Help to increase the accuracy of op-amp[lm324] as possible as

[booboo1]

Hi guys
I'm working with LM324 to learn how to use Op-Amp and all aspects of it. currently my circuit is this(a simple non-inverting Amp):

**broken link removed**

And here is a table of inputs and outputs and the gain of each input:

**broken link removed**

The gain of each input is odd to me. I'm trying to increase the accuracy of this circuit as possible as. I have applied 9.6mv in the input and got 4.75v in the output. or for 11.2mv I got 5.39v. What's the problem? IIRC the max input offset voltage is 7mv. Am I right? please show me by mathematical calculation that why the output is this? the gain isn't correct. should I bias the input offset current or offset voltage? How?

 

[JimB]

How accurate is your input mV measurement?

Was it necessary to start a new thread for this, this is basically the same topic as your other thread.

JimB

 

[Ian Rogers]

I thought the gain of a 324 was only in the order of 110... I have just checked the datasheet...

 

[ronsimpson]

The max gain is 100,000. Your gain is about 400. OK
Input offset is typically 3mV but could be as bad as 7mV.
SO; 3mV X 400 = 1.2V on the output.

I think you have about 2mV of offset. When you measure 0 volts the amp sees 2mV, because of the offset. Short out the +input and see what the output voltage is. Take that number and divide by 400 to see what the input offset.

Also the inputs have 20nA to 250nA of input current. This will also cause some offset across the 2.14k ohm resistor.

 

[MikeMl]

Here is your problem: First, read the data sheet for the LM324. Two things that are important for your circuit. First is the input offset voltage, worst case +-7mV referred to the inputs. The second is Voh, the highest output voltage that this opamp can pull up to (Vcc-1.5V)

I model it as an ideal opamp, but account for these two limitations of the LM324. V3 creates the +- 7mV offset. I run the simulation three times, with the offset at +7mV, 0mV, and -7mV. Note the three V(out) vs V1 plots.

I show the Voh limit on the plots. Note that the LM324 also has a Vol limit of ~20mV, so will not quite pull to ground, either...

You have the gain set way too high. Your gain is 1+ 837/2.14 = 392. With such a poor opamp (with respect to its input offset spec), you would do much better with two stages, each of which has a gain of about 30. You could also trim the offset out...

 

[crutschow]

.................
With such a poor opamp (with respect to its input offset spec), you would do much better with two stages, each of which has a gain of about 30. You could also trim the offset out...

But using two op amps won't affect the offset at the output as compared to one op amp.

 

[Tony Stewart]

The proper way is to select an OA with the Vio that satisfies your output offset needs for desired gain.

Since your gain is high now the bandwidth will be smaller than expected. Usually 1~2MHz GBW for this OA.

Note below I have used 1KHz 1mV to 1V with gain =1000 and the slightly less amount (+/- 988mV) is due to the gain BW product of 1000x 1Khz approaches the GBW product of the Op Amp

The Quiescent current must be small since we only use 10K to bias the midpoint with this split supply.

Java Simulator
photo below shows dummy OA supply lines for clarification but not in SIM.
I chose a Bipolar arrangement from one floating supply. ( poor mans split supply)

 

[Mosaic]

The OP didn't mention if he was using DC or a frequency based input signal.
Nor did he mention if the supply was a battery, SMPS or linear reg, as bad ripple on the rails could change the O/p. Were the measurements taken with a 'scope or a DMM? A DMM would average a non DC signal and any bad ripple.

I have used the 324 at lower gains with a regulated supply and decoupling caps and the results closely conform with the simulations.

 

[Tony Stewart]

The LM324 s a very low cost Quad Op Amp, with BJT outputs that suffer from 2-3 V saturation near Vcc but low saturation from Vee rail.

For full swing output a CMOS output which has much lower current capability so R's must be scaled accordingly well above 10k .

i would stick with R>> 1k on the LM324 and compute all errors sources .

 

[audioguru]

With its gain set to 400 then the output of an LM324 will be full of noise. Since it is "low power" then it produces pretty bad crossover distortion and has trouble with frequencies above about 2kHz.

 

[Tony Stewart]

Although I have used LM324 for linear control of resonant boost HV sweep regulators for laser printers with 1% accuracy before using accurate feedback, it does have crossover distortion which can be reduced with DC biased loads.

 

[simonbramble]

Back to the point of the original OP, he needs to learn about the fundamentals of op amps. As the contributors above have said, the main reason for your error is input offset voltage. This is a *DC error* that is effectively added (or subtracted - and it can be either) from your input voltage. So with a 1mV input and a 5mV input offset voltage, your amplifier could effectively be amplifying a 6mV or -4mV signal. Input bias currents, with the resistor values you are using should not be an issue (it amounts to the input bias current (100nA) x the parallel combination of the 837k and the 2.14k.) This only amounts to a 100uV or so.

Also, how are you generating the 1mV? If it is from a high value resistive divider network, are you measuring the 1mV with a fairly low input impedance voltmeter? this could also be contributing to the errors you are seeing. If you apply a low impedance voltmeter to the junction of 2 resistors, the junction voltage will change as the meter loads the resistor divider.

Generally LM324s are not good for amplifying mV dc voltages because of the input offset voltage.

Dont worry - this is teaching you much more about op amps than you can read in a book and professional engineers still get this stuff wrong!

If you want to amplify mV, I would use an op amp with a lower input offset voltage.

if you want to learn about the LM324 (and it is one of the world's most popular op amps), use a higher input voltage (1V or so).

 

[simonbramble]

More thoughts: As Tony says, you have a very high gain, so this will limit the bandwidth. the Gain x Bandwidth product is constant. It is worth applying an ac signal to the input and seeing how the output changes with gain (if you can get your hands on an oscilloscope). You should see that as gain goes up, available bandwidth goes down.

You might also want to investigate slew rate - this is the capability of the output to change its voltage quickly. The higher the output voltage (for a given frequency) the higher the slew rate of the amp needs to be to move from one voltage to another. this spec is independent of gain bandwidth (kindof), so you can be well within the limits of the gain bandwidth and still be bitten by the slew rate if you are asking the amp's output to move too quickly

All good experiments and most of them can be done with either LTspice or basic home lab equipment

 

[Mosaic]

As I am working on a circuit that has a 324 buffering voltage pulses for a differential amp, I captured some examples of its weaknesses. My area of interest is not in the actual pulses but stable samples about 8mS away from the pulse, so I can use the 324, otherwise the crossover distortion and slew rate would be problematic as the images show. The design of the diff. amp circuit requires an opamp output in the lower half of the voltage rail, under the crossover point.



The blue trace is the voltage input. Note how the yellow follower signal lags due to slew rate and entirely misses the trailing fast pulse due to its rapid rise time.

In the second image, a 1K load to ground or V+ rail on the follower's output ensures class A operation and sidesteps the crossover distortion at the expense of a few mA draw. That improves the total slew response permitting the follower to reach it's target voltage faster.