Using the dual channel feature on the ADS4226EVM

Lim Euncheon

Other Parts Discussed in Thread: ADS4226EVM, ADS4226, ADS6149

Hi,

I found this answer is useful.

http://e2e.ti.com/support/data_converters/high_speed_data_converters/f/68/p/2719/9350.aspx#9350

I changed the code to think simply about this problem. In addition, I did not change the code looked nicely. This is working code for dual channel data acquisition but not the final for ADS4226EVM.

---- makeCapture_ADS4226.m

clear all;

fs = 120;
fin = 1;
N = 32768;
coherentFin;
fs = fs*1e6;
fin = fin*1e6;
fprintf('\n\nCoherent Input Frequency is %8.2f Hz\n\n', fin);
adcName = 'ads62p2x';
adcChnl = 'AB';
% Initialize and connect the serial port
initSerial;

% make a capture
writeSerial_ADS4226(objSerial1, adcName, N, adcChnl);
ADCout = readSerial_ADS4226(objSerial1, adcName);
leng = size(ADCout)%!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
ADCoutA = ADCout(1:N);%!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
ADCoutB = ADCout(leng-(N-1):leng);

delSerial;
Bits = 12;
figure(1)
adc_performance2(Bits, fs/1e6, fin/1e6, 1, double(ADCoutA), 1)
figure(2)
adc_performance2(Bits, fs/1e6, fin/1e6, 1, double(ADCoutB), 1)

---- adc_performance2.m

function adc_performance2(NumBits, Fs, Fin, FundaBins, ADCData, useWindow)

x = 1:1000;
y = ADCData(x);
plot(x, y)

Then, I get above result.

I wonder whether the values larger than 3500 is complying with the design intention of FPGA-firmware or not.

If it does not comply, could you please revise the code a little?

I expect readSerial_ADS4226 file should be modified to solve this problem since it seems a 2's complement number problem.

Regards,

Lim, Euncheon

over 12 years ago

0 Richard Prentice over 12 years ago

TI__Mastermind 29220 points

Hi,

The EVM has a jumper selection for 2's complement data format or offset binary. The Matlab code was written with the expectation of offset binary, but the EVM may have been set up for 2's complement. The difference between the two formats is that the most significant bit is inverted. To get from offset binary data to 2's complement simply invert the most significant bit. To convert the other way, invert the msb. The easiest thing to do is change the EVM to offset binary by moving a jumper (JP14) to offset binary LVDS. Changing the Matlab code to expect 2's complement would involve inverting the msb of either the 12 bit or 14 data and then sign-extending through the rest of the 16 bit word - that could be done too.

Regards,

Richard P.

0 Lim Euncheon over 12 years ago in reply to Richard Prentice

Prodigy 160 points

Hi,

Moving a jumper (JP14) to offset binary LVDS does not give me the desired result.

The result is the same with both jumper setting:

3214
   3430
   3711
   4033
    267
    572
    820
    991
   1066
   1037
    907
    691
    410
     88
   3855
   3550
   3301
   3131
   3057
   3085

I expected, if I short JP19 to 2's complement LVDS position, that the result shows a sine wave but it was not.

The sampling resolution of ADS4226EVM is 12 bits. Therefore, 0 to 4096 values can be detected.

4033 and 267 are shown in the figure and 4033 should be less than 267 or 267 should be higher than 4033.

In readSerial_ADS4226.m, there are following statements:

d = fread(obj,obj.BytesAvailable/2,'uint16');
d = swapbytes(uint16(d));

        switch lower(adcName)
            case 'ads6149'
                d = bitshift(d,-2);
                bMask = hex2dec('3fff');

...

d = bitand(d,bMask);

Should I change this part to remove discontinuity or did I make a mistake in jumper setting?

Regards,

Euncheon

0 Lim Euncheon over 12 years ago in reply to Lim Euncheon

Prodigy 160 points

I solved this proplem with the code modification.

Thanks.

Regards,

Euncheon

Data converters

Data converters forum

Using the dual channel feature on the ADS4226EVM