Unity Gain Amplifier / Voltage Follower - Selecting resistor

The unity gain config has no resistors, it's a X1 as you see it.
The network on the output there isn't for gain.

So, from figure one and the many examples on Google..

Supply to Opamp - 5V from Power Supply.
Ground common to device being monitored.
Monitored Input to Non-inverting + of OpAmp
Input of Non-inverting - of Opamp tied to Output.
Output has no capacitor or resistor as in Figure 1 - it just goes straight to Analog In of arduino?

Sounds too simple?

That figure is showing the testing network they used to get the figures in the table. The R and C are dummy loads for the opamp during testing, just ignore them.

But you're right about it sounding too simple, these opamps won't work from a 5V supply. They are characterised for +/- 15V (which means the datasheet is very coy about saying much about their performance at lower voltages). One graph suggests they might just work at +/-5V supply, but they are certainly not rail-to-rail and won't work from a single +5V supply.

So I need another component then ?

I want to have the 'no current load' on the circuit being monitored - and ideally want to do it from the same supply as the circuit being monitored (which should be 5V).

Any suggestions on an Op-Amp capable of producing that from +5 only!?

The plan is as "CrossRoads" previously posted here: http://www.arduino.cc/cgi-bin/yabb2/YaBB.pl?num=1293425807/21#21
Finally making some ground on this - but yeh, he said single power supply there ( comprehension fail!).

Another thought is some sort of voltmeter circuit - I need it to be high impedance though - just like a multimeter - so it doesn't interfere with the readings.

The voltmeter would need a 0-5V scale (since the most the sensor can go is 5V and gnd when short cct.

I don't really want to bring a negative supply into the equation, so I think I need a rail-to-rail opamp one that can go right close to ground (the measurements we are interested in will be in the range of 0.7 ? to around 3V when cold. I was reading that Op Amps don't typically output that close (though rail-to rail does?)

And I need it to do that with 5V input and single supply (5V and GND).

TLV2231 ? Bad choice?

LM358 and LM324 are inexpensive and commonly available "single-supply" op-amps, and they're good with inputs down to Ground. Their outputs cannot go > Vcc - 1.5V, but that wouldn't be a concern if powered from/by tapping arduino +V_in. They have wide supply ranges.

Ahh, VCC(5V) - 1.5V = 3.5V (so I can read from 0 to 3.5V)..

I thought of LM358 (due to how common it is), but can't find in the datasheet where it is high impedance (i.e. like the JFET input in the TL072).

I need the high impedance so that we don't take current from the circuit being monitored.

LM358 might work, but I think rail-to-rail would be better (Gnd to 5V on input = Gnd to 5V on output).

That 3.5 - 5V would equal just below 21 deg in temperature - don't want to lose that much of the range.

Oh I see - Most of the Op Amps can only produce up to (VCC - 1.5V).
To get the full 5V - I need a supply of 6.5+ Vcc.

What are the downsides of using a DC-DC to get 6.5 from 5Vdc ?

5Vdc from the existing controller (mains powered), to the arduino 5Vdc, and then take from the 5Vdc to a DC-DC converter to get 7V with minimal power (OpAmp = low power).

tocpcs:
I need the high impedance so that we don't take current from the circuit being monitored.

The ADC in the Arduino has a typical input resistance of 100Mohms, and a sample capacitor of around 6.5pF. So it disturbs the circuit being measured very little. Unless the input voltage is changing rapidly and you want to follow those changes, you may not need an amplifier at all.

If you want an inexpensive rail-to-rail op-amp in a non-smd package, here is a suggestion: http://uk.rs-online.com/web/p/op-amp/0403165/.

Heh, I was thinking of that - the multimeter doesn't affect it - I can go to the analog pin of the MCU being monitored, and the - (gnd) of the sensor, and get a reading - it doesn't appear to drop it at all.

It appears the temperature sensor on the circuit being monitored has a diode on the grnd (looks like a shottky so I think -- lightning protection?)
And the positive is a 10k resistor.

Is the analog inputs really in the megaohms of resistance though?
How can I make my connection high impedance ? Does putting a 1Mohm resistor at the connecting point accomplish this (i.e. stops it drawing current?)

If I connect, I'll be loading down the temperature circuit (if it's wrong and the impedance isn't greater in the duino), I think that'll just cause it to think the sensor is 'hotter' than it is, until the current draw stops (as current increases, voltage decreases, so the temperature reading to the controller would be 'hotter').

Originally all opamps were designed to operate on +/-15V or +/-12V - this was universally accepted as a given.

Only more recently have people realised that this is inconvenient since the majority of electronics is digital and 5V was a more convenient supply. Original PCs had a +12V and -12V for several reasons (disk drive motors, serial comms and op-amp/analog interfaces). But now the emphasis is entirely on personal electronic devices where there is no need for such supplies, so a whole raft of low-voltage opamp designs appeared.

Most opamps that can run at 5V don't perform their best at that low a voltage, note, particularly frequency response is compromised. Most rail-to-rail opamps can't actually drive more than a few microamps right to the rail too, so in general you expect even a rail-to-rail opamp to only handle signals from 1.0--4.0V or 0.5--4.5V (on the output side that is). The input behaviour at at least one rail is usually fine (many opamp inputs can handle -0.3V or so on the inputs for instance owing to PNP or JFET input stages.

For the purpose of this discussion impedance means the same thing as resistance.

What's needed is a comparatively low impedance connection to the analog in, that way practically all of the voltage is across the high impedance.
If you have a 1000? resistor in series with a 100M? resistor, practically all of the applied voltage is where? It is across the 100M?, 1000? : 100M?.

You missed the part where I said to run the op-amp from "Vin", but you can't do that if you're stuck on running from 5V.

I caught that - VIN will be empty - the plan is to not use the voltage regulator and bring in +5 from the already regulated +5 at the other controller - (saves getting an additional supply and then wasted energy from that).

I was going to go and chuck a 9V supply in, but if I'm not going to have an issue wiring the input to the arduino (i.e. I don't want the arduino loading down the sensor), then I won't worry about the Op-Amp.
The entire point of the op-amp is so I don't load it down - if the analog inputs are indeed 100Mohm then I would hope that's fine? The sensor reading won't get loaded down, it'll be accurate and I'll get on with reading the voltage...

I like the idea of having a wee bit o' resistance betwixt the two.

tocpcs:
It appears the temperature sensor on the circuit being monitored has a diode on the grnd (looks like a shottky so I think -- lightning protection?)
And the positive is a 10k resistor.

It sounds to me that the temperature sensor arrangement has quite a low source resistance, so you can connect it directly. Can you provide a link to the datasheet for the sensor?

If it's on a long lead then you might want to include a 10K resistor between the sensor and the analog input pin, and a 0.1uF capacitor between the input pin and ground.

dc42:
It sounds to me that the temperature sensor arrangement has quite a low source resistance, so you can connect it directly. Can you provide a link to the datasheet for the sensor?

If it's on a long lead then you might want to include a 10K resistor between the sensor and the analog input pin, and a 0.1uF capacitor between the input pin and ground.

Nah no datasheet for it - though it's just a NTC 25 deg C 10k sensor.

Long lead --- from the sensor to the controller - yes, from the controller to my controller? Not so much

Can't hurt to have 10k resistor and the cap - so I'll add that.

Will test it today, connect the arduino direct and measure sensor voltage before and after connection, with any luck it won't change much at all..

No impact at all..

So does that mean the arduino and cable has a higher resistance/impedance than the jumper wires / fingers, etc that were the way?
Doesn't make sense at the moment, in a resistor divider, you have one resistor (the sensor), and the other a known value and you read the mid point of this.

What I find is if I read the voltage between the two cables that go to the sensor (i.e. think of a thermistor, you would wire it up with one side to the ground, and the other to the resistor, and measure the mid point of what you get back).

I wasn't expecting to find the same voltage between the two though - if anything, I'd have expected to find the 5V being supplied to the thermistor.
The diode on the ground side of the sensor is odd too.

Is there perhaps another reason for it to not care ?

I do recall that the controller reset / had randomness when I tried measuring once a while back, but I think I tried measuring resistance of the thermistors in circuit - this would mean I am taking current from the circuit, right?

What it all means is that a high-impedance circuit placed in parallel with a device under test has no more effect upon the D.U.T. than a voltmeter (which is a high-impedance circuit placed in parallel with a device under test).
[There is, in fact, (to split hairs) some negligible effect from that because it's a high or very high impedance, but not an infinite impedance.]

Back in the bad old days, voltmeters weren't the high-impedance wonders that they are today.