Input impedance of ADC10s, or something else (like ADCsetup) on a 40X2

[Eclectica]

Hi,

It may be that I'm worrying about nothing here, but just in case .

I have a load of NTC like this https://www.circuitlab.com/circuit/btt9cg/100k-ntc-circuit/ connected to ADC channels 19-27. There are a load of digital inputs and outputs on the other pins.

Thinking I have set up the ADC ok, there is always a 10mV or more drop across the 22k to the PICaxe pin with my Fluke 1587 DMM.

Surely the input impedence of the ADCs in the 100s of Megs or even a gig?

ADCset up is this:

adcsetup2 = %0000111111111000 ;boiler header, outlet, topstat, midpocket, botstat, bflow, bretn, tint, tout (ADC19-ADC27)

Am I missing something obvious? (no one is infallible after all ).

Regards
Eclectica

 

[nick12ab]

Are you sure it's wired like in the diagram? The one in the diagram wouldn't work properly anyway because the upper resistor of the potential divider is just a dead short to 5V.

 

[MartinM57]

Surely the input impedence of the ADCs in the 100s of Megs or even a gig?

It's relatively high but nothing like that - some current has to flow to charge up the ADC capacitor inside the PICAXE.

Until someone comes up with data sheet chapter and verse, I would guess (yes, I'm guessing) something in the order of a few 10's of Kohm

...but as Nick points out, your schematic also has a serious error in it at the moment...

 

[Technical]

That's a bad setup for a PICAXE, you want the total resistance from top to bottom of your potential divider to be more like 10k- 20k or so, not several 100k.
We assume the PICAXE connection is meant to be in the middle...
A PICAXE ADC is not an op-amp, in fact current has 'to be able to flow' to charge up the internal circuit to work correctly.
A 5k thermistor with a 4k7 pull down would be much better.

 

[hippy]

Thinking I have set up the ADC ok, there is always a 10mV or more drop across the 22k to the PICaxe pin with my Fluke 1587 DMM.

Surely the input impedence of the ADCs in the 100s of Megs or even a gig?

The effective resistance to 0V is probably high but perhaps not that high. From the voltage drop across the resistor and knowing its value, the current through it can be determined; about 450nA.

If you know the voltage at either end of the 22K you can calculate the effective input resistance of the ADC to 0V. Say it's 2.5V that would indicate around 5M5.

 

[Eclectica]

Thanks for the replies guys,

Yes, appologies, the schematic is wrong - sorry; trying to rush because the Mrs had to go out tonight and I absolutely had to get back to get tea ready !

The trouble with a lower resistance NTC is that the power dissipation leads to self heating - which is what I wanted to avoid. I know it would be better to pass through an instrumentation op amp but for 9 channels it gets kind of expensive! I could have gone for a physically larger NTC with a higher dissipation constant which would have helped I suppose, but the small size is useful in order to pack them into small tubes to gain access alongside part-filled thermopockets.

Ok. so at least there is a good explanation. I should be able to calculate this out in the lookup tables with luck and regain desired accuracy.

Thanks again - your responses are much appreciated.

Regards
Eclectica

 

[nick12ab]

An idea to prevent the thermistor from self-heating - power the potential divider with a PICAXE output pin only when you want to take a reading.