Op Amp / analogue help needed please

[max_torque]

I have a pressure sensor, the output of which is ratio metric to it's supply voltage, within a certain supply voltage window (approx 4.5 to 5.5 volts)

I wish to read that sensor into a sperate external device, which has its internal (precision) analogue Vref set at 4.096v ie i cannot power the sensor directly from this voltage as it is too low.

I am trying to avoid is having to have a precision voltage reference in the pressure sensor device,ie to keep the overal system ratiometric to the 4.096v refference in the external device


I can supply the pressure sensor with a seperate "Bulk" power feed (at around 11 to 12v) and i can break out the 4.096v Vref to the pressure sensor to act as a reference voltage.

It is not a particularly high precision device, i'd aim for around 1% accuracy or thereabouts


Options seem to be:

1)

use the 4.096v reference to generate a nominal 5v rail in the pressure sensor device, (that rail is now dependant upon the 4.096v reference), take the power for this rail from the 12v bulk supply - ie use an opamp with some 0.1% resistors to gain the 4.096vref up by 22%, then use a resistive divider with the same resistor values to drop the 0-5v output of the pressure sensor back down into the VREF range (ie divide by 22%)

2)

Use a non precision 5v LDO off the 12v bulk power, power the pressure sensors, somehow compare that 5v rail and the pressure sensor output to get a ratiometric value, then use the 4.096v supply to scale that output to that range.  This has the advantage that it would work with pretty much any VREF voltage potentially making the pressure sensor device more useful in future (ie i could use it for 3.3v VREF projects etc). Not sure how this would be done tbh in terms of amplification etc


3)

Another way i haven't thought of.....



So, analogue guru's, whats the best way of doing this???   

[ArdWar]

ratio metric to it's supply voltage

Use ADC with external Vref instead and use the sensor supply voltage (kelvin connected and buffered if sensor is far away ofc) as reference. No need to generate two precision voltages, or any precision voltages for that matter.

[Terry Bites]

If your sensor is truly ratiometric it will probably work fine at your lower ref voltage. Do the experiment. Lower ref volts usually only alter the scale, you fix that in software.
 If not you can rescale the ref you have and buffer it. You dont need pricey resistors, just a tweak with some regular types will do. They drift in abolute value but thats ok as these.
The pots are muliturn cermets. These circuits rely on resistor ratios to set the voltages. Its not uber super precise but plenty good enough for your 8bit spec. Spend your hard earned on 0.1% if you like!

Ideally you want to derive the bridge exitation from the ADC, if not a simple adjustable ref may be good enough.
A TL431 may well be good enough for your application.
See TL431 datasheet for applications - you may (will) want a current boosted output to supply the sensor bridge current.

Avoid long wires between the sensor and the ADC. You'll often see an instrumentation amp used to extract the signal from the bridge.
Its a good strategy and it rejects noise common to both nodes. If your ADC has a differential input then life is simpler and you can take full advantage of the ratiometric nature of your sensor without additional circuitry. If you are in an industrial or automotive environment you'll need to condition the 12V to keep your circuit safe.

Here are a few of practical way of buffering and rescaling or creating your ref.  In all cases the drift will be much less than 1% over supply volt variations (within reason) and temperature. A budget precision opamp is a good idea. op77 and op77 are industry, jellybean, standards. TL431s are ubiquitous. They're dirt cheap and incredibly useful. Get yourself a bag.

Have a look at AN43 and AN96 from linear technology for some useful ideas on sensor interfaces. Also https://www.ti.com/lit/pdf/sbaa532?keyMatch=SBAA532