Go Down

Topic: Measuring Voltage without common Ground (Read 2 times) previous topic - next topic

michael_x

My heating system has a NTC type outside temperature sensor.
With my multimeter I can measure about 2.5 V DC between the two wires disappearing in the wall.
If I disconnect the sensor wires I measured a resistance of about 1.5k at 5°C decreasing to 1.1k at 20° between the sensor wires.

Now I want to read this sensor in parallel to my heating system, without disturbing it too much.
Accuracy to about 1°C (about 25 Ohms/K, non-linear) is fine for the heating system and me.

I'm rather new to electronics and want to learn by doing.
Long term project is to read more data ( Hot water temperature, count burner / pump running time)

I fear I should not directly connect one of the wires (the one my meter reads as negative) to Arduino GND and the other one to an Arduino analog input, as I need to use a common GND for digital signals, in a parallel project, to be combined later. And I fear the hot water sensor might not have a common leg with the outside temperature sensor neither (not yet checked). Is this fear (Don't tie external signals to GND) correct?

I think there's an option about having Analog Ground separate from Digital Ground, but I can't find details how to separate them, and if yes, if that would help. Any hints, even if not directly usable here (I want to learn)?

I think I found an electronic Ground of the heating controller. Should I read the 2 voltages relative to that, and build the difference in arduino code?

I also ran across opamps like LM358 ? Is it possible to connect the sensor signal via high resistors and a loopback resistor, and then get the sensor voltage difference at the LM358 output pin relative to its Vss (GND)? Is this an intended usage of such amazingly cheap opamps?
If yes, asking for discrete circuits does not make sense, correct ?

kf2qd

An op-amp would be appropriate here. It would provide a High Impedence to to the sensor and a voltage output you could use with your Arduino. It could also provide some signal scaling so you could have a higher number of bits to work with at the A2D in the Arduino.

michael_x

#2
Mar 27, 2012, 05:37 pm Last Edit: Mar 27, 2012, 06:09 pm by michael_x Reason: 1
Thanks,kf2qd,  for the confirmation.
I already found this is my use case:


Scaling is done by relation of the pulldown / loopback resistors involved, I guess ...?
Or do I need to add appropriate zener-fixed voltages to come into an output range of 0 to 5 V ?

I'll order a pair, google more,  and experiment...  

Thanks. ( More hints are welcome, I'll keep an eye on this topic ;)

michael_x


Until it arrives, I play with an opamp in LTSpice.
I think I "wired" it right. The voltage is measured across a resistor with a current varying from 1 to 2 mA.
However, the output changes from about 2.4 to 2.8 V only. It does not really matter if the opamp resistors are 100k or 1M each.
If they are different, the output ratio is even less.
Do I just have a wrong LTSpice opamp model, and do not need to worry in real life later?
If yes, can I get ( or make ) a better opamp model somehow?

Quote
provide some signal scaling so you could have a higher number of bits to work with at the A2D in the Arduino

I hoped to get a voltage change from 1 to 2 V, and be able to scale that from 1 to 4 V or even 0.1 to 4 V  ;)

Could someone please help me with this ?!

michael_x

A slight error in my circuit.
And there's a "simple opamp" at ltspice (which does not care about outputs < 0V or > 5V)
Quote
I hoped to get a voltage change from 1 to 2 V, and be able to scale that from 1 to 4 V or even 0.1 to 4 V 

Could someone please help me with this ?!


1 to 2 Volts from an input varying from 1mA to 2mA is easy (in LTSpice theory). Scaling by a factor is easy as well, but does not help too much as 2 to 4 Volts does not give more accuracy. ( I could also modify the Vref ). 

Any expert on how to cut off a constant offset in an opamp input ? ( My idea to scale 1..2 to 1..4 )

Grumpy_Mike

Remove R4
Make R2 into a pot
connect the +ve of the op amp to the wiper of the pot.
The pot will then control the DC offset of the signal.

michael_x

Thanks for having a look, Mike


Remove R4
Make R2 into a pot
connect the +ve of the op amp to the wiper of the pot.
The pot will then control the DC offset of the signal.

Unfortunately, the resistors are named differently on my different pictures. Are you talking about the recent one ?
In the initial picture, it would mean to see 'R1 ... +ve ... R2' as a pot ?

Grumpy_Mike

I was talking about the schematic in reply #4

michael_x

Mike, I recently learned that both input resistors to +vin and -vin need to be the same size, to get the voltage difference right.
In the external link of reply #2 they are named R1 and R3. Then, if loopback and pulldown are the same (R4, R2), the ratio R2/R1 = R4/R3 is the gain.

According to that circuit, R2 on my reply #4 is on the wrong side of R4, and R5 is wrong anyway?
Seems I'm slightly confused ? 

Now, how to add (subtract rather) an offset, in the same opamp?
Or simply use the second half of the LM358 to do the offset scaling, once I got a common ground?

Thanks for hints, helping not to experiment/simulate into the wrong direction.

Grumpy_Mike

Quote
recently learned that both input resistors to +vin and -vin need to be the same size, to get the voltage difference right.

No you are confusing what you are doing with a differential amplifier.

michael_x

Quote
confusing what you are doing with a differential amplifier

Probably I'm not doing it right, but isn't that what I want to achieve: Measure a voltage difference, wich is not bound to my GND ?
And do some scaling like
   
    Vo = m * (V+ - V-) - b

I was fascinated when reading about opamps and thought it's so easy ...   


Grumpy_Mike

Quote
I think there's an option about having Analog Ground separate from Digital Ground, but I can't find details how to separate them,

It is not the ground but the supply, while you can separate them they are not isolated, and both need to be fed with 5V. They are separate to allow you to add some decoupling to try and isolate the effects of digital switching on the sensitive analogue circuits.

Quote
Measure a voltage difference, wich is not bound to my GND ?

That is not possible, the two voltages will have a path to ground through the resistors on the front end of the op amp. You need some sort of common reference and one of the wires has to provide this.

michael_x

Thanks Mike, for going back and rereading this thread ( and thanks for all the responses in general : )
Quote
the two voltages will have a path to ground through the resistors on the front end of the op amp. You need some sort of common reference and one of the wires has to provide this.


I found this schematic, and I'm not afraid of pulling all to Arduino GND via 1M resistors.
In practice, even 200k would not hurt or disturb the measurement too much, I think. 
I measure a VR of about 2V across that outside temp sensor, which is showing  1 .. 2 k  (NTC) when separated from my heating system electronics.

haydent


Docedison

The Op-Amp circuit drawn in the last Drawing (originally from the National Semiconductor Op-Amp Application notes Vol1) is a unity gain amplifier where R1/2 set the operating point of the op-amp @ 1/2 V(in) R is the sense resistor and R3/4 set the gain of the amplifier R1/3 should be equal in value so that 'equal' (Ibias) current is available for both the inverting and non inverting inputs of the amplifier and R4 sets the Gain of the stage. Gain is equal to Ri/Rf or R3/R4 and If I remember correctly in the text preceding and following that illustration the terms are explained (It's been about 20 years since I last used that circuit). One note here: The output of an LM385 (or any other Op-Amp in that family Vout can never be greater than Vsupply -1.2V or less than Ground +1.2V. If Vout must swing between ground and Vsupply a "Rail to Rail" type of op-amp is required (ex LMC622), all resistor values will remain the same for identical operation.

Doc
--> WA7EMS <--
"The solution of every problem is another problem." -Johann Wolfgang von Goethe
I do answer technical questions PM'd to me with whatever is in my clipboard

Go Up