Bench DMM project

Well, it's day I don't know how many and somethingty into this project, and I'm starting to get something that looks like a design. It's all a bit new.

It all started when I decided to recalibrate my cheap handheld dmm a few months ago, and discovered it's a bit naff. So I started pursuing accuracy, and eventually settled on using a 24 bit ADC from LT. Could be overkill, but in for a penny, in for a pound, they say. So I'm looking at 6 digits now.

LT also do a nice true rms converter for the AC ranges.

This is a general description of the design as it stands.

The ADC itself is incredibly easy to use. It just needs a reference, an input divider and a microcontroller. The hard part was learning how to program a PIC microcontroller. I settled on a 16F57 in the end. Thanks to the good folks on the forums here I discovered some holes in my reasoning!

So here's how things are looking so far. A 10 megohm input ladder divides down the input voltage to no more than 2.5v. The reference for the ADC is equal to it's supply voltage, at 5v, hence 2.5v since it's a differential ± input. OVP will be provided by zener diodes and a spark gap.

The ADC's serial output is clocked into a microcontroller pin and stored in 3 registers (this is one change from the first attempt at writing the program for this - it originally went straight to bcd conversion), where it is divided by a constant to scale the reading correctly - since the full scale count will be the same whatever reference is used it needs to be adjusted appropriately. Rather than try to do floating point maths, the result is left-shifted a number of times, and then right shifted by differing amounts which are added together to give an integer division. I included an offset to compensate for floating part being thrown away, rather than rounded. Works well on paper anyway. I'll post up the details when I have more time.

The result is then converted to bcd and output serially, including a code for blanking which can be gated off. This is used to control the sign display, which is permanently on, but blanked for positive readings. A clock output is provided for the 4 bit register which outputs to the bcd-7seg converter, and divided by 4 it clocks the digit output.

Auto-ranging is provided by the ADC indicating over-range, to switch up range, or the microcontroller deciding the display needs shifting up a digit, to switch down range. The 16F57 provides enough pins for there to be 4 "function" inputs, which will be mostly used to tell it where to put the decimal point for the non-voltage functions. Thanks to the folks here I have some approaches to creating a latching decimal to bcd circuit for momentary push-button inputs for these.

The actual switching will be done using reed-relays, unless someone has a better idea.

So that's the voltage ranges, auto-ranging and function switching, and display sorted out.

I've started tackling current ranges now, however there seems to be a bit of a problem in that the full scale of 2.5v is rather a lot to be dropping on a current shunt. So call for a x10 amplifier here, or full range control via the microcontroller.

The thing that's got me really scratching my head though is creating an ohms converter. I've found 3 designs on the internet so far, none seem really suitable, so some work to do there.

Ah well, time for bed.