how to protect analog inputs ?

[mehmet.karakaya]

Posted on February 20, 2013 at 18:04

hello forum,

I am using analog inputs of STM32F103 to measure current of a DC motor using ACS714

hall effect sensor

however at first trial the analog inputs burned becouse of high frequency ringing ( caused by the PWM ) at the measured signal

my question

can I use following circuit to protect analog inputs ?

I have checked user manual of F103 there it says max analog input voltage VDD+0.3 V

minimum VSS-0.3 V

however with this diode network max voltage will be VDD+0.6 Volts and minimum

VSS-0.6 Volts

any advice , thank you

#analog-input

[emalund]

Posted on February 20, 2013 at 18:22

for diode protection always use Schottkeys they are faster and Vf is lower.

and have a ceramic cap DIRECTLY between the two diodes.

I assume you are using a PCB with ground plane, otherwise good luck.

Anyhow even a shottkey may be too slow, you probably will need to add a transient absorber..

a very good article on the subject can be found here:

http://electronicdesign.com/boards/how-much-chip-transient-protection-enough

Erik

[jj2]

Posted on February 21, 2013 at 04:23

Agree essentially w/Erik's guidance - use Schottkey Diodes - ''some'' series R - and C - which still enables your capture of highest frequency of interest.  Unclear if you must protect one or multiple ADC inputs - in the case of multiple - compact & efficient multi-channel ''filters'' are available - often used in the quest to achieve agency (radiation) approval.

Most are not especially ''beefy'' - unable to withstand both high magnitude and sustained abuse.  It seems useful for you to impose a sufficiently robust filter/voltage clamp circuit - between your ''generator'' and MCU.  MCU really should not be tasked for this job - and use of MCU's diode protection scheme will win few fans.

Your report of ''burning'' inputs - with a motor nearby - suggests that your MCU may have ''absorbed'' the motor - ''acting as generator'' effect.  If a motor under reasonable load is commanded to stop (or sufficiently slow) often the motor will serve as a generator - and can pull-up the motor's voltage supply - often with fatal consequences to many Vbus components.

My experience teaches that these unplanned/unwanted voltage spikes may be powerful enough to wreak considerable damage - and should be countered away from your fragile MCU...

[frankmeyer9]

Posted on February 21, 2013 at 08:29

Your report of ''burning'' inputs - with a motor nearby - suggests that your MCU may have ''absorbed'' the motor - ''acting as generator'' effect.  If a motor under reasonable load is commanded to stop (or sufficiently slow) often the motor will serve as a generator - and can pull-up the motor's voltage supply - often with fatal consequences to many Vbus components.

 

That right, and neccessarily so by the law of energy conservation. The motor will ''sweat'' off the stored kinetical energy as heat and back emf. Your protective elements must be able to cope with the energy fed in by the motor. And small shottky diodes can't take more than a short transient surge. A typical suppressor diode, for instance, can stand 100 ... 10.000 Watt, but only for microseconds.

In short, measure your EMI, and design the protection accordingly.

I suggest to consult a hardware designer experienced with driving inductive loads.

[jj2]

Posted on February 21, 2013 at 14:57

Armed now w/2nd poster's affirmation - a method to ''reduce - possibly prevent'' such damaging voltage surge bears introduction.

Classical technique is to impose a properly sized/rated resistive load - across the VBus - upon the detection of such VBus ''rise.''  Depending upon the motor - this may well be a power R - and may require heat sinking and/or forced air cooling.  (it will get hot!)

MCU must properly detect such condition - switch in this ''VBus Pull-Down'' Power R - and carefully release it once the VBus returns to ''safe'' levels. 

Such ''prevention'' (or limiting) of offense will ease (and may lessen) your subsequent ADC protection efforts - which still must be carefully/skillfully designed and emplaced...

[mehmet.karakaya]

Posted on February 21, 2013 at 17:52

I am using LM2575-3.3 for CPU voltage supply

and I am driving the 24 V 200 W DC motor with mosfet H bridge

besides the output of ACS714 hall effecet sensor is LP filtered with 10K + 100 nF

something came into my mind

can I use 2x of the diode network side by side ( above picture ) ?

one with UF4007 and one with schottky diodes

becouse UF4007 has 0.6 - 0.7 V forward voltage

and this rises the analog input voltage to VDD+0.6 , VSS-0.6 in case of peaks

second) I have seen that the small signal schottky diodes reverse voltages are limited too to 30 V