External pull up/down/GND resistor's on op-amp output's ?

[MathWizard]

I have to learn the calculations for input offset voltages and bias currents very soon, after messing around with this circuit in LTSpice, I can have very different results by changing resistor size or op-amps. But it could be the Brodynov LM324/58 models I'm using too, some of his models give me problems.

Now I have a few caps left out of the schematic, I forgot about them, like on the zener reference, and a 1uF across the feedback 100k resistor, as 60Hz gets in there too from my hand or the scope probes. But just as a continuity tester, it seems very fast and sharp. The way it's set up now, 2 ohms trigger's it, but 2.2 won't.

I'm not sure yet how good common op-amp's really are, so I went with the higher gain to compare larger voltages. Then I'm also measuring them with my scopes, not a good DMM, so IDK really what they are doing yet anyways, or the affect probing has.


Just attaching a 1x probe on the input area, also seemed to add a lot of I think 60 or 120Hz hum. And just touching my finger to the outside package of the 100k input resistor, also makes the DC output change a lot, and couple in 60Hz. So I did put 1uF across the 100k feedback resistor. I never bother to try any other op-amp or use any comparator tho, I just wanted to make do with these LM324's I have.

For now I'll try the oscillator, then try it powered from the DMM using it's current source. But yeah I don't want the other section to turn into an oscillator, that 1uF cap should help but maybe it's too large, but the LED seems to work fine for me.

I have a great looking book here called "Operational amplifiers and linear integrated circuits" by Coughlin/Driscoll, I've been doing physical work all month, I need to take a texkbook month again.

[Zero999]

The input offset voltage is the problem here. It appears as a voltage source in series with on of the inputs. It's easy to model.

Here's an example.


I often don't bother using special models for generic op-amps such as the LM324. It's not worth the bother and seldom gives more accurate results, than LTSpice's built-in models, which also have parameters that can be changed to mimic common op-amps well enough. In the above example the open loop gain is set to 100k and GBW 1Meg, which is similar to the LM324, although the GBW is irrelevant to the simulation shown which is purely DC.

[MathWizard]

Yeah I measured the output voltage when both inputs are tied together to GND, on these ebay LM324's, and with +/-15V, the worst of the quad was about 1.1mV, the other 3 were about -500uV. With +15 and GND rails, they went to 800uV, and the other 3 went down to -800uV. So that's not too bad, these are marked TI, and their datasheet says 2mV is typical, and upto 5mV max, as the input offset, not quite the same thing I know, nevermind. Maybe I tied the NonInv to GND, made a buffer and that's the output offset voltage I measured. I made that above circuit from Zero99 too. Yeah tying the inputs together and to GND, the offset and gain are enough to drive the output to a rail.

I've gone back to BJT differatial pair circuits, and just played around with changing little things like resistors or beta's, etc.

Well another thing I forget about split voltage rails like +/-15V, and op-amp circuits, is that 2mV above GND could be called a small voltage, but it's also 15.002V above the negative rail. So yeah a good difference amplifier should only care about the difference, not the actual voltages.

[Zero999]

The easiest way to measure the input offset voltage of an op-amp is to build a DC coupled amplifier circuit with it and short the amplifier's input to 0V. The output voltage will simply be the input offset voltage, multiplied by the gain. To minimise the effect of the input bias currents, which pass through the feedback resistors, make the resistors as low as possible. Ideally the resistance between 0V and each input should be matched, although this is less important with the LM324 as the bias currents are tiny and the main error will be the offset voltage.  A non-inverting gain of 1000 for measuring the input offset and 100R and 100k are reasonable values to use. Another 100R resistor could go in series with the + input, but I doubt it'll make much difference.