This question is so stupid it hurts.


I am using an INA-122 instrumentation amplifier. I am supplying power from an Arduino, either 3.3V or 5.0V at my choosing. Without using any resistor for gain, and without applying any voltage to the +/- terminals I am consistently getting outputs of ~3.4V and ~5.1V, respectively. This happens even when i dont connect anything to the positive and negative input terminals. These are within range of INA's specifications. When I connect my load (0.1mV differences between the positive & negative terminal, as well as a right-leg-driver going to reference), I still am getting maxed-out outputs.

What am I missing here? I need to admit that I am a huge amateur.

I am following this DIY EEG guide


This is the general schematic, although the instrumentation and op amps I'm using differ slightly enter image description here

I have tried two of them, same result. I think the Arduino,GND circuit is single supply

  • \$\begingroup\$ Even though I have god-like qualities, my all-seeing-eye can't quite make out what you are doing. Post a schematic of the circuit you are attempting to build (use the schematic button in the post form, don't use fritzing), and if what you've built doesn't match what you are attempting to build, a photograph of what you've built. \$\endgroup\$
    – Neil_UK
    Aug 30, 2016 at 6:20
  • \$\begingroup\$ Please draw a schematic. Edit your question and click on the button that looks like electronic components. This will load the schematic editor. Otherwise it is pretty hard to understand what circuit you are actually working with. \$\endgroup\$
    – user57037
    Aug 30, 2016 at 6:20
  • 3
    \$\begingroup\$ That reads as if the inputs are floating. \$\endgroup\$ Aug 30, 2016 at 6:22
  • \$\begingroup\$ Schematic coming. I am following this instructable instructables.com/id/DIY-EEG-and-ECG-Circuit/step2/… \$\endgroup\$ Aug 30, 2016 at 6:30
  • 1
    \$\begingroup\$ My crystal ball... says ...with no input or feedback resistance, input offset and open loop gain will always saturate to one rail or the other. \$\endgroup\$ Aug 30, 2016 at 7:15

3 Answers 3


Yes, your output is saturating with nothing connected to the INA-122's inputs. Why wouldn't it? Your inputs are floating. They're floating with nothing connected, their floating with something connected, and they're floating with the right leg driver connected.

What ground is are your inputs referred to? Your load isn't giving them a return path, so where do you think the input bias current is going to go? It can't return to ground, as you've given it no path to do so, so it's going to charge up the input capacitance instead until the input (and thus output) saturates Regardless of the load on the inputs. This is normal and expected.

Assuming you are interested in the inputs, you want to include them in the rest of your circuit (instead of leaving it isolated and unconnected, as they effectively are now). Connect a 47KΩ resistor from each of your inputs to whatever it is you're calling ground for the rest of the circuit. Also, this information, while applicable to anything, is very specifically covered in the INA-122's data sheet. Page 7, Input Bias Current.

Which begs the question, why are you using something if you haven't read the datasheet? If you don't read the datasheet, then you don't know how to use it, and you'll use it incorrectly. If you want things to work, you have to read the datasheet. Not skim the parts you feel are relevant to what you're doing. Actually Read it. As soon as I figured that out, suddenly way more projects began working and there was a A LOT less headscratching. I know it seems like a pain, and you can get away with just skimming when it is a microcontroller or something digital (sometimes, other times you find out you missed that some peripheral works except in the way you need), but it is upfront pain that avoids significantly more amounts of work and frustration later. Additionally, it's extremely important to read the datasheets of anything analog. Even if you know very well how to use an op amp, the datasheet isn't going to tell you how to use an op amp. It's going to tell you if there is anything you can't or shouldn't do, or if there is something special unique to the specific op amp in question. So you always read through the whole thing.

Not that this weird. This is normal and would happen to any other in-amp. But there are also ones with extra weirdness that you would only ever figure out from reading the datasheet and carefully. There are of course reasonable exceptions to reading the datasheet (has anyone ever read the entire datasheet of an FPGA? Those things are like 1500 pages long!) but err on the side of 'I'll read all of it' whenever possible.

Remember, datasheets are like half pictures :D.

  • \$\begingroup\$ This answer single-handedly trained my brain into always reading datasheets completely any time i work with a new component \$\endgroup\$ May 1, 2019 at 18:00

The circuit shown has supply voltages of +/-9V.

You are supplying it with 5V and are surprised it doesn't work?

The amplifiers will saturate unless you maintain all internal and external nodes within the proper ranges. Converting a bipolar supply circuit to single supply is not a simple matter of replacing the amplifiers with 'single supply' versions, 99 times out of 100.

  • \$\begingroup\$ This answer is nonsense. There is no difference between a single supply and dual supply circuit except what we decide is ground. 9V and -9V is the same as 18V and 0V, but 0V in the first will be 9V in the second, but its just names. Voltage is relative potential. Ground is whatever we want it to be. And it is certainly not the problem with his circuit. Nor is that out of range for that amp. He also didn't replace anything dual supply with a single supply version. All he did was forget to give the input bias current a DC return path. \$\endgroup\$
    – metacollin
    Aug 30, 2016 at 13:52
  • \$\begingroup\$ To elaborate, 'single supply' is effectively what the input and output is referred to. And generally means the amp is ground referred and has a wide voltage swing. But you can use either in either situation, but expect clipping if using a dual supply amp. There is no real difference except in terminology and assumed part features. \$\endgroup\$
    – metacollin
    Aug 30, 2016 at 13:57
  • \$\begingroup\$ @metacollin The circuit is AC-coupled so the trivial issue of the input bias current you fixate upon (he has tried it more than one way, not only open) is not the main problem. He did substitute a single-supply inamp for a dual one if you read the question. And I referred to a bipolar supply circuit. The circuit shown cannot be operated from +5/0 no matter what op-amps he uses. Hope that is clear enough. \$\endgroup\$ Aug 30, 2016 at 15:30
  • \$\begingroup\$ @SpehroPefhany I think you read my post backwards, though it's probably my fault for providing a third-party-generated schematic of my set-up. I am not using the AD623 +/- 9V instrumentation amp as depicted in the circuit. Instead I am using the single-supply INA-122 instrumentation amplifier, which is allowed to be used by 3.3V or 5V single supplies \$\endgroup\$ Aug 30, 2016 at 17:09
  • \$\begingroup\$ Yes, I know, but the circuit won't work on a single supply. The inamp will be fine. \$\endgroup\$ Aug 30, 2016 at 18:02

Even though I put a question mark at the end, this really might be your answer:

Typical input resistance of a common in-amp is 1000 GΩ. What voltage will develop at your input terminals when the bias current of 80 pA is allowed to flow into your in-amp?

Spoiler: Enough to saturate your amp. Therefore, always provide a DC path from your inputs to a reference DC voltage (usually ground).

Simple test: Will your circuit's behaviour improve once you connect both inputs (pins 2,3) to ground?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.