Voltage scaling and ADC protection

[slanina]

Hello all,

I am desiging a circuit that will interface with the Allegro A1301 linear Hall sensor.
This sensor works at 5V and outputs Vcc/2 (2.5V) in neutral case - no magnetic field.
It has a sensitivity of 2.5mV/G.

I will interface this sensor to an ARM Cortex MCU that has a 10-bit ADC 0-3.3v range on board.

At 3.3V with 10-bits I get 3.22mV per ADC step, which means my sensor has more resolution than my ADC.

I am interested in measuring A1301 only the sensor output values between 2-3V, thus corresponding to 200 Gauss change either way and then I would scale this sensor output range to the full ADC range prior to conversion, so I get good resoultion.

I have designed a simple circuit for this:



Due to standard resistor sizes being used I am OK with Vref to be 1.9V so my range will be slightly more than 2-3V.

How can I make sure that input to ADC is not over 3.3V? Because even though I am interested in 2-3V range, it can happen that sensor output goes to 4V or more (in case there is some strong magnet close by) which would fry the ADC. Similarly, sensor output can drop to 0, which would make op amp voltage output negative.

Should I feed sensor output first to a voltage divider (10k/18k) and then to this circuit (with different resistor values now) or should I use clamping zener diodes before the ADC as suggested on some other threads ? However, I would not like to add noise to my signal.

As for constraints - there is single power supply (5V and 3.3V is available).

Any suggestions? Thanks.

 

[FvM]

A simple way to keep the ADC input voltage range without dedicated clamp diodes would be to power the OP with 3.3V, isn't it?

 

[slanina]

A simple way to keep the ADC input voltage range without dedicated clamp diodes would be to power the OP with 3.3V, isn't it?

Yes. But my sensor has to be powered with 5V and as such can give more than 3.3V output.
So OpAmp input can sometimes reach well beyond 3.3V which is not allowed per opamp datasheet.

 

[FvM]

So OpAmp input can sometimes reach well beyond 3.3V which is not allowed per opamp datasheet.

It's recommended to read the "Input voltage and current " paragraph in OP datasheet thorougly. In a short, it's no problem to drive the OP inputs into diode clamping as long as the 2 mA current limit is kept. This can be easily achieve by respective resistor dimensioning.

 

[slanina]

It's recommended to read the "Input voltage and current " paragraph in OP datasheet thorougly. In a short, it's no problem to drive the OP inputs into diode clamping as long as the 2 mA current limit is kept. This can be easily achieve by respective resistor dimensioning.

There they describe internal clamping operation by the diodes to ground. But for going above Vdd there should be external diodes as shown in the figure in the datasheet.

It looks like normal diodes; but I see people use 3.3 zener diodes in other schematics.

As for current, I have 10k resistor in series to the opa input. With maximum sensor voltage of 5V this gives 0.5mA of current which is <2mA.

But wouldn't it be better to clamp the opa output with a single diode prior to ADC and power opa with 5V? This way I only need to worry about 1 wire and that after all scaling and amplification being done already.

 

[FvM]

Yes, you are right. MCP601 has unusual single side clamping only. But with the resistor values in your schematic, maximum Vin+ voltage is 3.6V, so the maximum ratings are still kept without additional clamp diodes. 3.3V Z diodes are probably a bad idea because they start to conduct above e.g 2 V, causing severe voltage errors.

 

[slanina]

Since I am designing right now, I guess I will go ahead with some other op-amp which is rail-to-rail and single supply and has both clamping diodes.
I selected MCP601 because it is DIP and cheap.