Voltage reference value too low

[lukasz.kostka]

Hi

I've build a basic working circuit with MCP1541 which is a 4096 mV voltage reference. Thing is that I measure 4092 mV. I am a bit puzzled because I do not know how to set that voltage to 4096. Data sheet says that minimal voltage is 4055 and maximal is 4137. Bigger than 4096 is fine because I can trim it with a pot, but what can I do when I have less than 4096 ?

[HB9EVI]

like you say yourself 4,092V is well within tolerance; either you have to order a bigger stock and select a matching one, or you have to reconsider the circuit you feed with the reference; I don't see a reason, why the 4mV should do a big difference in the real world.

p.s. even if you find a matching one, there still is tempco and aging.

[Benta]

We're talking 1/1000 deviance here. Be happy that you got a part that good.

[tsman]

4.092V is well within the tolerance range of 4.055V to 4.137V and very close to the ideal 4.096V so I don't see the problem here. If you need something more accurate then you'll need to cherry pick a MCP1541 which is even closer to the ideal 4.096V or find a more accurate voltage reference. Trying to trim it with a pot is going to cause bigger accuracy issues.

What is the accuracy of your measurement device?

[lukasz.kostka]

I am using 12 bit DAC and ADC. That is why I've picked 4096 as my voltage reference.

[HB9EVI]

I did the same for several projects - with the MCP1541 - I didn't even bother to match the best; in 99,999% of all cases the circuit is going to work as well with just 4,092V.
To trim the voltage with a trimpot is not a very good idea, it just brings you more tempco and other drifts in the system

[lukasz.kostka]

So I have these options:

• Buy a more accurate reference.
• Live with what I have.

[HB9EVI]

that's about the choices. My guess is, living with it is perfectly allright; I built a simple DAC controlled PSU with a MCP1541 as reference, and it's doing fine - the DAC always brings higher non-linearities in the system than an offset of a few mV from the reference.

what is your project you're working on? a schematic would be helpful.

[lukasz.kostka]

I am working on a linear PSU. I will paste link to my kicad files as soon as I upload them to my github. I will write here when I posth them.

[lukasz.kostka]

I've attached my schematic. This is not yet finished board.

[HB9EVI]

yea, there is still a lot missing
however, I worked already several times with MCP4922; it's an ok DAC, but has its nonlinear behaviour which will mess up things much more than the missing 4mV from the reference

Additionally, you'll not get the voltage of the psu down to 0V, even if the LM358 is a single supply opamp; I used IORR MCP6282 and I got down to a Vmin of 25mV - that's not only due to the opamp, also the DAC never reaches 0V, nor 4.096V at fullscale.

I you want a straight voltage scale, say 0-20V in lets say 10mV step, you have to use a bipolar setup and make it possible to calibrate in software - just in hardware alone you'll never get the theoretical possible 0-20,48V in 5mV steps.

[lukasz.kostka]

Thx for all your replies. I will post a new thread not to junk this one with design issues etc.