A precision programmable voltage reference circuit capable of 0.001 to 4.095 volt output in 1 mV steps with an accuracy of 100 uV.

This project is based on a circuit designed by uChip and presented in the Sparkfun forum. There is a Github page for the original design.
Original Uchip programmable voltage reference with HP bubble display. Set for 4.000 volts


I have been wanting to build an adjustable precision voltage reference for some time now, and was really excited to find this design by uChip on the Sparkfun forum. It had most of the features and fit the specifications that I was interested in.
I made a few small design changes such as adding a wideband noise reduction capacitor to the REF5040 circuit, additional board routing to clear the area under the unused Vref pins, adding mounting holes, removing the solder mask from critical areas, and compacting the board a bit.

The original design is capable of generating an output between 0.001 volts and 4.095 volts with an after calibration accuracy of 100 uV. It uses a REF5040 low-noise, very low drift, precision voltage reference IC by Texas Instruments. The datasheet specifies a temperature drift of 3 ppm/°C Max for the high-grade version, and a long term stability of 45 ppm/1000 hr after 1000 hours.


The circuit worked great for 1 week, and then I had problems with the processor starting up when powered on. All voltages except for the negative analog, that is generated by the micro-processor were spot-on. Usually after several power cycles the system would start up and I would see the display and output reading 1.000 volts.
Later that afternoon the board started pulling excessive current, and the output voltage was dropping on the 5 volt regulator. The problem ended up being the 328P micro-processor.

I decided to do a major redesign so that I could add more features such as output monitoring, off-board I2C display, off-board rotary encoder with RGB LED, SPI isolator, and a LTC1152 op-amp for the output buffer. The processor was changed to a Teensy V3.2 board using a MK20DX256VLH7 Cortex-M4.
More to follow.
Any chance of this appearing as a kit at some stage?
After assembly and testing of the version 2.12 board I will make it available to order from the OSH Park printed circuit board service, and will most likely have several single bare boards available. As far as a full kit, I currently do not have the available time for the logistics of putting together a kit right now, but this may change in the future. If the 2.12 board works well, I will at least put together an easy order BOM from Digikey.
Thanks,
Barbouri
OK, the reason we were interested in a kit was;
1) The difficulty of hand soldering on such fine pin pitches
2) The need to program a micro we don’t normally deal with
Any suggestions on overcoming these hurdles?
What’s the point of routing away the board under the reference? This might make sense if board leakage or high voltage creep were an issue, but neither of those will have any impact on the REF5040 so the benefit of the PCB “peninsula” escapes me.
Nevertheless, a very cool project and way cheaper than those Datel calibrators:
Hi Len,
The reason for the PCB “peninsula” is to reduce thermal and mechanical stress on the voltage reference IC.
See Linear Technology’s Application Note AN-82 page 7, for a detailed explanation.
Thanks,
Barbouri
Please let me know when the pcb and bom are available, i need to get one of these asap. thanks (post would be to 4509 queensland australia) Am extremely interested.
All the board files and BOM are in the Programmable voltage reference v2.12 assembly post.
Greg (Barbouri)