Why do opamps in this configuration get very hot?

[disorder]

I have two unused opamps (each "amp" is a from a different IC) and they are setup such that each output goes to an input and the remaining input is grounded. This is what I know as the usual way to handle unused opamps. Now because I am dealing with two separate dual opamp ICs (the other "amps" on each IC are being used), and because this a proper PCB, I can't avoid the connection that joins one input of one amp to the input on the other. I believe this is why both ICs are getting excessive hot. But I am failing to understand why this is the case?? Any ideas??

EDIT - NEW SCHEM OLD ONE WAS INCORRECT SORRY!

[armdnrdy]

http://e2e.ti.com/blogs_/archives/b/thesignal/archive/2012/11/27/the-unused-op-amp-what-to-do.aspx

Which "proper" PCB are you referring to? Do you have  photo or link?

[Seljer]

You've got a feedback loop on the positive input, you've probably created an oscillator. Or two oscillators, and you're running their outputs into each other.


Grounding the positive input with the feedback wire going from the output to the negative input would probably be the better choice. And if you haven't got a bipolar supply, tying the positive input to Vcc/2 voltage would be better still.

[octfrank]

You're feeding back to the positive input, very bad. In this config the transfer function is (if I did this correctly):

Vout/Vin = (A-1)/A

Where A is the gain of the opamp. You have Vin set to 0 so you are basically running at infinite gain (Vout/0), drawing lots of current and burning up the opamps.

[disorder]

I @#$%ed up again on that schematic!! Sorry, I should note I fed back into the inverting input (PIN 2). Sorry again, even with feeding back into PIN 2 and having the PIN 2 connected together both opamps still get hotter than hell.

[Ice-9]

+ op amp input to ground, feedback loop from -input to output, no connection between the op amp feedback . should do it.

[octfrank]

Well could be latch up/phase inversion, TL072 can suffer this if the input is too close to a power rail. Drive the non-inverting input with a reference voltage like 1/2 Vcc rather than tying to a rail. I am assuming that gnd is also the negative power rail for the TL072.

[induction]

+ op amp input to ground, feedback loop from -input to output, no connection between the op amp feedback . should do it.

+ op amp to Vref is better. Shorting it to ground rails the output low. If you're not using an adapter maybe you won't mind the extra current consumption, but with a battery you will.

[disorder]

This is a +/-12V power supply and they are NE5532.. I didnt have the NE5532 package in EAGLE so I used TL072.

[armdnrdy]

This is a +/-12V power supply and they are NE5532.. I didnt have the NE5532 package in EAGLE so I used TL072.

You can change the name that shows up on the schematic by changing the value.

[R.G.]

This is a bad way to inactivate unused sections. Opamps should be tied output to inverting input, and the + input to a vref that's within the common mode range of the opamp. Check the datasheet to find this. As noted, TL072s will invert their output phase if the inputs get too close to V-.  It's not all that good for any input whose datasheet does not explicitly say something like 'the input common mode range includes the minus power supply' or 'rail to rail input'.

You'd probably have gotten away with it if the outputs of the opamps were not tied together. It is, in general, a bad idea to simply tie low impedance points together. They will go fight over millivolts. There is an assortment of unfortunate, smoke-releasing results that can happen.

You might get an OK result simply isolating the opamp outputs, leaving each output connected to its respective (-) input. Otherwise, you're going to have to supply a Vref to the + inputs as well as disconnecting the outputs.

[disorder]

Thanks all, this makes sense now. I always knew the dangers of leaving opamp pins floating but never knew the proper procedure for dealing unused opamp.

RG, you've sparked a few more questions. I am using a bipolar +/-12V supply, and the NE5532 has a common mode input voltage range of +/-13V however is not a "rail-to-rail" opamp, what should the +input be tied to? I now see why it's bad form to tie +input to GND when using a single pole power supply though.

Also, I broke the pins of one of the NE5532 before going into the socket so I could measure current. None of the pins (output, inputs, and power) were seeing anything above 20uA. Wheres the heat come from?

[armdnrdy]

Also, I broke the pins of one of the NE5532 before going into the socket so I could measure current. None of the pins (output, inputs, and power) were seeing anything above 20uA.

[disorder]

Do you not agree with my use of the word "broke" 

I mean I individually disconnected different pins to the NE5532 and hooked up a current meter. I was expecting to find a point at which the current measured excessively high. Basically, I was looking to find out where the heat/current was coming from or going to for that matter.

[armdnrdy]

The word "broke" gave me the impression that you cut a pin off of a perfectly good op amp just to check the current.

I only break the pins off of ICs on accident. Like when I purchase ICs online and they get shipped stuck in foam in a regular envelope...then the post office runs them through their sorting machine with rollers because they are not marked
"HAND SORT ONLY" "FRAGILE"

The pins usually break off when I try to convert them from pancake ICs back to a usable form. 

[R.G.]

RG, you've sparked a few more questions. I am using a bipolar +/-12V supply, and the NE5532 has a common mode input voltage range of +/-13V however is not a "rail-to-rail" opamp, what should the +input be tied to? I now see why it's bad form to tie +input to GND when using a single pole power supply though.

Actually, it can be tied to any voltage inside the common mode range of the opamp in the application's power supply as used. If you're using +/-12, you need to put the + input somewhere between +/-10V.

10V? Where'd that come from? The datasheet says the ICMR is +/-13 but the conditions are probably stated as with a +/-15V power supply, or two volts less than the rails. So I subtracted 2V from each of your actual power supplies. With a bipolar supply, ground is a fine bias voltage.

So with that not a problem, you're likely seeing the results of tying the two outputs together - or there is a different problem.

Also, I broke the pins of one of the NE5532 before going into the socket so I could measure current. None of the pins (output, inputs, and power) were seeing anything above 20uA. Wheres the heat come from?

Good question. Could be that mysterious "different problem" I mentioned. It could be oscillation from the two outputs tied together, or it could be oscillation from poor power supply decoupling, or some other quirk of the layout.

[PRR]

I want to know why the outputs are tied together. If the PCB forces that, it's wrong. Get your sharp knife and bright light, cut the trace.

Since you have +/- supply, ground would be a fine place to tie the input.

With perfect opamps in this connection, both outputs would be Zero exactly.

There are no perfect opamps. In particular there is an Offset Voltage, typically a few milliVolts either way.

So one output could be +0.005V and the other could be -0.005V. 0.010V different.

These two *different* voltages are then connected together with a PCB trace. What is the Ohms? Let's say zero Ohms. 0.010V/zero is "infinite" current from one opamp to the other!! Even if the copper is 0.1 Ohms, that is 0.010V/0.1= 0.1 Amps.

The TL072 will only pass ~~0.030 Amps into a short (internal self-protection).

At +/-12V and 0.030 Amps, each opamp is dissipating 12V*0.030A= 0.36 Watts.

The conservative spec for a DIP-8 package is 300mW or 0.3 Watts.

0.36 Watts in a DIP-8 will be HOT.

(At room temps, even 0.6W in a DIP8 will not be instant death, so the chip is probably fine, but does not need the abuse.)

*None* of this current is doing you any good. These two "idle" opamps may be eating more power than all the rest of your pedal.

Connect inputs together (sometimes), NOT outputs (some exceptions).