Battery percentage on door contact sensor erratic (sent on interrupt)

[gohan]

I am using the vcc library and also a voltage divider to read higher voltages than 3.3v and I make the average of the last 4 readings before sending the value and it has been working very well so far.

[GreyLinux]

@gohan @Yveaux I think this might be the option to send in the average reading .To be fair this was opening the contact every couple of seconds and in reality this would be alot longer between reads . Thanks for your help.

[rozpruwacz]

@greymarvel what kind of battery are You using ? The voltage on the battery may not be constant in short periods, it depends on the load and bettery internal resistance. The higher the internal resistance the bigger are voltage swings in response to load swings. I have two identical boards, differing only in the batery holder, one is for cr2032 (higher internal resistance) and the other is for cr123 (lower internal resistance). These boards have exactyl the same circuit and software, the only difference is the battery type. And the one with cr2032 give me very unstable battery voltage readings while the other very is stable. So for sure it depends on the battery type and the current consumption at the moment of voltage probing.

[GreyLinux]

@rozpruwacz I'm using 2 AA batteries . I think it might be worthwhile me using @Yveaux vcc library as a reference and see if that improves things. Also I think I might delay a little longer before the read to allow it to settle more after the contact interrupt .

[rozpruwacz]

@greymarvel AA batteries should have much lower internal resistance, so I guess that voltage reading should be stable. Good luck finding the problem :)

[gohan]

try adding a little ceramic cap on vcc or near the analog pin you are reading from.

[GreyLinux]

@gohan

This is how it looks I know it's a little messy but a little cap between the high impedance point and A0 will help ?

[gohan]

are you using a voltage divider on pin A0? if so, a small 0.1uF ceramic should do. If you are using a booster is your only option. In addition you can also consider using a single LiFePO4 battery and use the vcc library to get battery voltage

[GreyLinux]

@gohan yeah it's hard to see but the black wire behind all the radio wires goes to A0 from the divider. I'll put a little cap on this .

[gohan]

You may also consider some Easy Pcb you can find on openhardware.io that would make a much cleaner solution for your node

[GreyLinux]

@gohan yeah I have some on order. I'm just learning with what I've got till they come and the sensors I've made won't be seen so not a problem for the moment . Thanks again .

[GreyLinux]

@gohan well adding a cap from the impedance point to A0 has 100% stabilized my readings

4630 MCO:SLP:WUP=-1
4632 TSF:TRI:TSB
4634 MCO:SLP:MS=0,SMS=0,I1=1,M1=1,I2=255,M2=255
4638 TSF:TDI:TSL
4640 MCO:SLP:WUP=1
4642 TSF:TRI:TSB
Sensor value: 862
Battery Voltage: 2.78 V
Battery percent: 13 %
4751 TSF:MSG:SEND,4-4-0-0,s=3,c=1,t=16,pt=2,l=2,sg=0,ft=0,st=OK:0
4759 MCO:SLP:MS=5,SMS=0,I1=255,M1=255,I2=255,M2=255
4763 TSF:TDI:TSL
4765 MCO:SLP:WUP=-1
4767 TSF:TRI:TSB
4769 MCO:SLP:MS=0,SMS=0,I1=1,M1=1,I2=255,M2=255
4775 TSF:TDI:TSL
4777 MCO:SLP:WUP=1
4780 TSF:TRI:TSB
Sensor value: 863
Battery Voltage: 2.78 V
Battery percent: 13 %
4886 TSF:MSG:SEND,4-4-0-0,s=3,c=1,t=16,pt=2,l=2,sg=0,ft=0,st=OK:1
4894 MCO:SLP:MS=5,SMS=0,I1=255,M1=255,I2=255,M2=255
4898 TSF:TDI:TSL
4900 MCO:SLP:WUP=-1
4902 TSF:TRI:TSB
4904 MCO:SLP:MS=0,SMS=0,I1=1,M1=1,I2=255,M2=255
4911 TSF:TDI:TSL
4913 MCO:SLP:WUP=1
4915 TSF:TRI:TSB
Sensor value: 862
Battery Voltage: 2.78 V
Battery percent: 13 %
5021 TSF:MSG:SEND,4-4-0-0,s=3,c=1,t=16,pt=2,l=2,sg=0,ft=0,st=OK:0
5029 MCO:SLP:MS=5,SMS=0,I1=255,M1=255,I2=255,M2=255
5033 TSF:TDI:TSL
5036 MCO:SLP:WUP=-1
5038 TSF:TRI:TSB
5040 MCO:SLP:MS=0,SMS=0,I1=1,M1=1,I2=255,M2=255
5046 TSF:TDI:TSL
5048 MCO:SLP:WUP=1
5050 TSF:TRI:TSB
Sensor value: 862
Battery Voltage: 2.78 V
Battery percent: 13 %
5156 TSF:MSG:SEND,4-4-0-0,s=3,c=1,t=16,pt=2,l=2,sg=0,ft=0,st=OK:1
5165 MCO:SLP:MS=5,SMS=0,I1=255,M1=255,I2=255,M2=255
5169 TSF:TDI:TSL
5171 MCO:SLP:WUP=-1
5173 TSF:TRI:TSB
5175 MCO:SLP:MS=0,SMS=0,I1=1,M1=1,I2=255,M2=255
5181 TSF:TDI:TSL

however a multi meter gives me 2.90v and the reading is 2.78v i understand it wont be 100% accurate but this far out is a little weird my other sensors are about 1 or 2 volts if that . any ideas?

[rozpruwacz]

@greymarvel check the actual values of the resistors in voltage divider how far they are from the nominal values. Resistors have some tolerance so they are not exactly 1kOhm or whatever, this makes the division ratio different than You take into account calculating voltage from ADC reading.

[gohan]

also the internal reference is not 100% accurate, so you can't have a multimeter precision. There is a battery measurement code that you can add some correction to it, look into the examples

[GreyLinux]

@rozpruwacz I've checked the values and they don't seem to be that far from expected but as you say they definitely weren't the stated resistance . I think @gohan is right I will add the correction in the sketch and try and get it reasonably close. Thanks again everyone for your help .

[BulldogLowell]

As the analog pins share a single A2D, rapid use of it can cause the previous read to impact the next. A common workaround in this situation is to read twice on each input, ignoring the first.

try that:

// read analog pin
  (void) analogRead(BATTERY_SENSE_PIN);
  int sensorValue = analogRead(BATTERY_SENSE_PIN);
  // calculate battery voltage
  float vBat  = static_cast<float>(sensorValue * (V_MAX/1023));