Microcontrollers and stuff A blog by Derk Vedelaar

Precision voltage reference

I recently stumbled upon the chip ref102p (datasheet, pdf). This is an extremely precise voltage reference that gives 10 volt +/- 0.0025v without calibration. I thought this is a great way to test my multimeter (which is an extremely cheap one, because well, I don't need an expensive one).

After some searching I found out I am not the only one thinking this. Some more searching made me find that there are more series of chips doing this. For example the Intersil 21090BFB and the 21009BFB ranges. These Intersil chips have the values of 1.25v, 2.5v, 5v and 7.5v. Especially the 1.25v is really nice in my opinion, because <2v will give an extra digit on my multimeter.

I chose to make a small unit which has the values of 1.25v, 5v, and 10v to test my multimeter. This is the result of the module:

There is a small dc-dc converter which takes 5v from micro usb and will scale it up to about 15 volt. Then there is a 12 volt voltage regulator to make a nice stable 12 volt and a capacitor to smooth it out even more. Then all the chips need is power and ground and that's all. The two Intersil chips are in a SOIC package which is annoying to solder on normal protoboard. That's why I chose to buy small converter boards from SOIC to DIP. The chips can easily be soldered on the converter boards using an hot air gun and some solder paste.

Usually I put a bunch of wire on the bottom of the PCB to wire everything up. But somewhere on the internet I saw someone make really nice traces on the back of his pcb with solder. I tried that too:

I think it looks quite nice and I don't have the problem anymore of wires that come loose.

To the interesting part. How is my multimeter doing?

The 1.25v is bang on with my multimeter, well, on the 2v setting. On the 20 volt setting it actually gave 1.26 volt. The 5 volt is 1 hundredth of a volt off, and for the 10 volt reference, it couldn't decide between 10.02 and 10.03 volt. Apparently in the 20 volt setting the higher the voltage the more it is off.

There are some people that calibrate their multimeter using these references, but since my multimeter is almost perfect, I won't change anything about it. I will probably only make it worse anyway.

I also own a oscilloscope, a really cheap do it yourself one. This one was less precise than my multimeter, but still quite good. 1.25 volt shows 1.256 volt. This voltage can just like with my multimeter be measured with one extra digit. 5 volt is bang on and 10 volt actually shows 10.20 volt.