Unexpected op amp behavior...

[kutalinelucas]

Hey guys...brand new to the forum and pretty excited to be involved looking at the number of viewers in so many categories...I will be chipping in my tuppence when I can, but first I'd really like to open with a question if thats OK

I build a circuit a few weeks ago and thought I cracked it with little problems, now I've come back to calibrate it and its not behaving as I'd expect it...
I have a couple of op amps which are giving unexpected results, and was hoping somebody could see what's happening.

The sensor part of the circuit consists of two op-amps each with a separate matched cold-junction compensation IC (http://www.linear.com/product/LTK001) to read temps from two K-type thermocouples, powered by a regulated split 12v (-6v to +6v) The result is fed into the ADC of a PIC18F4550.

And here's the problem...when I heat the aluminium with the thermocouples fixed inside, the voltage measured across the thermocouple terminals increases with temperature as expected, but the output, measured against ground sits at about 0.3v until the thermocouple output measures about 2.5mA at 70 degrees. Then the voltage rapidly increases to ~5v, settles and continues to rise fairly linearly with temperature. From the data sheet, it seems the common mode voltage range would be between -5.25v to 5v but analog theory really isn't my forte and I'm in a bit of a pickle!

Here's the schematic for the sensor board...

**broken link removed** (please note, the voltage regulator towards the left of the schematic is not grounded as shown)

Any ideas would be welcomed, and thanks for looking

 

[kripacharya]

Well, had a look at your schematic, and several points I could not understand :

1) what is the purpose of U7 & U8 ?
2) I do not see how you are achieving the split +/-6v. Your GND symbol is locked to 12v gnd, and goes to several places. But I do not see any virtual mid-point gnd ?
3) couldn't figure out the purpose of the LT1025's config either ?

am I missing something.... or is your virtual gnd & schematic messed up in some way ?

 

[kutalinelucas]

Hi, thanks for getting back to me. Just to address your queries...

1. I'm assuming by pure magic (as this is my first experience with op amps) U7 and U8 are uA741's which split the regulated 12v into 2 6v rail-to-rail supplies. I added two separate supplies for each op amp just in case noise became an issue. The splitter came from an example posted here https://tangentsoft.net/elec/vgrounds.html, which does in fact work when measured against pin 6 of U8 and U7 (virtual earth). The circuit works, but does not respond (measured against common earth) until the PD across the thermocouple terminals is greater than 2.5mA.

2. As I posted above, the ground shown underneath the 12v reg is a mistake and I'm not sure how it got into the schematic

3. The 1025 config was taken directly from its data sheet, as there is no other documentation on these ICs. I, possibly naively assumed I could connect the pair as described in the data sheet and measure the output from ground, common to the 1025...

I know I'm missing something, but really don't know enough about op amp behaviour yet to figure it out

- - - Updated - - -

hang about, I think you're right this is an old schematic...give me a minute I'll try and sort it out...my mistake

- - - Updated - - -

Really sorry about that...here you go http://s7.postimg.org/76dteumsb/Breakout_circuit.jpg

 

[FvM]

The 1025 config was taken directly from its data sheet, as there is no other documentation on these ICs. I, possibly naively assumed I could connect the pair as described in the data sheet and measure the output from ground, common to the 1025...

I can't relate your schematic to the datasheet configurations. I doubt that it keeps the necessary voltage ranges of all pins.

As a first point, you should cleanup the virtual ground splitter, it's at least missing a ground connection at the output.

I have problems to read the cicruit details from the lowres JPG schematic.

 

[kutalinelucas]

I've uploaded a higher res image of the sensor circuit here... **broken link removed**

Fvm, I'm not sure what you mean by you cannot relate my schematic to the datasheet, assuming you've looked at the second link I posted (and ignored the first), and I assure you, there is a stable -6v and +6v on the required pins of all ICs.

Also, could you please clarify what you mean by cleaning up the ground splitter? I'm not an electronic engineer by definition, but just trying to muggle my way through and learn as much as possible on the way

 

[FvM]

The LT1025 circuit is O.K. now, the virtual ground circuit isn't, it's still missing a ground connection of the OP output, please compare with example circuits in your links. Without the connection, it can't work.

Not directly related to the original question, but a processor based thermocouple sensor don't necessarily need analog cold junction compensation. It can also digitize the uncompensated thermocouple signal, use another temperature sensor (e.g. Pt100, silicon diode, PTC, digital) to measure the cold junction temperature and do all compensation and linerization in software. This can be done with 5V single supply amplifiers and lower component count.

 

[kutalinelucas]

I also have a thermistor connected to one of the ADC channels, I just thought for the sake of £6 it would be quicker to do it through hardware...could you possibly embellish on the problem with the "OP output"...I'm not in any way looking for a free ride and rarely get this stuck, but I'm worried you'll log off before I get it right!

 

[phongphanp]

Depend on range of distance .
Depend on type of Opamp .
... etc

 

[kutalinelucas]

Depend on range of distance .
Depend on type of Opamp .
... etc

Sorry, thats confused me even more. To get back to FvMs comment, should the gnd terminals of the 1025 be grounded to the (virtual earth) output of the 741?

 

[phongphanp]

You should consult electronics books , it's more basic problem.

 

[kutalinelucas]

Yeah thanks for that, very constructive. I'm not asking for any hand-outs...just a point in the right direction from those in the know. Unfortunately it really has stopped me in my tracks and I can't see the wood for the trees

 

[kripacharya]

Yeah thanks for that, very constructive....

LoL !!!

I think the key to your troubles is very likely the whole virtual earth setup. Lets focus on just that initially and make sure you have it setup right at all places, with no missing links/ extra shorts etc etc. If you don't understand how this is done, then google for 'virtual earth' and you'll get a bunch of leads.

In short (no pun intended), a virtual earth is just a stable voltage created which is usually midway between the 0 & +ve rails. The voltage source which does this should be able to handle currents in both directions. That's what your U7 and U8 seem to be doing. But then you hadn't connected them anyplace. And no, the other opamp's power pin is NOT connected to this earth.

Next - your LT1025 connections is beyond my understanding. Maybe 'cos I don't know the pinout functionality of this IC.

Anyhow, first step is to sort out the VE concerns, and then we'll see what to do next

 

[kutalinelucas]

This is what is pickling me...do you have any suggestions on how I can bring u7 and u8 into the circuit without affecting the split-rail supply?

 

[phongphanp]

Can u sent data sheet of all Opamp to me? At phongphanp@yahoo.com .

 

[kutalinelucas]

Of course...didn't mean to be flippant earlier, its just this problem has halted my progression for 3 days and I'm getting frustrated!

 

[schmitt trigger]

LoL !!!

I think the key to your troubles is very likely the whole virtual earth setup. That's what your U7 and U8 seem to be doing. But then you hadn't connected them anyplace.

Me too believes that this is EXACTLY the problem. The outputs of U7 and U8 ARE the virtual ground but both are unconnected.

 

[kutalinelucas]

I agree, but I don't know how to connect them. If I place say a 0.1nF cap between vgnd (pin6) and common gnd, the LTKA0 output still only amplifies when the thermocouple is at 2.5mA...or simply with a wire, still no response until 2.5mA

 

[schmitt trigger]

It is very simple.
In the Linear Technologies datasheet circuit:

1) All The V+ pins go to your +12V
2) All the V- pins go to your 12V return (and no, this is NOT ground)
so far, everything is like that in your schematic..
3) Last and most important, all the nodes that have an attached ground symbol, go to the output of the 741 opamp.

Simple.

 

[kutalinelucas]

Unfortunately not! If I connect the virtual earth to the common earth on the sensor board, the op amp voltage jumps to 5.5v both isolated and connected to my computer and does not amplify at all...I've just noticed that when I hold the thermocouple rod, the output rises linearly from until the thermocouple hits 2.5mA, it then goes crazy and rockets up to 10v!

I don't know, don't think I've ever been this stuck.

 

[schmitt trigger]

sensor board?
computer connected?


There is a system issue here...Most likely a ground loop.

Do not connect to the computer at all. Troubleshoot the board all by itself. Connect the thermocouple, a separate bench power supply, and then read the voltage values with a DMM as the couple heats and cools.
After, and only after the stand alone board is working, then connect to the computer...

If all else fails, then you will have to use an eight-brick or similar power supply.....with 12 V in, +/-6V out, isolated.