Low noise input protection for a 16bit ADC?

Thread Starter

MikeA

Joined Jan 20, 2013
362
I'm trying to protect the inputs of a 16bit ADC (ADS1115) that will be powered by 5V, and if I understand the spec sheet right, it will have an absolute max input range of -0.3V to +5.3V.

I kind of care about resolution (which will be 0.076mv) but it's not critical to be super low noise, but would prefer to stay under 1mv noise.

What would be a reasonable solution here?
 

Papabravo

Joined Feb 24, 2006
21,159
The value of the resistor will have an effect, since it will be a source of noise. Your 1mv of noise does not specify a bandwidth and so is a bit ambiguous. What is your actual requirement?
 

Thread Starter

MikeA

Joined Jan 20, 2013
362
The value of the resistor will have an effect, since it will be a source of noise. Your 1mv of noise does not specify a bandwidth and so is a bit ambiguous. What is your actual requirement?
Very low bandwidth. No more than 215hz.

DAC is 860 samples per second, so reading from all 4 inputs will never be faster than 215 samples per second per input.
 

Papabravo

Joined Feb 24, 2006
21,159
The 10K resistor will have a noise figure of almost 13 nV/√Hz. at room temperature. We still don't know what the low noise requirement of this application is because the bandwidth of the input signal is unspecified.

Edit: At 215 Hz you will probably be OK. At that frequency, you could also consider putting the input protection on a low pass active filter instead of directly on the A/D converter. In this way the value of the protection resistor will not interact with the input impedance of the A/D converter.
 
Last edited:

carloc

Joined Oct 8, 2018
13
I'd rather have a look at diodes reverse current, Schottky ones can easily go up to 100uA and above. This current into your 10kohm resistor can drive errors up to 1V into your measure.
 

Thread Starter

MikeA

Joined Jan 20, 2013
362
At that frequency, you could also consider putting the input protection on a low pass active filter instead of directly on the A/D converter. In this way the value of the protection resistor will not interact with the input impedance of the A/D converter.
If the A/D has 10MOhm input, will there be much interaction at all?
 

Thread Starter

MikeA

Joined Jan 20, 2013
362
A/D converters don't necessarily have a high input impedance. Some are in the 20K range. As always the datasheet should be your guide.
The datasheet claims 6 to 10 MΩ input.

I drew the circuit above with the Schottky diodes in LTspice and saw there is a large difference near 0V. Some Schottky models had 30+ mV output at 0V input, and some had negligeble uV output. Is this the reverse current mentioned above at play?
 

Papabravo

Joined Feb 24, 2006
21,159
The datasheet claims 6 to 10 MΩ input.

I drew the circuit above with the Schottky diodes in LTspice and saw there is a large difference near 0V. Some Schottky models had 30+ mV output at 0V input, and some had negligeble uV output. Is this the reverse current mentioned above at play?
I'm not sure what you are seeing, since at 0V on the resistor input neither diode is forward biased and neither diode should conduct any appreciable current when reverse biased by less than it's breakdown voltage.
 

dendad

Joined Feb 20, 2016
4,452
Are you trying to read from 0V to 24V range on the input?
A Voltage divider will be needed, with about 7:1 ratio I think.
Not the 10K/1M you have shown.
 

ebeowulf17

Joined Aug 12, 2014
3,307
The datasheet claims 6 to 10 MΩ input.

I drew the circuit above with the Schottky diodes in LTspice and saw there is a large difference near 0V. Some Schottky models had 30+ mV output at 0V input, and some had negligeble uV output. Is this the reverse current mentioned above at play?
Yes, that's almost certainly what you're seeing.

It's worth noting that there are at least two different common applications for Schottky diodes, with very different requirements. Small signal Schottky diodes are useful when you need especially low Vf and/or quick response. Large Schottky rectifier diodes are used in high power rectifiers, where the fast switching speed and low Vf both translate to better efficiency. You can imagine that a diode that only has to survive 30V reverse bias and 200mA of current is built in a completely different way than one that has to handle hundreds or thousands of volts and hundreds of amps of current!

It would be worth looking up some of the part numbers and seeing what you've been simulating. I bet in some cases they are large, high power packages that wouldn't be suitable for your application anyway. Try narrowing it down to small signal Schottky diodes and you'll probably find a more reasonable range of leakage currents. Also, note that in some cases, the diodes with especially low leakage values accomplish that in exchange for very high Vf values, which defeats the purpose of using a Schottky for your clamping circuit.

In my limited experience with I/O protection circuits, I've used BAT42/BAT43 and BAT54 diodes quite a bit, and been happy with them. I imagine there are better options, but they might be worth looking at.
 

ebeowulf17

Joined Aug 12, 2014
3,307
Do you have any idea what kind of issues you need to protect against? What range of inputs will be fed into this circuit? Is it just transient noise spikes you're worried about, or are there higher voltage rails in the system that might get shorted to the ADC input?

The reason I ask all this is because there were questions about how large of a value you should use on the protection resistor (10k in your image above.) A lower resistor value would mean less noise and less voltage-divider action. If you're only worried about brief transients, you can use a small capacitor between input and ground, and use a much smaller resistor value. On the other hand, if you're protecting against a possible 24VDC supply connection, then you need to carefully limit how much current can run through the protection diodes, meaning a much larger resistor value.

This Digikey tech article describes a lot of the considerations that go into input protection, although it's not a perfect match, since it's focused on digital inputs, not analog:
https://www.digikey.com/en/articles/techzone/2012/apr/protecting-inputs-in-digital-electronics

And here's one more specifically aimed at analog input protection:
https://www.analog.com/en/technical-articles/protecting-adc-inputs.html
 
Top