Hi, I am relatively new to Arduino and the Arduino forum.
I had a question about supplying a 2.5 voltage to the Arduino Uno AREF pin to improve the resolution of the 10 bit ADC. The voltage regulator I have is an LM2937 from TI. I see on the packaging that the current output of the voltage regulator is 400 - 500 mA (assuming this is a maximum) and I was just wondering if this was too high for AREF. I am trying to avoid frying my Arduino.
Is there any additional resistors or electronic elements that I should attach between the LM2937 output and the AREF pin? If so, how would this affect the AREF voltage? I am aware that there is a 32kOhm internal resistor between AREF and ground. My reasoning behind using this regulator is an alternative to the using a simple voltage divider from the 5V source with two equivalent resistors.
Here is my anticipated schematic. I was planning on using the 5V Arduino supply voltage for Vin, connect the ground pin to the ground pin next to the AREF pin, and the Vout to the AREF pin.
When the Arduino first powers up, Aref is connected to Vcc until your code changes it to external. So I think it would be a good idea to put a 1k resistor between the output of the above circuit (after the capacitor Cout) and then a 0.1uF capacitor from Aref to ground.
. I see on the packaging that the current output of the voltage regulator is 400 - 500 mA (assuming this is a maximum)
Yes, that's the maximum.
The relationship between voltage, current, and resistance (or impedance) is described by [u]Ohm's Law[/u]. "Resistance" is the "resistance to current flow"... More resistance = less current. And, higher voltage = higher current (assuming the same resistance or impedance).
The AREF input is high impedance and very little current will flow. (I haven't checked the datasheet to find the exact impedance.)
Are you saying to put a 1k resistor in series with Cout (10uf) to ground? And then add an additional 0.1 uf cap between the AREF line and ground?
Instead of connecting the voltage regulator (or reference) directly to the Arduino, connect it with a resistor.
I had a question about supplying a 2.5 voltage to the Arduino Uno AREF pin to improve the resolution of the 10 bit ADC.
What voltage(s) are you measuring?
For 1V or less you can use the optional-internal 1.1V reference. I've got an auto-ranging project where the software automatically switches to 1.1V when the ADC reading is less than 200. Then if it gets a reading that maxes-out the ADC at 1023, it switches back to the 5V reference.
I will look into getting a voltage reference , but for now all I have is a voltage regulator. Presumably I can use this to see if it works by connecting it in series to the AREF. I am assuming I can keep the same configuration when I get this voltage reference. Wouldn't this resistor affect the analog voltage reference?
Am i safe to use the voltage regulator in the mean time, with a series resistor and stabilizing 10uF cap?
FYI
I am measuring two voltage signals from LMT87 temperature sensors and converting them to temperatures. The range of temperatures i am interested in is <2.5 V which is why I have been looking to improve the ADC resolution using the AREF.
Since I am interested in measuring the temperature difference between two sensors, an alternative I have been considering is using a differential amplifier to take the difference in voltages, which in my application would be <1.1V. In this case I could use the internal 1.1 V Uno analog reference, but at this point I will lose the information of each individual temperature value.
BTW, I like your idea Doug of autoranging to determine the optimal ADC setting based on the voltage input. I will look to upgrade my software in future revisions, especially if I pursue the unity gain differential amplifier route, in which case I can have one signal < than 1.1V, and then two separate streams containing the individual voltages from each temperature sensor using a 2.5V analog reference.
had a question about supplying a 2.5 voltage to the Arduino Uno AREF pin to improve the resolution of the 10 bit ADC.
Just so you know you are not improving the resolution, that is fixed at 10 bits. What you are doing is reducing the size of each voltage step. These two things are not the same thing.
The logical thing, to me, would be to scale the temperature of interest with an opamp between the temp sensor and analogue input.
That way you can also remove the 0.5v offset @150C, and use the max. range of the A/D.
e.g. you might only want -10C to +92.4C (0.1C resolution).
Temp difference is easy in software.
Leo..
Grumpy_Mike:
Just so you know you are not improving the resolution, that is fixed at 10 bits. What you are doing is reducing the size of each voltage step. These two things are not the same thing.
I am aware I am not improving the actual resolution of the ADC, I was more so referring to the temperature resolution, (or "effective" resolution") improving as a result of lowering the AREF and in turn reducing the side of each voltage step. But thanks Grumpy Mike.
Wawa:
The logical thing, to me, would be to scale the temperature of interest with an opamp between the temp sensor and analogue input.
That way you can also remove the 0.5v offset @150C, and use the max. range of the A/D.
e.g. you might only want -10C to +92.4C (0.1C resolution).
Temp difference is easy in software.
Leo..
I am only concerned with a narrow window of temperatures ~ 20 - 37C which means the largest voltage difference I can have is ~230mV (assuming I do not adjust the gain via an op-amp). What benefit would I have by scaling (adjusting the gain)? Regarding software, I am using Matlab's Arduino package and am relatively solid. I was more interested in the electronics help.
Based on crossroads suggestions I have ordered an ADR291 and LT1019. Can anyone explain the difference between series and shunt voltage reference, and tell me which of these configurations would best match my needs for supplying a steady 2.5V to AREF.
Scaling with an opamp.
Say you take 25C as your center point.
That is according to the datasheet 2.298volt from the sensor.
Feed that voltage (~2.3volt) into the + input of an opamp, or to the + input of several opamps.
That can be done with a voltage divider, say 1k/2.3k (or with a trimpot) from the 3.3volt rail.
That is now the pivot point. 2.3v from the sensor compares with 2.3v from that ref voltage, if the opamp has full feedback (out to - in), the opamp's output will be 2.3v@25C.
The opamps output will also be 2.3volt@25C if you use feedback (gain) resistors.
e.g. 22k from sensor to +in and 220k from out to +in gives you 11x gain.
That gain results in about 30 digital steps per degree C.
Leo..
P.S. You might have to find an opamp that works on 5volt and has rail to rail outputs.
Crossroads is correct, my concerns are groundless.
I like that. It sounds like a really smart way to say "I was wrong" , which is always painful. ;D
I'll have to remember that for the next time I make incorrect or inaccurate statement in a post,
(which I have been known to do).
We can always count on Crossroads to set the record straight. ;D
I am only concerned with a narrow window of temperatures ~ 20 - 37C which means the largest voltage difference I can have is ~230mV (assuming I do not adjust the gain via an op-amp).
I'm going to go out on a limb here and say that the preferred way to deal with small inputs is to make them big , which is why instrumentation circuitry is so often based on amplification (using an op amp amplifier like WAWA suggested), thereby changing a scenario of small differences in small numbers to large differences in large numbers. A gain of 100 to change the voltage difference to 2.3 V , accomplished with a two stage amplifier , each stage having a gain of 10. (single stage gain of 100 is a bad idea).
In addition, I might suggest a window comparator that sends a logic signal to the uP when the voltage is within the desired range, so you can ignore it when it is out of range and read it when it is in range, eliminating the need to constantly poll it. I would use an LT1215
Some number crunching.
The sensor, with the range the OP wants, produces already 229mV between those two temps.
20.8 times amplification would be the absolute maximum to stay in range of the A/D converter.
I sugested 11x to start off with. That can easilly be done with one opamp.
This gain is already more than practical. It gives you 0.033 degrees per digital step.
Leo..
Wawa:
Scaling with an opamp.
Say you take 25C as your center point.
That is according to the datasheet 2.298volt from the sensor.
Feed that voltage (~2.3volt) into the + input of an opamp, or to the + input of several opamps.
That can be done with a voltage divider, say 1k/2.3k (or with a trimpot) from the 3.3volt rail.
That is now the pivot point. 2.3v from the sensor compares with 2.3v from that ref voltage, if the opamp has full feedback (out to - in), the opamp's output will be 2.3v@25C.
The opamps output will also be 2.3volt@25C if you use feedback (gain) resistors.
e.g. 22k from sensor to +in and 220k from out to +in gives you 11x gain.
That gain results in about 30 digital steps per degree C.
Leo..
P.S. You might have to find an opamp that works on 5volt and has rail to rail outputs.
Hi Wawa, I like your suggestion.
Is there anyway you could draw out what you were saying in a circuit so I can visualize it better. Are you using the op amp as a differential amplifier, because you mentioned both a reference voltage (equivalent to center point going into the noninverting input, and the voltage signal from temp sensor going to inverting input.
Wawa:
Some number crunching.
The sensor, with the range the OP wants, produces already 229mV between those two temps.
20.8 times amplification would be the absolute maximum to stay in range of the A/D converter.
I sugested 11x to start off with. That can easilly be done with one opamp.
This gain is already more than practical. It gives you 0.033 degrees per digital step.
Leo..
Are you suggesting to go back to the 5V analog reference, or stick with 2.5 from an external analog voltage reference?
Also, I should probably state again that I am using two different temperature sensors (both LMT87s) and am interested in the difference in temperature. So with your suggestion I am assuming I would need multiple op-amps?
