STM32L47x ADC maximum input voltage

[Markus Vogel]

Posted on April 26, 2017 at 08:40

Dear all,

I have the following situation:

• STM32L471/L476 processor
• Using the ADC with the internal reference buffer (2.5V) as analog eference
• One of the ADC input signals comes from a rail-to-rail OP-AMP with 3V supply voltage, so this input signal can be up to 3V, and therefore higher than the ADC reference voltage
• Normally this ADC input signal will be within 0 to 2.5V but it might occassionally exceed that range and then go up to 3V

Now my question is:

Is it a problem when the ADC input voltage is above the reference voltage? I know that some ADC's can get damaged in this case. In the datasheet I found that

• GPIO VIN max = 4V (STM32L476xx Datasheet, DocID025976 Rev 3, Table 18)
  • This absolute maximum rating I would fulfill
• ADC VAIN Conversion voltage range max = VREF+ (STM32L476xx Datasheet, DocID025976 Rev 3, Table 62)
  • This looks like it is just where the ADC converts but not like a do-not-exceed limit

So if the ADC in this configuration can get damaged by a 3V voltage I would look for some alternative (other reference, limit input signal externally) but it would be the best if it could withstand it.

Additional information:

• I cannot use the 3V supply of the OP-AMPs as it is switched on/off but I need a constant analog refernce since I require the DACs all the time
• Using another 3V supply would be possible but costly (this is a device prototype)

Hope you could undestand my problem, otherwise just let me know.

Thanks in advance and kind regards

Markus

#stm32l4 #analog-inputs-for-adc

[S.Ma]

Posted on April 26, 2017 at 10:14

As a general rule, being above the normal max operating means accelerated ageing before permanent damage.

Above AMR (Absolute Max Rating), the chip may be damaged anytime.

At AMR, the chip may be operational up to a certain time (typically 1000 hours?)

In between... 

[AvaTar]

Posted on April 26, 2017 at 11:04

I don't have a L47x board in my collection, which is already too large. So I didn't read the datasheet / reference manual for your specific part.

But I bet it (the datasheet) defines the maximal input voltage of the ADC, and supposedly identical to the maximal supply voltage (3.6V on other parts).

Quite sure there is no harm done in your case, but your ADC readbacks will be meaningless when above Vref.

If you can set Vref equal to Vcc as well, a supposed damage would be illogical.

[Markus Vogel]

Posted on April 27, 2017 at 09:26

Unfortunately I could not find a maximal input voltage for the ADC in the datasheet. There is a limit for V_AIN, but it is described as ''Conversion voltage range'', see following picture from the datasheet. That the conversion voltage range spans from 0 to VREF+ is obvious but still leaves the question for maximum ADC input voltage.

I also would not expect harm but I hoped to get some definite answer on it.

Do you by chance know if an input-voltage-limit is defined for another ST-processor's ADC? Maybe that's a point to start from as I would expect ST to use similar ADC circuitries in different processors.

Thanks already for your quick reply and kind regards

Markus

[RomainR.]

Posted on April 27, 2017 at 10:11

From the STM34L476 datasheet chapter 6.2 Absolute maximum ratings below:

All STM32s have GPIO FT and FF type. This corresponds to the maximum and minimum guaranteed values.

• FT stands for 5 Volt Tolerant GPIO.
• FF stands for 3.6V Tolerant GPIO.

Generally Analog IO are FF_a or FT_a and Non-Analog IO are FT, I suggest you consult the datasheet of your device (chapter 4 )

In your case, I think there is no risk of destruction for the analog input and the ADC as long as your signal is less than 3.6V.

On the other hand if your VRef + is the internal reference 2.5V you will not be able to convert the variations of the input signal from 2.5V to 3.6V, there will be saturation of the conversion result.

Change VRef to 3.3 (VDDA) or make a conditioning of your input signal by scaling to meet the full scale of your ADC.  

[Bertrand DENIS]

Posted on May 04, 2017 at 19:10

I confirm that there is no risk to damage the product in the case you are describing above.

[Markus Vogel]

Posted on May 09, 2017 at 14:02

Ok, perfect!!!

Thank you everybody for your help, really appreciate it