Instrumentation amplifer oscillates.

[fnpfar]

Hi folks

I have built an instrumentation amplifier with a LM324 single rail supply op amp,
soldering on a board to avoid breadboards parasitic capacitance. The schematic that
I followed is published on the datasheet "Typical Applications":



The problem is that when I conected it to the scope, instead of output a DC line when
aplying differential voltage, it outputs a pulsating wave. It is oscillating at 50Hz, but the
supply voltage is conveniently filtered of ripple so I think this oscillation may be produced
by parasitics on the board or something?


(1v/div, 5ms/div, 0v at the bottom)

I have tried to conect a capacitor as a peak detector, because they have the amplitude
that the amplifer should have on DC, but when conected it loads the amplifer and the
voltage falls to nearly 0.

Other way I figured to meausure this peaks is to conect it to a buffer amplifier and drive
from here the capacitor peak detector, but why is this circuit oscillating? It shouldnt.

Thank you for all

[Andy Watson]

How, exactly, have you connected the differential input? - is there a common mode path to ground?
Also, when you probe the output you are adding a few, possibly tens of picofarads load which could cause instability - try "isolating" the scope probe with a few hundred ohms resistance directly at the tip of the probe.

[fnpfar]

There wasnt a patch to ground from the imput, my fault... Now I added a 4.7K  conecting the (-) input to ground and the peaks have disappeared but now the gain has been reduced to almost 0, when with these 2K , its voltage gain should be 100.

(The amplifier is conected to a solar cell that produces 0 to 0.2v to meausure light intensity.)

[Andy Watson]

You mentioned "single supply" op-amp - does this mean that your ground/reference is also the most negative supply rail? If so, you might want to think-through what output voltages you would expect to see at the outputs of the op-amps.

[fnpfar]

Yes, ground is the lower rail. On the output I expect a range from its minimum offset voltage (that appears to be like 0.5v), to saturation (supply voltage is 10v so i think this amplifer should saturate at 8v or so, since it is not a rail to rail amplifer).

[Andy Watson]

Consider 50mV differential input and calculate what voltages you would expect to see at R3 and R6 - it should become obvious why you'll be heading for difficulties without a negative supply. You should find that you can select a common mode voltage the will push the amplifier into range where it will operate - try again with -50mV differential input

And, IIRC as you noticed the output of the LM124 is pathetic especially near to the negative supply rail.

[fnpfar]

Yes!.. It was that.. Im adding common mode voltaje and works perfectly.

Thanks a lot for your help Andy!.. I will continue tweaking this so I can find a good range.

Greetings

[jesuscf]

That happens when one of the op-amps in the LM324 saturates to the negative rail, in this case 0V.  One solution is to shift the reference of the op-amps a few volts up.  You can use the fourth op-amp in the LM324 IC to create that reference as shown in the first attached figure.  In this case I made it 5V.  The second figure shows the output of the simulated amplifier (NI Multisim) working as expected . 

[Andy Watson]

The LM124 is a very old design. I believe it suffers the same problem demonstrated in this video:
https://youtu.be/VgodYtiD_F0

[capt bullshot]

I remember Bob Pease writing "the LM324 is a lousy (lousy is a technical term) amplifier"

Thats true from my own experience, but this this is still usable if you know its limits. The crossover problem is one of these, simple fix would be to provide a load resistor to the negative rail in the way that the output doesn't have to cross over.  That is what Pease recommended.