Why is my op-amp frequency response not ideal?

[djstar]

I am having a play about with the differential amplifier using a LM324 and have shorted both inputs and set them to 3V sine wave at 100Hz.
i understand that in a ideal op-amp the output should be 0, how ever i am unsure of what is happening at the output...

i increase the frequency every 1Khz and record the voltage on the output which is in the mV range. when i plot the results i can see that the gain slowly increases then all of a sudden the gain increase rapidly then as you further increase the frequency it rapidly decreases. The same type of repression you get with a band pass filter.

my only explanation i can see from reading books is that it is to do with the compensation capacitor in the device. could some one shed some light on the subject?

 

[analogdesign]

If you shorted both inputs you are operating the LM324 as a comparator and are, in effect, measuring how the offset changes with respect to frequency. I suggest you configure the device as a unity gain follower. Connect the 3V sine wave to the + input of the op amp and feed the output back to the - input.

Are you ac or dc coupling the inputs? If you are ac coupled you will get a bandpass response.

 

[djstar]

looking back at the results i realized that the results were for a 3 op amp instrumentation amplifier that i was messing with rather than the single diff opamp. the inputs of the IA are shorted with a wire and the function generator is connected directly to the inputs

you mention about a bandpass response, this is what it looks like in the graph. what is causing this response?

 

[analogdesign]

Oh, so it isn't an LM324. Is the input to the instrumentation amplifier dc coupled or ac coupled? If there is a series capacitor then there is a zero at C times the input resistance of the IA. Typically an IA works down to dc, so either your setup is ac coupled for some reason (check the output of the function generator) or your data is wrong.

 

[Mike_In_Plano]

With both inputs shorted together, it's surprising that your getting any output as the input offset voltage would likely have the output at the high or low rail. If you did have a part with Vos=0, you'd be getting a lot of noise.
I'd suggest a useful experiment for the hobbyist. Simply make a gain circuit, i.e. 101, and check the output gain as you increase the input frequency. Try this again with a gain of 11.

What did you learn. Is it useful?

Mike In Plano

 

[analogdesign]

With both inputs shorted together, it's surprising that your getting any output as the input offset voltage would likely have the output at the high or low rail. If you did have a part with Vos=0, you'd be getting a lot of noise.

Mike In Plano

Mike, the OP didn't actually have the op amp connected that way. He or she was using an instrumentation amp, not a raw op amp. That's why the device wasn't railed.

 

[djstar]

Thanks for all the message guys, just to clear things up I have attached a picture of the 3 opamps I have connected together, what I want to know is why the output is behaving like it is in the graph.

Claire

 

[NascentOxygen]

Your graph is a little short on detail.

The amplitude peak---is that something of the order 5dB or more like 50dB? What happens at 80kHz?---the graph takes a dive, or you didn't explore beyond 80kHz? What are your values of R2 and R3?

You monitored the output waveshape and can vouch for it being sinusoidal throughout? What input amplitude were you using?

 

[djstar]

I do apologise for the lack of detail, some one needed to use the test equipment urgently so that's all the data I could pick up.
The output from the sig gen is a sine wave with a amplitude of 1 PK-Pk the peak is at about 15KHz at about - 8dB ( input 1V output 350mV)
r3 = 27K and r2 = 1K

 

[NascentOxygen]

If these are OP-AMPS, it seems that the pair of unity gain buffers achieve nothing much, leaving a single stage with ideal gain = 0. This leaves your circuit's response to be wholly determined by non-idealities, and passive element inaccuracies.

With the output being the difference between two similar terms, it's inaccuracies and imbalance which determines any non-zero result. Construction techniques need to observe balance so there are no unintended capacitances introduced to upset the balance.

The resistor arrangement around the first pair is not a potential divider, the ends connect to nominally-equal points, so it's just unity feedback.