adjustable vref voltage ic

[MCUprogrammer]

Hello to everyone

Is there any voltage reference integration you can recommend me?
The processor I use
dsPIC33EP512GM710. I am using 24 channel adc. I am looking for a stable vref integrated. Is there a product we use and you are satisfied with? Output needs to be set. It would be better to have 1.2 to 5.2 volts.
_________________
Best Regards...
MCUprogrammer
_______________________________
Work Hard

[temtronic]

There's a HUGE problem in having an adjustable Vref circuit.
If you use a pot to adjust/set the Vref output, it WILL change over timeduw to temperature and humidity, maybe even due to vibration. At the very minimum you'll need a '20 turn pot' to 'dial in' the Vref you want. Then there's wiper 'noise'.All pots do this as material gets scrapped off the 'resistor' by the wiper. The more you adjust, the worse it gets. Think of the 'crackle' in a volume control.
If you use 2 resistors to 'set' the Vref output, you have another set of issues. While more stable than a pot, you need to calculate the correct values. Not that hard but ordering 1,233r resisitors isn't easy ! Even using standard values, you'll have to 'trim' . One technique is to use 4 resistors in a 1-2-4-8 set, removing a combination to get exactly the voltage you want. Takes up board space,cheap,but expensive to calibrate. Also you'll need .1% resistors. Standard 1% values, well, they're +-1%, so 100r can be 99 to 101r and be within 'spec' yet your vref could be at least 4% wrong.

Normally you choose a vref device to be just below the Vdd max of the ADC . A 5V ADC will use a 4.096 Vref. do the 'math'...it's a magical voltage.

Be SURE that the Vin of the Vref device is well filtered and isolated from Vdd of the PIC. The more stable the power supply, the more stable the Vref will be. Also use bypass caps ! to eliminate ANY noise getting into the device.

[MCUprogrammer]

Is there a product you can recommend?
_________________
Best Regards...
MCUprogrammer
_______________________________
Work Hard

[Ttelmah]

Yes. Even going digital and using a digital pot, the actual accuracy will be
way below a good Vref.

Maxim have a note on adjusting one of their Vref's with a digital pot.
https://www.maximintegrated.com/en/design/technical-documents/app-notes/6/629.html

However even with careful setup, only 0.5% accuracy. Any 'gain' in signal
resolution by lowering the Vref, is more than lost because of the lower
accuracy.

For some reason when I post this as a URL, the whole post does not show..

I use the LTC6655. Usually at 2.048, or 3v.

[chaphill]

Sorry I am late to the discussion. Without knowing the actual needs of the OP I would like to throw in info about a circuit I have been using for decades. It uses a precision reference, e.g. LT1019 which is available in multiple voltages, and an MDAC to such as AD5543 to set the output voltage. This technique has been used for decades as the reference to adjust hv power supplies. Note that I can not tell you what the "accuracy" is but the voltage is controllable to better than 0.05% so the precision is pretty decent. In the circuit we are not concerned with the actual output voltage, we have a feedback from the signal generated by the supply that we use to control the circuit. Be aware this is not a cheap circuit, it is probably over $30 in small quantities.
"Some genius" decided to change the circuit to use the pwm outputs of dsPIC33EV256 and PIC24HJ128 processors with an RC filter. The feedback circuit is certainly a lot more busy. I sure miss my MDACS.

[temtronic]

There is a nasty 'little' problem if you use an adjustable Vref circuit.

Who and WHEN tells the PIC WHAT the actual Vref is !!!!
How often do you test the ACTUAL Vref output ?
ANY resistive divider WILL change value over time, temperature and humidity.
I KNOW that trimpots, even 20 turn units, change over time, so I did the 'math' and used 2 fixed resistors with 0.1% tolerance.
That could be 1-5% so HOW does the PIC know that and modify the 'math' for whatever calculations it uses Vref for ?

[chaphill]

In the circuit that I described the DAC I mentioned has a tempco of about 1 ppm, so even over a 100 deg. C range the delta is going to be on the order of 100 ppm from the DAC. For practical purposes you are looking at the variation of the reference IC and often the op amp buffer as a function of temperature. The 100 ppm from the DAC is probably insignificant.
I am assuming the DAC is set by the dsPIC in this case, so it depends a bit on the programmer but if the programmer knows what he just set the DAC to he should know or at least be able to calculate the output of the ref. voltage because he is the one that set it.

[temtronic]

light bulb came on....dim like today's weather...
just WHAT are the 24 sensors ?
what is their accuracy ?
While it'd be nice to have a 1.02400000000 vref, it's not really needed if the sensors are only good to say .01v
Also what's the 'update' or 'scan' time needed for the sensors ? Always best to read a sensor multiple times and get an average.