Please help on odd reference voltages

[aniruddha]

Please help me with the (schematic) design of reference vtgs. of +2.65V and +3.673V.
Accuracy and stability and stability are the most important criterion as they will be fed as analog ref. vtgs to PIC .

 

[gramo]

Why not use a Pot? You could set a reference voltage quite easily, and you should already have a rather stable power supply for your PIC

Wind the POT's to the required resistance and check them with a multimeter and your set.

If for some reason your voltage regulator decides to conduct harder/less than the voltage Vref (the reference for your ADC conversions) on your PIC is also going to be affected proportionally.

Eg; Vout from reg rises too 5.2 Volts, therefore the output of the potentiometers will rise by 0.2V but so too will what ever Pin (probably Vcc) you have as a reference for the ADC conversions, so your PIC will still read 2.26 or 3.673 volts (well the analogue equivalent), as its a digital representation of Vref

 

[gramo]

Except dont use 10K pots, as the total input impedance shouldnt be more than 2.5K for accurate ADC conversions. A 2K pot would be suffice;

http://www.futurlec.com.au/test13.jsp?category=POTRECT&category_title=Rectangular%20Potentiometers&main_menu=POTENTIOMETERS&sub_menu=POTRECT

 

[ante1]

This circuit will be sensitive to temperature changes.

 

[aniruddha]

Thanks, will try... have any specific pot for suggestion?
Will the pot give required accuracy?

I dont think stability would be a factor because after all its the ratio that counts ... temperature coefficients would get cancelled ... hence insensitive to temp. changes?

 

[Ldanielrosa1]

If you are in no particular hurry, then I have a few TL431s and daughterboards I could spare.  They need two 1206 SMD resistors each for biasing.  They stand on edge and act like a two pin zener then finished.  PM me if you're interested.

 

[gramo]

The accuracy will depend on where you set it - use a multimeter and wind too exactly where you want it, be sure to test the output without the screwdriver on the pot as the downward force may alter its normal reading.

The input impedance shouldn

 

[aniruddha]

cant construct a pot ... sorry

instead trying to find standard resistor values to satisfy the combination-ratio ...
trying hard to find them ...

thanks for the help guys

 

[Ldanielrosa1]

With the resolution you seek, standard values won't get you there in one step.  I recommend choosing one common value, and for the other leg combining resistors in parallel to trim the value down.

For example:  with a TL431 I would use a 10k 1% on the bottom leg which always sees a nominal 2V495, which I would measure with ref shorted to cathode.  For the upper leg I would start with 750R and measure the actual resistance.  After measuring the shunt voltage, expecting  2V655 to 2V709, I will know how far the '431 is off the nominal voltage and about how much reference current the chip needs.  Then I would select a resistor to parallel with 750R to get the target value of 516R to 729R, which could be from 1k6 to 26k.