Input circuit to adc

Ok so I've finally got back to doing some work on my little bench dmm project. I was reviewing the help I've received on here so far (thanks again folks ) and puzzling over mixed power supply requirements - when I discovered fully differential op-amps, which I think makes a very good solution.

I can't build anything to test yet because all my electronic stuff is still packed away following the house move, and is likely to stay packed for a couple of months at least.

I've drafted a notional circuit (attached) which I think achieves what I want:
* A permanent un-switched first stage of the input divider between the meter input and any input circuit - so the input is always divided by at least 10 - this means the input voltage will never exceed the ratings for the reed relays I'm using (unless Something Really Bad happens)
* An amplifier to bring the voltage back up and give more practical scaling
* Differential input to the ADC
* Accommodate mixed split/single supply requirements
* Work with + and - inputs

The adc block on the drawing is a dummy just to show how the adc will be connected.
The Vocm input on the fully diff. amp is used to set the mid point for it's outputs to half the adc's reference voltage, so the adc can see both positive and negative inputs.
This particular adc gives best accuracy when Vref=Vcc.
The first op-amp is to give a very high impedance input - I was originally looking at CA3160 (thanks ron)
The second op-amp has differential outputs, but is connected with a single sided input.
The analogue supply voltage is + and - to accommodate + and - inputs
The analogue supply voltage is quite high (yeah ok I need to lower that!) to ensure linearity and for good power supply rejection
The digital supply 0v is connected to analogue 0v at the PSU
I haven't shown any protection.

It simulates OK, but I have some questions:
In principle, is this a good design (ie, what is wrong with this circuit)?
I'm now confused about the input voltage I should be allowing to the adc. My previous situation was that the "-" input would be fixed at +2.5v, so the "+" input could go between +5v and 0v, so the range was -2.5v to +2.5v. With this new arrangement, it looks as though I can use adc differential input equal to the full reference voltage. Is this true? Relevant spec on page 3 of the data sheet (attached). I'd really appreciate someone else looking at this, because all this is still pretty new to me.
Am I better to use the first op-amp as the gain stage, with the fully diff amp having a gain of 1? Or leave it as it is?
Any beginner mistakes I've made?

Thanks in advance

uC ADC protection

Howdy, didn't review ckt, just suggest my std protection on Any ADC input that could get out of spec.
A 10k inline (ADC input impedance >> 10k) to a 5.1V zener in parallel to a schottky to ground. Zener shunts overvoltage, schottky shunts < ground.
This keeps most of my analog fauxpas' from roasting the uC. Good Hunting... <<<)))

With this new arrangement, it looks as though I can use adc differential input equal to the full reference voltage. Is this true? Relevant spec on page 3 of the data sheet (attached).

Click to expand...

As I read it, the operating differential voltage input must be between –VREF/2 and VREF/2, or -2.5V to 2.5V in your circuit; meaning that the usable differential input is not equal to the full reference voltage value of 5V, but +/- half of that. Outside of those limits, an overrange condition will result.

Thanks both. Looks like everyone else has more interesting things to read - either that or my circuit is otherwise perfect! I attached a png of the circuit for those who aren't using LTSpice

I question the need for U1. Why not give the input Op Amp a gain of 10 (classic non-inverting amp with RF=900 ohms and Rjunction=100 ohms), feed its output directly to the ADC + input and ground the ADC - input? The differential input to the ADC will be the same.

This design originally started with a direct connection from the divider ladder to the adc, intended to be an isolated, floating input but I ran into various issues with this, so chose to connect signal ground to the meter's "-" input and keep the highest value on the ladder permanently in circuit, and compensate with an amp. The amp was going to be how you describe, but at some point during my investigations these last few days I discovered op-amps with differential outputs.

The method you describe isn't truly differential and loses the adc's common mode noise rejection, which is important since this meter will I hope measure down to 3μV. It seems the general consensus I've read is that using one of these "fully differential" op-amps is the best way to drive an adc with a differential input, but I wanted to run my initial scribble of a design past the folks here because most of the regulars know vastly more than I do.

It's slow progress because I'm learning as I go.

That's a reasonable train of thinking, but I have my reservations about the maintenance of common mode rejection through the system as a whole since the first op amp is not an instrumentation amp. I defer to others on that matter. A microvolt meter opens up additional considerations, too, that may need to be addressed, such as dissimilar (thermocouple) junctions in the interconnections. That's out of my realm of personal experience.

I know, and I did start off by looking at inst. amps before I'd really settled on how to approach the design, then sort of forgot about them. I'm still on big learning curve with this. At some point (next year, it's looking like) I'll actually build a prototype and see what reality has to say I know there's a world of stuff I haven't got my head round yet - I've been picking bits out of relevant literature to arrive at a basic design, however I think it's not sensitive enough to need to worry about thermocouple effects. I did start looking at guarding though - again not sure if the design is sensitive enough for it to make much difference.

It's not intended to be a microvolt meter per-se, just a meter with microvolt resolution. I'm looking for versatility.

Anyway, you've certainly given me more food for thought and some pointers where I should dig next, ccurtis - I'll certainly re-visit instrumentation amps - not sure if it's worth using one when it's acting as a voltage follower though.

Thanks again