why condition become true in if/esle ?

Hi, the below simple code is making trouble that : here Vxp and Vxn both are same value(32.000); so the difference must be 0 but here program still enter in if condition. instead it should enter in else condition. why it's happening ? anyone have the idea what cause make condition false ? even I check values by insterting break point. however same code in another matlab function block is running fine. could anyone help in this regard ?

for kbit = 1:Nbit

if Vxp - Vxn > 0

B(kbit) = 1;

Vxp = Vxp - Vref*2^(-kbit);

else

B(kbit) = 0;

Vxn = Vxn - Vref*2^(-kbit);

end

end

Thanks

Answer 1

[madhan ravi]

see https://www.mathworks.com/matlabcentral/answers/429435-matlab-value-error-while-creating-vector#answer_346543 for explanation

if abs(Vxp-Vxn)<1e-4

[Sarfaraz Ahmed]

Thanks ravi. Actually, I cannot set abs(Vxp-Vxn)<1e-4 because I have differennt values of Vxp and Vxn in every edge of clock through sampler. so there are also some negative values. this is one or few case in which I am getting similar values. I can list my code down here. Please have a look and suggest how can I do this ? so that other values may not get disturb.

function [y,y1] = ADC(Vin, Vip)

coder.extrinsic('stem');

coder.extrinsic('get_param')

sim_t=get_param('ADC_Sign_E_ref_2','SimulationTime')

persistent simt;

if isempty(simt)

simt=0;

end

simt=sim_t;

Nbit = 7;

Vref = 64;

% generating empty plot

ax1=subplot(2,2,([3,4]));

ax2=subplot(2,2,([3,4]));

axis ([ax1 ax2], [0 0.2 -64 64]);

title('Dout Scaler Value');

xlabel('Time(s)');

ylabel('Amplitude');

hold on;

persistent Dout;

if isempty(Dout)

Dout = zeros(1,2);

end

persistent B;

if isempty(B)

B = zeros(1,Nbit);

end

% Conventional Set-and-Down SAR ADC

% 64C, 32C, 16C, 8C, 4C, 2C, C, C

Vxp = Vip;

Vxn = Vin;

for kbit = 1:Nbit

if Vxp-Vxn>0

B(kbit) = 1;

Vxp = Vxp - Vref*2^(-kbit);

else

B(kbit) = 0;

Vxn = Vxn - Vref*2^(-kbit);

end

end

Dout = B(1)*64 + B(2)*32 + B(3)*16 + B(4)*8 + B(5)*4 + B(6)*2 + B(7)*1 -64 +0.5 ;

stem(sim_t,Dout);

hold on;

y = Dout;

y1=simt;

disp('test');

Please help in this regard. Thanks

[madhan ravi]

the above tolerance condition will satisfy regardless of the values

[Sarfaraz Ahmed]

Thanks ravi. It's again enerting in if condition. I see seepratly the difference between similar value. in blockA the difference of two similar values is (Vxp-Vxn= -1.42e-14) and in blockB (Vxp-Vxn= -4.533e-12) therefore, when it execute blockB the program does not enter in if condition (which is correct) and produce fine Dout. but in blockA it enters in if condition. I dont know how to handle this problem even both block executing same code.

Thanks

[madhan ravi]

-1.42e-14 keep tolerance -1e-16 for -4.533e-12 keep -1e-16

[Sarfaraz Ahmed]

Thank you. I made typing mistake it basically (Vxp-Vxn= 1.42e-14) so I keep tolarance 1.42e-14. I only keep tolarance when both values are equal otherwise I have to omit this tolarance as it makes affect on further calculation in the loop.

[madhan ravi]

Anytime :)

Answer 2

[Stephen23]

"...here Vxp and Vxn both are same value(32.000); so the difference must be 0..."

Evidently not true.

"why it's happening ? anyone have the idea what cause make condition false ?"

Sure, read these:

https://www.mathworks.com/matlabcentral/answers/57444-faq-why-is-0-3-0-2-0-1-not-equal-to-zero

https://www.mathworks.com/matlabcentral/answers/316889-why-two-equal-numbers-are-not-equal

http://www.mathworks.com/matlabcentral/answers/102419-how-do-i-determine-if-the-error-in-my-answer-is-the-result-of-round-off-error-or-a-bug

http://www.mathworks.com/matlabcentral/answers/140656-about-numerical-difficulty-of-equality-comparison-operation-between-two-double-real-numbers-in-matlab

This is worth reading as well:

https://docs.oracle.com/cd/E19957-01/806-3568/ncg_goldberg.html

http://faculty.tarleton.edu/agapie/documents/cs_343_arch/papers/1991_Goldberg_FloatingPoint.pdf

All programmers need to understand that calculations on floating point numbers can accumulate floating point errors (like your example does), and do not expect floating point mathematics to be equivalent to the symbolic maths that they learned in high school (e.g. operations on floating point values are not commutative). They write their code accordingly, by comparing the difference of floating point values against a tolerance:

abs(A-B)<tol

[Sarfaraz Ahmed]

Thanks Stephen. yes, I know this abs(A-B)<tol even I used this expression in my other program but here problem is the Vxp and Vxn values are changes everytime also there are positive, negative both values so I can't set this expression othewise other values might get disturb. these are few cases in which I get similar values. please can you check above listed code ? Thanks

[Stephen23]

"Vxp and Vxn values are changes everytime also there are positive, negative both values so I can't set this expression othewise other values might get disturb"

This makes no sense to me. Why does it matter if variable values change: that is the whole point of variables! Why would using variables in some operation "disturb" some other values? Why does it matter if A and/or B are negative or positive? Nothing you wrote in your comment seems to have any relationhsip to my answer or the code that I gave you.

Please explain your comment clearly.

[Sarfaraz Ahmed]

Alright. let me explain bit more.

The below code is same in 2 different blocks. in blockA the difference (Vxp-Vxn= 1.42e-14) which is greater then 0 that's why it enter in if condition. in blockB (Vxp-Vxn= -4.533e-12) which is less than 0 so it doesn't allow to enter in if condition (which is true). I am expecting same thing in blockA but it's not happening.

in the case blockA Dout= -0.5

in the case blockB Dout= 0.5

now let's consider by setting tolarance in blockA only because blockA is not giving correct value

abs(Vxp-Vxn) < 1.3e-14

by setting this in blockA, my Dout value totally disturbed and Dout = -63.5 which should not happen. if you see in the for loop below it always check and decrease Vxn and Vxp acording to the code logic below.

The dout values of both blockj should be identicle and it must be equal to 0.5 (which blockB is ging already).

function [y,y1] = ADC(Vin, Vip)

Vxp = Vip;

Vxn = Vin;

for kbit = 1:Nbit

pp=abs(Vxp-Vxn);

if abs(Vxp-Vxn) < 1.3e-14

B(kbit) = 1;

Vxp = Vxp - Vref*2^(-kbit);

else

B(kbit) = 0;

Vxn = Vxn - Vref*2^(-kbit);

end

end

Dout = B(1)*64 + B(2)*32 + B(3)*16 + B(4)*8 + B(5)*4 + B(6)*2 + B(7)*1 -64 +0.5 ;

May be this info is enough to explain. I accept your tolarance expression and concept but it's making affect on futher calculation in the loop. so what I does, I only take tolarance when both values are similar otherwise no toalarnce.

Thanks Stephen.

[Stephen23]

@Sarfaraz Ahmed: your code has a few interesting features, e.g.:

• you have an Nbit variable to select the number of ADC bits, but then you hardcoded B(7), B(6), etc. So when I tried your code with Nbits=4, it threw an error.
• you enlarge B on each loop iteration. It would be more efficient to preallocate this before the loop.
• No code comments, no help, no input/outut specifications, not description, no named algorithm.

I am not completely sure what your algorithm is. Can you please give a link that explains the algorithm you are trying to implement.

[Sarfaraz Ahmed]

Thanks Stephen. Yes I can list here code inside that block : here Vin and Vip are different discrete values (sometime similar values) as I mentioned above. The problem is, same code is running in another blockB and that block is giving correct value when both values are similar. but bllockA is not giving correct value because it enters in if when both values are similar. if you say, I can give you my simulink design and you can check in the model by inserting breakpoint at the 2nd rising edge of clock.

%blockA

function [y,y1] = ADC(Vin, Vip)

coder.extrinsic('stem');

coder.extrinsic('get_param')

sim_t=get_param('ADC_Sign_E_ref_2','SimulationTime')

persistent simt;

if isempty(simt)

simt=0;

end

simt=sim_t;

Nbit = 7;

Vref = 64;

% generating empty plot

ax1=subplot(2,2,([3,4]));

ax2=subplot(2,2,([3,4]));

axis ([ax1 ax2], [0 0.2 -64 64]);

title('Dout Scaler Value');

xlabel('Time(s)');

ylabel('Amplitude');

hold on;

persistent Dout;

if isempty(Dout)

Dout = zeros(1,2);

end

persistent B;

if isempty(B)

B = zeros(1,Nbit);

end

% Conventional Set-and-Down SAR ADC

% 64C, 32C, 16C, 8C, 4C, 2C, C, C

Vxp = Vip;

Vxn = Vin;

if(Vxp-Vxn <=1.4211e-14)

x=-1;

else

x=Vxp-Vxn;

end

for kbit = 1:Nbit

if x > 0

B(kbit) = 1;

Vxp = Vxp - Vref*2^(-kbit);

else

B(kbit) = 0;

Vxn = Vxn - Vref*2^(-kbit);

end

x=Vxp-Vxn;

end

Dout = B(1)*64 + B(2)*32 + B(3)*16 + B(4)*8 + B(5)*4 + B(6)*2 + B(7)*1 -64 +0.5 ;

stem(sim_t,Dout);

hold on;

y = Dout;

y1=simt;

disp('test');

is this information is sufficient ? Thank you.