I have a question regarding the measurement of thermistors using a differential ADC like the MCP3421 that is unipolar. Attached are two circuits, one the bridge circuit that you can find in the MCP3421 literature (http://ww1.microchip.com/downloads/en/DeviceDoc/22003e.pdf) (see page 28), and the one I modeled and used in my application. Try as I might, I cannot figured out what the advantage of the bridge circuit is other than linearizing the output somewhat. In addition, my research suggests that the bridge-based output loses a bit of information due to the voltages that the circuit produces with a unipolar ADC. So why advocate the bridge circuit over the 'direct' approach I show on the right, where both ADC inputs remain positive, any excitation voltages disturbances will affect both inputs equally, and the voltage difference vs. temperature output is just a function of a couple of lookup tables?
any excitation voltages disturbances will affect both inputs equally,
False assumption. Read how voltage divider works on wiki.
Hi Magician, and thanks for the reply. Having gone over the section, I am still baffled re: your response.
The most compelling reason I did stumble across (elsewhere) to use a bridge circuit is the possibility of using a three-wire system for the RTD (http://www.globalspec.com/reference/10952/179909/chapter-7-temperature-measurement-rtd-measurement-circuits) and hence eliminate errors due to line losses. Is this what you were referring to?
I was referring to circuitry shown on the right. Voltage variation is not equally distributed between ADC- and ADC+. R2 makes difference, for one input:
Vout = Vin x R1 / ( R1+R2+R3); and for another
Vout = Vin x (R1 + R2) / ( R1+R2+R3).
Thanks for the clarification. I appreciate that and I didn't express myself well. I am not worried about the overall excitation voltage because I'm using a 0.05% accurate voltage source (series reference LM4132A @ 2.048V). Rather, I was hoping that any changes in excitation to the thermistor wires due to EMI, etc. would be canceled. However, I now realize that accounting for line losses to and from the thermistor is going to require some careful rethinking. Thanks again.
So why advocate the bridge circuit over the 'direct' approach I show on the right,
I think it's because the bridge configuration presents a balanced impedance to the inputs of your differential amplifier. Any common mode signal will just be rejected and only the difference (i.e. the change detected by your sensor) will be amplified.
Imagine this, if you have any "external" noise/interference that got picked up, chances are the noise will be picked up by both lines, and therefore should/will be rejected at the inputs of your differential amp. Only the true signal (i.e. the data you want),will be accepted by the diff opamp. So it has better noise immunity.
Here you go, see this... page 2.
chances are the noise will be picked up by both lines, and therefore should/will be rejected at the inputs of your differential amp. Only the true signal (i.e. the data you want),will be accepted by the diff opamp. So it has better noise immunity.
That's exactly why.
edit a little bit ;)
chances are the common mode noise will be present at both inputs with the same amplitud and therefore should/will be rejected by your differential amp. The diff amp will not amplify (or very little) if there is no difference between the inputs. Only the true signal (i.e. the differential signal, the signal you want), will be amplified by the diff amp. So it has better noise immunity.
This subjects will help.
-Common Mode Rejection Ratio (CMRR) of the diff amp.
Hi and thanks for your answers!
I understand the need for common mode noise elimination in the context of very small voltage measurements for things like strain gage's. They require serious amplification so eliminating as much noise from the circuit is a good thing.
On the other hand, the only times a thermistor will feature a resistance remotely close to the 'opposite' resistor in the bridge is within a very narrow temperature band. For example, the NTCs I have been using feature a range from >10k to 150 Ohms for the temperature range I am interested in covering. Not sure how a bridge circuit can have impedances that are close to balanced unless you employ a variable impedance resistor in the bridge? Or perhaps a ladder of resistors in series that are shorted by transistors as needed?
FWIW, I got that second circuit 'inspiration' from LTC with its easy drive ADCs. See this design note (http://cds.linear.com/docs/en/design-note/dn379f.pdf) as an example.
Not sure how a bridge circuit can have impedances that are close to balanced unless you employ a variable impedance resistor in the bridge? Or perhaps a ladder of resistors in series that are shorted by transistors as needed?
Yes, that's how you will need to balance your bridge (Zeroing) for the temp you want to establish as reference. Ten turn precision pots are regularly used for that. Another way is to set an offset voltage level added to the diff amp. If the signal is converted to digital, corrections can also be done by adding or subtracting a "DC voltage" value to the samples to compensate for the difference and zeroing the bridge.
Yes, that's how you will need to balance your bridge (Zeroing) for the temp you want to establish as reference.
What is important, the resistors/impedance used in the legs of the wheatstone bridge could be of different RANGE - ie. in one leg ~100ohm, in the other 100kohm, end you will get the same results.
And, we talk impedance, so in one leg you may have resistors (or a potentiometer) and in the other resistor and capacitor/inductor, again even with quite different range, and with AC excite voltage you get again the same results.
So with a single value of the potentiometer in the first leg you may compensate to zero the thermistor of ANY value in the second. Used for precise measurements for maybe 200 years already..
So with a single value of the potentiometer in the first leg you may compensate to zero the thermistor of ANY value in the second.
Yes, of course he just needs to do it once. If the "zeroing" is constantly done then the bridge will always be balanced at any temp with no differential output. The ten turn pot is to allow precise adjustment and recalibrate for drift errors from time to time if necessary. Another way is to use 1% or lower tolerance resistors with high stability.
FWIW, I got that second circuit 'inspiration' from LTC with its easy drive ADCs. See this design note as an example.
And in their App Note, they also say it's for High Impedance sensors.
Delta-sigma ADCs, with their high accuracy and high noise
immunity, are ideal for directly measuring many types
of sensors. Nevertheless, input sampling currents can
overwhelm high source impedances or low-bandwidth,
micropower signal conditioning circuits. The LTC®
2484 family of delta sigma converters solves this problem by
balancing the input currents, thus simplifying or eliminating the need for signal conditioning circuits.
So if you're using their LTC2484 part #, with a high impedance sensor, then you can use their simplified circuit.
Looks like their LTC chip does the balancing.
The bridge configuration is more "generalized."
So the good example for arduinists would be:
1. use ie. 10k potentiometer in the first leg of the bridge
2. mount a pointer across a dial/scale on the pot. rotor
3. couple the pot. rotor with a servo
4. use the op-amp as a comparator
5. zero the bridge with help of the servo
6. read the temperature on the pot.'s scale