Simulink - Obtain fixed-width vector from MATLAB function block output

Question

Nicolas 2024-2-27

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/2087636-simulink-obtain-fixed-width-vector-from-matlab-function-block-output

编辑： Fangjun Jiang 2024-3-1

采纳的回答： Fangjun Jiang

在 MATLAB Online 中打开

Hello,

For context, I am using the STM32 Hardware Support Add-On and Simulink in order to deploy a model on a STM32 MCU.

After some signal processing, the model extracts 2 values (double), and I want these values to be sent through USART to another device.

This is the model I've come up with :

My goal is to generate a uint8 vector containing :

'A = ___ | B = ___'

My first idea for this seemingly trivial task was to use sprintf :

function data = fcn(valA, valB)
msg = sprintf('A = %06.2f | B = %06.2f\r\n', valA, valB);
data = uint8(msg);

I got this error message :

Error:'data' is inferred as a variable-size matrix, but its size is specified as inherited or fixed. Verify 'data' is defined in terms of non-tunable parameters, or select the 'Variable Size' check box and specify the upper bounds in the Size box.

However, since I'm running code generation tools, I cannot use variable-size signals, so ticking the "Variable Size" box is not an option. I have tried many things, but could not solve this issue. Here is what I ended up with :

function data = USART(valA, valB)
%% Min/max values definition
% Min/max values for A and B are defined to keep them at a fixed number 
% of characters
max_A = 999.99;
max_B = 999.99;
valA = min(max_A, valA);
valA = max(0, valA);
valB = min(max_B, valB);
valB = max(0, valB);
%% Truncating values 2 digits after decimal point
% A and B are being rounded 2 digits after the decimal point
valB = round(valB * 100) / 100;
valA = round(valA * 100) / 100;
%% USART frame generation
% If the frame exceeds a set number of characters (50), it is truncated.
% If the frame character count is below a set number of characters (50), it
% is padded with spaces until it reaches this count.
% Note : this messes up any carriage return/new line.
message = sprintf('A = %06.2f | B = %06.2f', valA, valB);
if numel(message) > 50
    message = message(1:50);
elseif numel(message) < 50
    message = [message, repmat(' ', 1, 50-numel(message))];
end
data = uint8(message);
end

My reasoning is that :

valA or valB have a maximum character count (both before/after the decimal point) ;
The sprintf formatting ensures that if the values are smaller than the maximum character count, the missing characters are padded with zeros ;
The truncating/padding instructions ensures that signal width is constant (redundant).

This functions runs great in MATLAB, and generates a fixed length uint8 vector no matter what values valA and valB were given, positive, negative, small, big...

However, in Simulink, same error remains.

The only case where I can get any output from data is when I use hard coded values :

function data = fcn()
valA = 123.123;
valB = 456.456;
message = sprintf('A = %06.2f | B = %06.2f\r\n', valA, valB);
data = uint8(message);

This compiles and gets deployed on the target device, but hard-coded values have no practical use.

How can this be solved ? How can I get my desired output ?

Thanks in advance,

Nicolas

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Nicolas 2024-2-27

编辑：Nicolas 2024-2-27

I would also be grateful if someone could share an example model that sends data from a variable through the USART Write block.

Thanks again,

Nicolas

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Fangjun Jiang 2024-2-27

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/2087636-simulink-obtain-fixed-width-vector-from-matlab-function-block-output#answer_1417526

编辑：Fangjun Jiang 2024-2-27

Use the Compose String block and String to ASCII block?

https://www.mathworks.com/help/simulink/slref/composestring.html

https://www.mathworks.com/help/simulink/slref/stringtoascii.html

2 个评论
显示无隐藏无

Nicolas 2024-3-1

Thanks, this is the block I needed (and I couldn't find).

Quick note for anyone using this block with STM32 features : for some reason, Compose String block's %f formatting option didn't work for me, and I had to make another subsystem to convert my doubles into strings.

Not pretty, but it seems to work pretty good.

Fangjun Jiang 2024-3-1

编辑：Fangjun Jiang 2024-3-1

在 MATLAB Online 中打开

What is the issue, @Nicolas ? You had the correct format to generate the fixed width numerical numbers for a double data type.

 valA = 123.123;
valB = 456.456;
message = sprintf('A = %06.2f | B = %06.2f\r\n', valA, valB)
message = 
    'A = 123.12 | B = 456.46
     '
compose("A = %06.2f | B = %06.2f\r\n", valA, valB)
ans = 
    "A = 123.12 | B = 456.46
     "
compose('A = %06.2f | B = %06.2f\r\n', valA, valB)
ans = 1×1 cell array
    {'A = 123.12 | B = 456.46←↵'}