LM324 opamp output variations

[earckens]

A pcb made according to the schematic (attached below) uses LM324 opamps. With no input signal to the current transformer pin 3 should be low, and pin 2 follows feedback from output pin 1 (high).
Two of the opamps give 0.0V on their pin 1, but two opamps measure approx. 2.5V on their outputs.

All components are correctly soldered with correct values on the pcb, hence I would expect five identical results.

Pcb traces to pins 2 and 3 are just millimeters long, and as short as possible.

Jumpers SJx are not soldered.

I swapped IC's around, but still the same circuits of the five do show this behaviour, hence unrelated to the IC's.

Why are two outputs not near 0V?

 

[Tony Stewart]

Voltage on all input pins may tell you why the output is low or high. 1st stage has a gain of +101..
Maybe some diodes are reversed D1 etc Does layout match schema?

2nd stage is a compare with vbl. threshold going to a filter for a 2.5V compare in stage 3.

 

[earckens]

Voltage on all input pins will tell you why the output is low or high.
Maybe some diodes are reversed D1 etc

No, diodes all are of correct type and orientation.

The opamp stage in question has a x1000 amplification, so the output has to be either low (0V) or high (Vcc - 1 or 2V).

So for an output being 'low' then 2V is pretty high, isn't it?

 

[Tony Stewart]

1st stage problem If Vio = 2 mV , Av=1k then 2V out LM324B spec is +/-3mV @ 25'C
Is there Vio on that line? I assume CT's are connected (?) . Otherwise . install jumper

 

[rjenkinsgb]

Is the gain 100 or 1000?
As shown, it should be 100 x, plus the input value?

The 324 has a possible worst-case input offset of 3 - 4 mV, so if the gain is 1000 that could throw off the output.

A 1M or higher from pin 2 to VCC would give a slight cancelling offset.

 

[earckens]

Is the gain 100 or 1000?
As shown, it should be 100 x, plus the input value?

The 324 has a possible worst-case input offset of 3 - 4 mV, so if the gain is 1000 that could throw off the output.

A 1M or higher from pin 2 to VCC would give a slight cancelling offset.

My apologies, you are correct: x100.

If I understand you correctly the input offset (3 to 4mV) should be a reason for the output offset.. Yet, with x100 that would be 0.5V at worst?

And >=1M from pin 2 to Vcc: good point. But that will then also have implications for input voltage on pin 3, which would need to be higher than now for the output to go high.. right?

 

[Tony Stewart]

Pullup +offset on on p3 requires more input so Pot threshold must be reduced.


If IC1A-p1 = 2 V then p2, p3 = (1/101)*2= 20 mV.. +/-Vio
So why is p3 = 20 mV with no current.

 

[earckens]

So why is p3 = 20 mV with no current.

Leakage from D1 (Schottky, cathode to Vcc, anode to junction R3/R4/D4/C2)?

 

[earckens]

I replaced D1 (Schottky) with 1N4148 and voltage on pin 3 dropped from 16mV to 0.0004V (Fluke 289Pin 1 dropped from approx. 2V to 0.0002V

Case solved, thank you Tony Stewart

 

[Tony Stewart]

What is max 2ndary current ? * 1k must not < -0.3V from datasheet yet input Veb can handle 50 mA

I would move D1 to gnd for + clamp. and add jumper burden while ,R3 shud be 10k ~50k then reduce C

But if your secondary current is low, maybe you are ok now

 

[earckens]

But if your secondary current is low, maybe you are ok now

That is the point: the input from the current transformer must trigger this circuit at low values. Reason being that this results in higher detection sensitivity.

 

[Tony Stewart]

As long as you verify Do Not Exceed levels. PNP's are pretty tough. So you are good to go

 

[earckens]

Tony Stewart Thank you Tony, your help has been invaluable.

 

[MrAl]

Hi,

Just to add a couple notes here.

First, a non inverting amplifier stage is 1+Rf/Rin so with 1M and 10k we have 1+100 which is a gain of 101.
With 1mv input offset that means 0.101 volts output. With 2mv input offset that means 0.202 volts output. I've never seen one of these op amps with an input offset greater than 2mv at 25 degrees C ambient, and that is probably because at other temperatures it could go even higher. If the max was 5mv though it would still only be 0.505 volts output, and all these are plus and minus.

To get +2v on the output we'd have to see about 20mv input offset, which would mean a bad op amp chip. If there are other components on the input however, we could see effects from those parts also. Schottky diodes are notorious for higher leakage and can look like a 20k resistor with a reverse voltage. Compare that to the 500M reverse resistance of a 1N4148 diode.
So how much leakage does it take to force 20mv extra on the input of the input impedance is 10k?
It takes a whopping 2ua to get an extra 20mv across a 10k resistor! Ha!
So what input voltage could cause that with a 20k resistor? Well, about 40mv.
So with just 40mv and some extra leakage we could see an extra 20mv input.
This analysis is approximate to just get some idea what it takes to mess up a seemingly good op amp circuit.

I've seen worse. I've actually seen an inexplicable near zero output offset when it could have been more like 10mv.
I've never seen a higher than expected output offset without a good reason for it however.