# Dual op-amp problem.

Discussion in 'General Electronics Chat' started by Zortz, Jul 19, 2014.

Not open for further replies.
1. ### Zortz Thread Starter New Member

May 2, 2014
16
0
Hey!
Op amp: LM358PE3

What i am doing:
I have thermocouples and i am trying to amplyfy the signal.

Setup:
Non-inverting, R1 = 1k, R2 = 250k, Vcc = 4.8dc
Thermocouple + to opamp +, thermocouple - to ground.

Problem 1(main focus now):
amp2 is amplyfing less then amp1(40 times difference)

Problem 2:
saturation at 3.7 volts while supply is 4.8

What i have tested:
1. i have checked all resistors and i have swapped them, no help.
2. i have swapped chip 4 times thinking i have damaged it somehow, no help.
3. Tested inverting solution - same problem.
4. Tested op-amp without thermocouple as voltage follower - working.
5. Tested non-inverting solution without thermocouple and gain of 10 -working.

From the tests i am thinking that op-amp is working and problem is related to high gain or thermocouple.

I am slowly losing my mind here..everything works at amp1 then i put thermocouple at amp2(100 % same circuit as amp1) and it fails.

I have no idea what to test next to find the problem..help

Thank you!

2. ### Alec_t AAC Fanatic!

Sep 17, 2013
5,977
1,136
Shouldn't R1 go to the inverting input? What does connect to that input? A schematic would be helpful.

Zortz likes this.
3. ### Zortz Thread Starter New Member

May 2, 2014
16
0

Where Vin is thermocouple+. And thermocouple- goes to ground.

LM358 pinout. I am 100% sure that i have correct connections(i have verified this multiple times).

Last edited: Jul 19, 2014
4. ### shteii01 AAC Fanatic!

Feb 19, 2010
3,507
512
Problem 2.
Your rail is 4.8 volts. "Normal" op amps never reach their rails. The fact that you only get as high as 3.7 volts is not just normal, it is very normal. If you want to get closer to the rail, then you need to use rail-to-rail op amp. Rail-to-rail op amps are "special" op amps that are designed to allow the user/designer to get as close to rail as possible, usually withing .2 to .4 volts. So if you used rail to rail op amp, you would have gotten 4.4 to 4.6 volts instead of your observed 3.7 volts.

If you look in the datasheet, you will see that output high (Voh) is Vcc-1.5. So let us examine your findings. You Vcc is 4.8 volts. You maximum output then is:
Vcc-1.5=4.8-1.5=3.3 volts. You actually got 3.7 volts. I would check the drift on your power supply. See if power supply drifted up a little, I have had this happen to me in laboratory experiments with Tektronix bench power supply. Also this illustrates that you ALWAYS be 1.5 volts below the power supply/rail. No matter what you. This is limitation built into the chip.

Last edited: Jul 19, 2014
Zortz likes this.
5. ### Zortz Thread Starter New Member

May 2, 2014
16
0
Thank you!
You solved problem 2 for me, new thing learned!

For problem 1 however i dont know. I am using them as seperate op-amps with identical circuit and one of them is always giving wrong solutions.

6. ### Zortz Thread Starter New Member

May 2, 2014
16
0
Did some more tests:
Test V = 4 mA
Gain = 250
Expected Vout = ~1V

___amp1 amp2
IC1 1.4___0.6
IC2 1.37__0.72
IC3 1.14__0.6
IC4 1.0___0.67

Would expect to get same results when i just swap the IC from circuit.
So, i think chips are broken.
Wondering if i managed to kill them all with static or they were like this when sold.
I need rail-to-rail anyways so...​

Last edited: Jul 19, 2014
7. ### shteii01 AAC Fanatic!

Feb 19, 2010
3,507
512
You could just use larger rail. The LM358 has normal maximum rail/supply voltage of 30 volts, absolute max. rail/supply of 32 volts (but you don't want to do that), so max usable is 30-1.5=28.5 volts.

Also check your resistors. 1.4/.004=350. Maybe instead of gain of 250 you setup gain of 350. And resistors have their own tolerances, you might have 1 kOhm resistor that is really 0.980 kOhm, so check them out with 3 or 4 digit ohm meter to get a better idea of what resistors you REALLY have.

Also check out the input offset voltage, datasheet says typical is 2 mV and it can go as high as 3-7 mV. So take your 4 mV, add 2 mV input offset, and run it though 250 gain: (4+2 mV)*250=1.5 volts. There is also note about V out =1.4 volts, but I don't know what it means:
"(1) VO≃1.4V, RS=0Ω with V+ from 5V to 30V; and over the full input common-mode range (0V to V+ −1.5V) at 25°C."

Zortz likes this.
8. ### Zortz Thread Starter New Member

May 2, 2014
16
0
Thank you for input.
Yes the actual gain is 251.03.
Resistors are equal at both sides, i have also swapped them around - no change at all.
How can i measure input offset?

9. ### crutschow Expert

Mar 14, 2008
13,505
3,376
Short the (+) input to ground. Measure the output voltage and divide that by the circuit gain. That's your input offset voltage.

Zortz likes this.
10. ### MrAl Distinguished Member

Jun 17, 2014
2,562
518
Hi,

As others have pointed out, input offset can be a huge problem in circuits with high gain because the input offset affects the output reading so much.
Unfortunately for this application the LM358 is not that good of a choice because of the relatively large input offset. They make much better op amps for this kind of application and i suggest you go look for one. That will improve the circuit a hundred fold.

If you insist on using the LM358, then you will have to find a way around the input offset problem. For operation at one fixed temperature you can use a resistor biasing scheme to null out the offset, but if the temperature changes much then the network will have to be able to track temperature too, and i dont believe it's worth the effort to do this.

Because you have two 'identical' circuits you have a somewhat unique situation which allows comparative testing of components. This is an advantage because at least for now you dont care about absolute values just that both circuits work the same way.
For example, if you suspect one op amp is worse than the other you can simply swap op amp packages. If the problem also moves then it is one of the op amps. If you swap two resistors (in identical positions in each circuit) and the problem moves, you know it is one of the resistors.

For this circuit though the first problem to solve is the op amp input offset spec, which is too large for your application. Fix that first and you should see a marked improvement.

Zortz likes this.
11. ### Alec_t AAC Fanatic!

Sep 17, 2013
5,977
1,136
What are you using to provide your 4mV test input?
What temperature range are you hoping to measure in?
How will you apply cold junction temperature compensation?

Last edited: Jul 20, 2014
Zortz likes this.
12. ### Zortz Thread Starter New Member

May 2, 2014
16
0
Did some tests:
gain ~250
____1___2(mV)
IC1 447 6.8(?)
IC2 330 8.5(?)

gain ~1
____1___2(mV)
IC1 6.1 2.5
IC2 3.2 2.8
One thing i noticed: when i pulled out feedback resistor then ICx.1 went saturating high but ICx.2 did not change at all.
I did dublicate my circuit for another IC, so now i have 2 ICs at same time, they behaved same way.

MrAl:
I now ordered LM833P, it has
input offset voltage 0.15 mV(typical)
Not perfect but alot better.

Alec_t:
What are you using to provide your 4mV test input?

I did use potentiometer.

What temperature range are you hoping to measure in?
How will you apply cold junction temperature compensation?

Circuit will measure the "temperature" of motorcycle exhaust.
I am using this data to sync out carburators. I dont really care about temperature, i only need to know if one exhaust is hotter or colder then the other.

So i am planning to just amplify the thermocouple into ADC and display results on LCD.

Thanks for help guys!

13. ### Alec_t AAC Fanatic!

Sep 17, 2013
5,977
1,136
Are you absolutely sure you haven't got the (+) and (-) inputs of opamp2 reversed? R1 should go to the (-) input. Those readings suggest opamp2 is trying to drive its output low and is stuck as close as it can get to the 0V rail.
Since you are only interested in the difference between the two thermocouple voltages you could use just one opamp, in differential mode.

Zortz likes this.
14. ### Zortz Thread Starter New Member

May 2, 2014
16
0
Yes sure. Tried swappin + and -. Both will saturate.

I have 4 sensors so i guess it is simpler to just ADC each.

15. ### Alec_t AAC Fanatic!

Sep 17, 2013
5,977
1,136
If you plan to have one opamp per thermocouple then you'll need a different circuit. Changes in thermocouple voltage with temperature may be small (uV per degree), but the absolute voltage will be much bigger if you're checking exhaust temperatures. Hence with a gain of 250 the present circuit will always cause the opamp to saturate.
Which type (J etc) of thermocouple do you have?

Zortz likes this.
16. ### Zortz Thread Starter New Member

May 2, 2014
16
0
K type.
At room temp it reads dead 0. Heated max i could get is 25 mV.
Exhaust will be 15 mV max or engine will overheat(i tested this and got ~12 mV when engine was very very hot-its air cooled)

Could you explain more about absolute volate? I am a bit lost here.

Apr 5, 2008
15,806
2,389
Hello,

I am closing this thread as it violates AAC policy and/or safety issues.

Quote:
6. Restricted topics.

The following topics are regularly raised however are considered off-topic at all times and will results in Your thread being closed without question:

• Any kind of over-unity devices and systems
• Automotive modifications
• Devices designed to electrocute or shock another person
• LEDs to mains
• Phone jammers
• Rail guns and high-energy projectile devices
• Transformer-less power supplies
This comes from our Tos: