3

I am building a simple inverting operational amplifier circuit where I am using a chip with dual amplifier. I am only using one of the two amplifiers. What should I do to the unused amplifier's input pins? Should I leave them floating or is there a good way to sort of terminate them.

dr3patel
  • 1,620
  • 3
  • 22
  • 31

1 Answers1

6

It depends a bit on the op-amp. If it's a bipolar op-amp with high input bias current in a non-precision circuit (referring to the other op-amp) it may be safe to leave the inputs open. Think LM324 in some non-critical circuit.

One common practice is to connect the op-amp as a voltage follower and tie the non-inverting input to some voltage within the input and output common-mode ranges. Since the unused op-amp is operating normally, all the datasheet guarantees apply, so this could be considered a best practice approach.

You can also tie the inputs to potentials such that the output (left open) is railed high or low (there may be some small power consumption advantage).

I don't suggest connecting the inputs to the same potential since the output of some units might drift around and slightly affect the other op-amp. Similarly, leaving inputs open on an amplifier with low input bias current (those with bias current cancellation or JFET/CMOS input op-amps) can cause pickup that will show up in power consumption and can show up as crosstalk in the amplifier that is being used.

Edit: As the actual amplifier OPA2614 appears to be a non-unity-gain stable very high frequency type, it would likely be best to either make a gain-of-five amplifier or apply a small DC voltage of a few hundred mV on the inputs and leave the output railed.

Spehro Pefhany
  • 376,485
  • 21
  • 320
  • 842
  • Yes, I am using the bipolar op-amp with ultralow input bias current. – dr3patel Jun 21 '15 at 15:58
  • @dr3patel bipolar and ultra low bias current? what part? maybe it has some fets for the input pair – Vladimir Cravero Jun 21 '15 at 17:59
  • http://www.ti.com/lit/ds/sbos305d/sbos305d.pdf – dr3patel Jun 21 '15 at 18:45
  • 1
    @dr3patel Arg. That's very high input bias current: -6uA typical, and quite high input noise current with an amplifier that is really high GBW (290MHz). The 2614 isn't even unity-gain stable, so the voltage follower would be a **bad** idea. I'd say tie the inputs with 100mV < Vin < 300mV across them and leave the output railed. – Spehro Pefhany Jun 21 '15 at 18:59
  • @SpehroPefhany: So I simulated using the 2614 spice model provided by TI where I connected both the non-inverting and inverting pins to the 200 mV source and connected output to the 12V rail. My op-amp rail-rail is 12V to 0V. But this configuration is drawing significant amount of current at the output i.e 400mA. Did you mean to say leave the output unconnected or connect it to 12V rail? – dr3patel Jun 21 '15 at 19:23
  • Leave the output open (it will sit close to the rail 'be railed') not connected to the rail. Glad you simulated this first. – Spehro Pefhany Jun 21 '15 at 19:29
  • I do the same as Spehro suggests but in addition I use zero ohm (or moderate value) resistors to perform the connection for the feedback and connection to the input. Then if I need to use the opamp in the circuit at later date I can easily connect to it by removing the resistors and using their solder pads. – Kevin White Jun 21 '15 at 19:43
  • 1
    @SpehroPefhany: yes leaving the output open (railed) works fine for both in simulation and breadboard prototype. – dr3patel Jun 22 '15 at 20:24
  • @SpehroPefhany: one last thing I didn't understand is the connection at the inverting and non-inverting inputs. Isn't it bad to physically apply the same voltage to both pins? Will it not drive the op-amp unstable as you also mentioned this earlier? – dr3patel Jun 22 '15 at 23:54