ADC noise reduction with software filter 

+David Ramirez Glad it was helpful for you. Thanks for the comment.

+EMBEDONIX It is possible that a more optimized solution can exist. I only spent a few hours on this. You can also take a look at my Solar BMS project that is open source and there is way more code there at least a few thousands of lines. There is a lot of room for optimization there also but for low volume products you better spend a few more cent on a more powerful processor maybe with more memory that wasting time optimizing the code. Optimizing the code will be more expensive in low or even medium volume.

At the amount of noise I get in this example your filter will still be noisy and it will not react as fast to larger changes. The main idea of my code is to just ignore 1 to 2 least significant digit changes in both directions as that is distracting to see but when there is a more significant change than that I want a fast update to that value.

Glad the code was helpful for you.

I think that is based on signal strength. So when signal is below a certain level audio will be muted.

How do you make the measurement ? Ideally you will use some isolated way to measure then there will be no influence on your circuit. I guess you have just some restive divider and that pushes to much current in your IO pins that it locks the microcontroller maybe can even damage the internal diode that protects the pins.

I wrote this about 5 years ago and do not have time to look at the code now. The idea was to simply have a more stable ADC reading by ignoring the reading noise if it was just 1 or 2 least significant digits. 5 - The 3 has to do with the bargraph indicating the delta error on the LCD probably not relevant for you. The 3 is the middle of the bargraph representing zero deviation from that you can have 1 or 2 digit negative or positive. The absolute will display the value without the sign since that is already seen in the bargraph. The microcontroller I have has a 12bit ADC so 0 to 4096 digit output and the supply voltage is about 3.3V so ADC input voltage will be from 0 to 3.3V or 3300mV then dividing the 4096/3300mV you get that aprox 1.244 6 - again as mentioned above that is the small bargraph set to zero (middle) so 3. 7 - Just display one of the readings that is part of the 128 ADC readings to see how far a single reading is from the average to get a sense of the ADC noise. You have a 10bit ADC so 0 to 1024 so resolution and measurement accuracy will be lower with the PIC microcontroller. For you is probably enough just to make an average to get a better reading since temperature changes slowly with time. Also try use a low pass filter before the ADC input to reduce any possible noise from the long temperature sensor wires. I'm guessing you are using an NTC thermistor.

1- am using a LM35 as it specifies the hja of ... the measurement data output is stable with Multimeter. 2- respect to the power source is also stable. 3- am using reference voltage from 0 volt to 1024 mVolt and this is also stable I mean that the above three parameters do not vary simpre remain termometro except that varies with the tempeatura but no sudden vaiaciones .... I even I have placed 100 stable mVolt in emtrada the adc and still varies widely ... .for thus I imagine that the error is internal .. that's why we want to implement this filter because it works for you great. Thanks for the prompt response...

Your ADC will make fast sampling of the signal so it can measure peak voltage. I do not know your schematic but try to use a low pas filter realy close to the ADC input made of say a 1Kohm resistor and a 22uF between ADC input and ground. I'm sure this will hugely improve the ADC reading. If you notice in my video the ADC reading without any filtering is the first display value and while it has some variation is no more than 4 to 5 digit. In your case 5 digit variation will be about 0.5C variation without any averaging or filtering. But since is a slow signal best way to filter is in hardware with a low pas filter as mentioned then if you want you can add some software filter for even better accuracy.

ok .. I will adapt the filter as you recommend, then you commented, thank you very much for your time greetings'll be in touch.

+Sachin Singh Hi, I just took a look at the code and yes it looks that way but is not. All the code is in the GPIO_LCD1602 there is no excuse for this :) but was just a short experiment I done in a few hours so in not well organized not to mention I'm more of a HW engineer not so much SW but this is an old video I learned a bit since then :)

+Nihilokkk You can say that but the scope was not exactly that since you keep the signal "analog" except for narrow hysteresis portion where you do nothing. On top of that not many will know what a schmitt trigger is and will not search for that. Thanks for the comment.

+electrodacus I meant what about implementing a Schmitt trigger. But you are right its a bit a different concept that is not fully useful here.

+Nihilokkk Yes almost no software is needed for that just an if ADC value higher than x1 set to 1 and if ADC less than x2 set to zero. It is probably used all the time not sure if is called a software Schmitt trigger or not. There are a lot of this sort of triggers in my current project the Solar BMS where I need to check multiple cell voltages and set upper and lower limits for charging or load and same with currents. Solar BMS is open source software and hardware so if you are interested take a look at my google+ page plus.google.com/+electrodacus/posts Software is done and code will be released in a few weeks at most.

It is not empty but I think I know what you mean. Go in to example folder and check the GPIO_LCD160.c file.

:) Maybe the Solar BMS project is a better rival to the house project.

***** brilliant design and build, so as the house