Op Amp

[metrologist]

OK, now if I have the inputs shorted together or open, I get 4V out. If I ground pin 3 I get 30mV. It does not seem to matter what I do with pin 1.  If I connect my supply to pin 3 and ground, I get 30mV no matter input voltage. If I connect both inputs to supply, I get around 3V with 0V input and 4.2V with 60mV input. If I just connect a
short jumper to pin 3 and touch the insulation, the voltage goes down a couple volts, if I touch pin 3 I get ~600mV out.

[Wimberleytech]

Connect pin 1 to ground and connect pin 3 to ground.  Then measure every pin with respect to ground and fill out this table.

[metrologist]

I had to do some manual labor for a while.

1: 0V
2: 0V
3: 0V
4: 0V
5: 3mV
6: 4.7V
7: 3mV
8: 3mV

[rstofer]

I may have missed it but did you specify exactly which package you are using?  If so, my bad...  Skip the rest!

In any event, the pinout for the PDIP would be the center image and it is, by far, the more common package.

The datasheet makes a reference to another device as justification for the bizarre pinout on the S8 package.  Kind of like "we screwed up before and we thought we would do it again".

[metrologist]

I don't know which package I have. Does the the mfg make it possible to tell? They seemed to make some effort to say the presumed packaged is labeled as 1078, but does that mean any other package would have an additional marking? It's a small IC, but it is not like there couldn't be a few more letters printed on it. Perhaps the only thing left to do is try alternate pinouts. Or just do something else entirely. This sucks!

[rstofer]

The SO-8 package should have one edge beveled as in this package drawing:

http://cds.linear.com/docs/en/packaging/05081610_G_SO8.pdf

The PDIP is just an ordinary DIP with no beveled edge:

http://www.interfacebus.com/ic-package-plastic-dual-in-line-through-hole-description.html

We really need to get this squared away because you could be wiping out chips by the bunches.

If it is a PDIP, you are putting the power and ground where you shouldn't.  Here is the standard pinout:

http://www.learningaboutelectronics.com/Articles/Summing-amplifier-circuit.php

[metrologist]

Is the SO-8 package the same as a pdip, dip, or any other acronym one could formulate? Just pissed at the industry now.

No, the edge is not beveled. Of course, a beveled edge is the only illustration I recall seeing in the datasheet. So that was another reason I thought I might have that 'qpdip' config.

Seriously, what's the difference between dip and qdip? plastic is all I was able to find. Good that does me because aren't they both plastic? I googled this crap at first too. Got no answers for my time.

[metrologist]

BTW, these are board pulls and I measured several chips on a board and found all pin 2 had continuity, and all pin 6 had continuity. How am I getting blunted?

[Wimberleytech]

I had to do some manual labor for a while.

1: 0V
2: 0V
3: 0V
4: 0V
5: 3mV
6: 4.7V
7: 3mV
8: 3mV

These numbers look good enough.  Now, can you drive pin 3 to 20mV or something close?  Then repeat the table.

[rstofer]

Is the SO-8 package the same as a pdip, dip, or any other acronym one could formulate? Just pissed at the industry now.

No, the edge is not beveled. Of course, a beveled edge is the only illustration I recall seeing in the datasheet. So that was another reason I thought I might have that 'qpdip' config.

Seriously, what's the difference between dip and qdip? plastic is all I was able to find. Good that does me because aren't they both plastic? I googled this crap at first too. Got no answers for my time.

The only similarity is that they all have 8 pins.  That beveled edge is not JUST an illustration, on the SO8 when you look end-on, you will plainly see the bevel.  There is no mistaking it!

Look at the datasheet and notice that the two configurations have different letters after the LT1078 bit.  Just knowing it is an LT1078 is meaningless if you don't know what package it is.  The letters tell you.  OTOH, so does the very pronounced bevel running the entire length of the left side of the package (pins 1-4).

It's all there on page 2 of the datasheet.  You can see from the shaded edge of the package outline which devices have the oddball pinout.

http://cds.linear.com/docs/en/datasheet/10789fe.pdf

There's nothing in the datasheet that discusses qdip.  I'm not sure I know what that is.

At Digikey they only show 2 packages:  PDIP (a regular DIP package for through hole mounting) and SOIC for surface mount.  That kind of simplifies things.  Is your device intended for through hole mountint?  If so, use the PDIP pinout.  If it is intended for surface mount, use the SO8 pinout.

https://www.digikey.com/products/en/integrated-circuits-ics/linear-amplifiers-instrumentation-op-amps-buffer-amps/687?k=lt1078

[metrologist]

I had to do some manual labor for a while.

1: 0V
2: 0V
3: 0V
4: 0V
5: 3mV
6: 4.7V
7: 3mV
8: 3mV

These numbers look good enough.  Now, can you drive pin 3 to 20mV or something close?  Then repeat the table.

Pins 1/3 were grounded, so should I remove ground from both pins for this? Previously, "If I connect my supply to pin 3 and ground, I get 30mV no matter input voltage." That was on the output pin, so not sure about the others, and pin1 was floating.

I left pin1 grounded:

1: 0V
2: 0V
3: 10mV
4: 20mV
5: 600mV
6: 4.7V
7: 12mV
8: 12mV

The output seems to track the input in this configuration, ~70mV in hits the output rail. I am guessing this is using just the B section of the chip.

There must be a reason for the supplied reference circuit, but it says gain=100 and it also has notes that input is 0.03 to 1.8V, and output is 0.03 to 2.3V. That doesn't make sense 

Is the SO-8 package the same as a pdip, dip, or any other acronym one could formulate? Just pissed at the industry now.

No, the edge is not beveled. Of course, a beveled edge is the only illustration I recall seeing in the datasheet. So that was another reason I thought I might have that 'qpdip' config.

Seriously, what's the difference between dip and qdip? plastic is all I was able to find. Good that does me because aren't they both plastic? I googled this crap at first too. Got no answers for my time.

The only similarity is that they all have 8 pins.  That beveled edge is not JUST an illustration, on the SO8 when you look end-on, you will plainly see the bevel.  There is no mistaking it!

Look at the datasheet and notice that the two configurations have different letters after the LT1078 bit.  Just knowing it is an LT1078 is meaningless if you don't know what package it is.  The letters tell you.  OTOH, so does the very pronounced bevel running the entire length of the left side of the package (pins 1-4).

It's all there on page 2 of the datasheet.  You can see from the shaded edge of the package outline which devices have the oddball pinout.

http://cds.linear.com/docs/en/datasheet/10789fe.pdf

There's nothing in the datasheet that discusses qdip.  I'm not sure I know what that is.

At Digikey they only show 2 packages:  PDIP (a regular DIP package for through hole mounting) and SOIC for surface mount.  That kind of simplifies things.  Is your device intended for through hole mountint?  If so, use the PDIP pinout.  If it is intended for surface mount, use the SO8 pinout.

https://www.digikey.com/products/en/integrated-circuits-ics/linear-amplifiers-instrumentation-op-amps-buffer-amps/687?k=lt1078

The chip is a small surface mount. There is a small bevel, but I've seen wider bevels before.

There are no other marking on the chip. It has a stylized LT in the corner, then I think a date code of 406, then below that is 1078. No other characters.

The datasheet says the PART MARKING is 1078 for the "odd" pinout, so everything I have tells me it's the odd pinout.

[metrologist]

My last try at this. After looking at the lt1078 datasheet a bit more and the notes below the reference schematic:

INPUT RANGE = 0.03V TO 1.8V
OUTPUT RANGE = 0.03V TO 2.3V
 (0.3mV ? VIN+ – VIN– ? 23mV)

I don't get the input range because at 30mV, the 100 gain amp would be railed. The note in parenthesis makes sense for the output range.

I set up a 500k bourns pot with a battery at each end, center tap set to 1mV. I connect battery ground to the inverting input (circuit -) and center tap to non-inverting input (circuit +). I see ~2V output. I increase input voltage until I hit the output rail at ~43mV input.

I repeat the above with a jumper from the inverting input (-) to circuit ground. 1mV input yields about 2mV out. Increasing input voltage to ~75mV hits the output rail.

I suppose that configuration is creating a virtual ground at the A- input, and then there is the ratio of 1M/20k, which is about the gain I am measuring. Maybe the A channel is canceling some noise and offset, if I want to pretend it's doing something useful.

I think I am just going to use it this way by wiring my shunt between ground and the non-inverting input.

[Wimberleytech]

I don't get the input range because at 30mV, the 100 gain amp would be railed. The note in parenthesis makes sense for the output range.

They are probably referring to "common mode" input range.

I set up a 500k bourns pot with a battery at each end, center tap set to 1mV. I connect battery ground to the inverting input (circuit -) and center tap to non-inverting input (circuit +). I see ~2V output. I increase input voltage until I hit the output rail at ~43mV input.

I repeat the above with a jumper from the inverting input (-) to circuit ground. 1mV input yields about 2mV out. Increasing input voltage to ~75mV hits the output rail.

I suppose that configuration is creating a virtual ground at the A- input, and then there is the ratio of 1M/20k, which is about the gain I am measuring. Maybe the A channel is canceling some noise and offset, if I want to pretend it's doing something useful.

I think I am just going to use it this way by wiring my shunt between ground and the non-inverting input.

That is close to what I suggested some time back.  I would expect more accuracy than you are seeing but if it works for you, that is fine.
I have wondered if you are getting some artifacts that you do not see with a DMM but would see with a scope.

[metrologist]

I read about common mode something as (V1+V2)/2 and differential mode as (V1-V2)/2.

How would each of these modes be wired? I suspect differential mode could be modeled as a battery between the + and - inputs to the op amp, and common mode wired as that, and by tying the - input to circuit ground.

Since voltage is not an absolute, it the inputs are independent of the circuit, then what are V1 and V2 measured relative to?

Where would you expect to see more accuracy? I'm still not sure if the circuit is working or only half working.

[metrologist]

I got this done...not what I see on my actual circuit, but I have no idea what I'm doing there. But the plot is what I was expecting.

[Wimberleytech]

Here, use this.  Take a look at the offset voltage in the sine wave and notice that I have a 0.04 volt bias at the non-inverting input.  This circuit indeed requires some offset away from ground.

[Wimberleytech]

Sadly, it appears that the 1078 is not happy with input at ground despite what the datasheet says on the facepage textual description.  If I am to believe the simulations anyway.  I dont have any I can test on the bench to confirm.

<update>
It may be more of an issue that the output cannot really go to ground.  Taking a look at the output, I see issues with a true ground output.  There is a note in the datasheet about.  Truly, no single-supply opamp can output ground.

[danadak]

RRIO opamps can swing to rails, but with no load.

Once you place some load on an output, the outputs being CMOS
P and N channel devices, all have non zero Rdson, so with load
they no longer swing to rail.


Regards, Dana.

[metrologist]

Thanks for all the information and support. I am especially pleased to have finally downloaded LTspice and to have spent some time playing with the circuit elements here. That exercise allowed me to see similar behavior that my test circuit was presenting, and now I have experienced a tool that is far more useful than anything I've used before.

I have simulated some configurations with the LT1078 that I think will work for what I want to do. 

[Wimberleytech]

Very happy about your progress!!  LTSpice is a very powerful tool.  I have used many $expensive$ simulators throughout my career and LTSpice performs very well by comparison.