How to convert reading ADC Value to Float?

Question

I want to convert ADC Values to float numbers. This is my code:

uint32_t ADCValue;
char str[15];

          ADCValue = HAL_ADC_GetValue(&hadc1);
          ADCValue=ADCValue*3.3/4095;
          sprintf(str, "%d", Voltage);
          HAL_UART_Transmit(&huart2, (uint8_t*)(&str), 6, 100);
          HAL_UART_Transmit(&huart2, (uint8_t*) "\n\r", 2, 100);
          HAL_Delay(500);

I am reading value of 4095 at 3.3V but I am transmitting 3 to buffer. I want to transmit exact value of voltage.

`ADCValue=ADCValue*3.3/4095` --> `ADCValue=(ADCValue*10)*33/40950` — LPs, Jan 13 '17 at 08:30
@LPs - why the extra multiplication by 10? 3.3 was already changed to 33. — Support Ukraine, Jan 13 '17 at 08:38
@LPs - I see... So instead of sending the truncated value `3`, it will be `33`. Nice idea in case floats are not an option. — Support Ukraine, Jan 13 '17 at 08:45
@4386427 I guessed it's an embedded platform, so better to work with int instead of float. ;) — LPs, Jan 13 '17 at 08:46
@Lps I am reading 33 instead of 3.3 . I think this is not exact solution. — coollime, Jan 13 '17 at 08:46
@Dummy You send 33 but you know how to "decode it" on the slave device. — LPs, Jan 13 '17 at 08:47
@Dummy ... or you can transform `33` to `3.3` yourself by some simple string manipulations. — Jabberwocky, Jan 13 '17 at 08:49
@Dummy just to eliminate all possible misunderstandings: which string do you want to transmit for `ADCValue = 3457` ? — Jabberwocky, Jan 13 '17 at 08:54
@MichaelWalz If ADCValue=3457,I should read 3457*3.3/4095= 2.78.I want to read 2.78 Volt. — coollime, Jan 13 '17 at 08:59
Since you don't appear to be doing any advanced math calculations, why do you need float numbers to begin with? Does you MCU even have a FPU? — Lundin, Jan 13 '17 at 10:13

Lundin · Accepted Answer · 2017-01-13T11:45:00.210

6

There is no apparent reason why you need to use float numbers here in the first place. They are needlessly slow. Simply do this:

uint32_t millivolts = ADCValue*3300/4095.

For an input of 4095, you get the result 3299mV. This is possibly more accurate than what you would get through float numbers, because of floating-point inaccuracy.

edited Jan 13 '17 at 11:45

answered Jan 13 '17 at 10:20

Lundin

195,001
40
254
396

While I agree that floats are not necessary here, there is no reason why they would be less accurate here. Unless you back it up with something, this answer would be better of without such remark. – user694733 Jan 13 '17 at 10:44
In case of using CMSIS DSP float functions we have to convert ADC output to float. This is bad and make ADC DMA useless, What if hardware could output float directly. – mohammadsdtmnd May 25 '22 at 05:04

Jabberwocky · Answer 2 · 2017-01-13T09:39:09.273

5

You probably want this:

ADCValue = HAL_ADC_GetValue(&hadc1);
float fADCValue = (float)ADCValue*3.3 / 4095.0;
int txlength = sprintf(str, "%.2f", fADCValue);
HAL_UART_Transmit(&huart2, (uint8_t*)(&str), txlength, 100);
HAL_UART_Transmit(&huart2, (uint8_t*) "\n\r", 2, 100);
HAL_Delay(500);

Caution: this only works if the float type is available on your platform.

If floating point is not available you can use this:

ADCValue = HAL_ADC_GetValue(&hadc1);
ADCValue = ADCValue * 330 / 4095;
int len = sprintf(str, "%d.%02d", ADCValue / 100, ADCValue % 100);
HAL_UART_Transmit(&huart2, (uint8_t*)(&str), leb, 100);
HAL_UART_Transmit(&huart2, (uint8_t*) "\n\r", 2, 100);

This is a quick hack and there is certainly room for improvement.

Demonstration here.

edited Jan 13 '17 at 09:39

answered Jan 13 '17 at 08:39

Jabberwocky

48,281
17
65
115

My platform may not be available for float type. – coollime Jan 13 '17 at 09:00
1

@Dummy see my updated answer. Now there is a solution for the case where floating point math is not available on your platform. – Jabberwocky Jan 13 '17 at 09:10
@MichaelWalz It worked.Now I am reading in float number.Thanks man. – coollime Jan 13 '17 at 09:14
@4386427 good idea, answer updated. But still could not spot any other decimals related problems. – Jabberwocky Jan 13 '17 at 09:31
1

You need to have the `02` in format string for decimals: http://ideone.com/RBJCmn – user694733 Jan 13 '17 at 09:37
@Dummy the program was not quite correct, the answer has been corrected now. – Jabberwocky Jan 13 '17 at 09:43

score 0 · Answer 3 · answered Jan 13 '17 at 08:42

0

This should do the trick

float Adcvalue_f;
uint32_t ADCValue;
char str[15];
....
ADCValue = HAL_ADC_GetValue(&hadc1); 
ADCValue_f =ADCValue*3.3/4095;
sprintf(str, "%2.2f", AdcValue_f);   //Adjust the .2 as per required precision
HAL_UART_Transmit(&huart2, (uint8_t*)(&str), 6, 100);
HAL_UART_Transmit(&huart2, (uint8_t*) "\n\r", 2, 100);

answered Jan 13 '17 at 08:42

Rishikesh Raje

8,556
2
16
31

With this code,I am reading just empty data :) What am I doing wrong ? – coollime Jan 13 '17 at 08:54
empty data at other UART or at `str` – Rishikesh Raje Jan 13 '17 at 09:13

chux - Reinstate Monica · Answer 4 · 2017-01-13T17:11:49.510

0

In addition to other fine answers, consider:

Integer division truncates the factional part of the mathematical quotient.

Instead, perform a rounded integer division. With unsigned integer math, add half the divisor before the division.

// ADCValue=ADCValue*3.3/4095;

unsigned Voltage = // millivolts
Voltage = (ADCValue*3300 + 4095/2)/4095.

edited Jan 13 '17 at 17:11

answered Jan 13 '17 at 17:06

chux - Reinstate Monica

143,097
13
135
256

How to convert reading ADC Value to Float?

4 Answers4