Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.
Generally voltage regulators are not very good voltage referencies due to their average temperature compensation, and this include zener shunt regulator such as 341.
There is however one exception, and I have been using it as voltage reference for 16-bit and more A/D converters. I am talking about the "old" 723 voltage regulator. The circuit has on average ≈0.003%/°C temperature coefficient.
If you have some room on your PCB - go for it !!!
Obviously one has to be realistic here: if you are happy with ≈1% precision, then any voltage regulator is good, and you can even use local +5Vcc as a reference ..
Provided you are calibrating your units the absolute value does not normally matter. On the assumption that there is no noise on the reference, the issue is temperature drift. On the TL431C (0 to 70C) the worst case given by TI is 16mV. This does not include the difference that the programming resistors and the Iref current will contribute. Assuming a reference voltage of 2.5V the variation is 16/2500=0.64%. On an 8 bit A/D the resolution is 1/256=0.39% so you will get a two count error based on the thermal performance.
Keep in mind the variation is dependent on temperature range and the reference value. For the above range the TL431 is good for a 7 bit a/d. If you only working over a 10 degree variation the performance would significantly improve.
For a 12-bit converter, assuming the input voltage is 0-5V, you have 2.5V/4096=0.61mV/ count.
Thus, the TL431 will give you an error of about 26 counts. That is just the thermal drift. And if you do not have a way of calibrating the system, then you need to add the errors due to the tolerance of the TL431 at room temperature. That is another 50 counts or so.
Therefore, I would not recommend it. If you use a 12-bit A/D, spend a little more money and get a suitable reference, one that will give you just a few counts of error over the required temperature range.
Of course, it is your circuit and you decide on its accuracy, but what's the point of having a 12-bit A/D that only produces 8 bits or so?
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.
