e.ti.com/blogs_/archives/b/precisionhub/archive/2015/09/04/… . So to ensure proper operation one has to low pass filter the input signal of the ADC. \$\endgroup\$ – Christian B. Feb 29 '20 at 12:28
0
\$\begingroup\$

Ok i think i figure out my problem, in the Diamond analyzer i provided sample clock: buf clock and I should have provide sclk as the sample clock in the analyzer, now when Im using sclk i getting this results: enter image description here

so i get an clean output and for what I asked here the problem solved, so for who using Diamond Lattice pay attention which sample clock you provide in the Analyzer.

Thanks for all the helpers.

now the problem is that I expecting to sample an input signal of -70dBm+ and in reality i have success only with -35dBm+.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.