Stupid op-amp question

[audioguru2]

Hi GPG,
It is all in the datasheet of the LM34 temp sensor that was linked earlier:
http://www.national.com/ds/LM/LM34.pdf
but they sure don't explain it very well.

Their block diagram shows an emitter-follower output (I hate emitter followers because of their unequal pull-up and pull-down ability) and 2 resistors to ground.

The emitter-follower can source current very well (for an analog meter to ground), but only the resistors sink current. Those resistors have high value to keep the quiescent current down, so any external current into them (like the 25uA from our diff amp) raises the voltage across them. That voltage is the "temperature", so the accuracy is ruined.

From the specs and curves in the datasheet, the current in those resistors is only 15uA at 75 degrees F. But the 25uA from the input of our diff amp will pull their voltage (which is the output voltage) up.
So our diff amp's output adjusts itself to be the same as the reference voltage, which is what Darrin measured.

Therefore our diff amp's input needs a high impedance buffer.
Even the tiny input current of an LM324 follower will cause an error of up to nearly 2%.

View attachment 36003

 

[darrins]

audioguru,

I tried a couple of different things. Removing the resistors from the non-inverting input, as you suggested, appears to work fine.

You've mentioned in a couple of your posts that the op-amp will affect the accuracy of my temperature sensor. I'm no theory expert (I'm a mechanical engineer), so could you explain, in very simple terms, how the diff op-amp will degrade the accuracy of my temp sensor?

Thanks

Darrin

 

[audioguru2]

Hi Darrin,
Didn't I explain it well?
Your LM34 sensor needs a high impedance load, like an opamp voltage follower, which can feed the 10K resistor of your diff amp.

How much temperature accuracy do you need? The LM34 has an error, your diff amp resistors have error, the offset voltage of a voltage follower and of the opamps in your diff amp cause errors and even the tiny input current of a voltage follower causes an error. Those errors can simply be cancelled by adjusting your reference voltage, which will cause an error anyway and will also be cancelled with adjustment.

You probably don't need the good common mode rejection of a full hi-z diff amp (like in the datasheet) because your sensor is far away but the reference is local. Therefore your sensor will need a good shielded cable anyway. You should also add the bypass cap and RC-damper to the sensor like figure 4 (page 7) in the datasheet.
A full diff amp (made from the same IC so that all opamps have the same offset voltage error) has twice the offset voltage error of a buffered diff amp which takes time to stabilize after power-up.

So just add the voltage follower and let us know how well it performs!

 

[darrins]

Thanks audioguru.


I will use one of my spare amps as a voltage follower between my sensor and my diff amp. I'll post my results when I get a chance.

One more question. I've noticed that when the differential input goes negative, i.e. when my sensor voltage is greater than my reference voltage, I'm still measuring some small voltage at the output of my diff amp. I'd like the output to go to zero when this occurs. Is there a way to do this? ???

Thanks again.


Darrin

 

[audioguru2]

Hi Darrin,
The minimum output voltage of an LM324 opamp will be pulled down to about 10mVDC or less with a 2K load to ground. It is spec'd to be able to drive a 2K load nearly as high as it can drive a 10K load. With a 13V supply the IC and resistor will stay cool and you can use a 1.2K resistor so the minimum output voltage will be even lower, but with a 30V supply the IC will be warm at 15V output and the 2K resistor will dissipate 392mW (quite hot for a 1/2W resistor) at 28V output.