EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: rob.manderson on November 19, 2016, 12:31:48 pm
-
G'day,
I'm putting together a design using an LM324 op amp. The circuit only uses 1 of the 4 op amps in the package. I've searched but can't find much guidance on the question of what to do with the unused ones. Leave the inputs and outputs floating? Gut feel* tells me to set them up as unity gain non inverters and ground the input. (I.e., connect the output to the - input and ground the + input). Leave the output otherwise unused. Seems to me this will minimise noise pickup and possible adverse effects on the one op amp I am using due to internal chip effects.
Thoughts and recommendations?
* Of course gut feel shouldn't be factored into design decisions like this! :)
-
You have the right idea. Connect the output to the (-) input, but careful where you put the (+) input. It should be somewhere in the middle of your range. For example, if you're powering the opamp with +9V and -9V, then ground the input. If you're powering the opamp with +9V and 0V, then grounding the (+) input makes it sit on a rail, which is often a very bad thing. You'd want to have it sitting at some bias voltage of 4.5V, which you probably have kicking around anyway as a bias for your active opamp. If not, a simple voltage divider can provide the bias.
I happen have a paper on this, and some other stuff, kicking around somewhere for some reason. I'll see if I can dig it up and I'll post it if I find it.
-
You have the right idea. Connect the output to the (-) input, but careful where you put the (+) input. It should be somewhere in the middle of your range. For example, if you're powering the opamp with +9V and -9V, then ground the input. If you're powering the opamp with +9V and 0V, then grounding the (+) input makes it sit on a rail, which is often a very bad thing. You'd want to have it sitting at some bias voltage of 4.5V, which you probably have kicking around anyway as a bias for your active opamp. If not, a simple voltage divider can provide the bias.
I happen have a paper on this, and some other stuff, kicking around somewhere for some reason. I'll see if I can dig it up and I'll post it if I find it.
I think it'll be ok to ground the input with an LM324 because it's a single supply part. According to the datasheet it has an input common mode voltage range including ground.
Thanks for the input.
-
See, look here: https://www.maximintegrated.com/en/app-notes/index.mvp/id/1957 (https://www.maximintegrated.com/en/app-notes/index.mvp/id/1957)
So as I said, if you're powering from a dual supply, you can simply ground the (+) input, though it does no harm to use the voltage divider anyhow. For example, if you have +9/-9 supply, the divider will see 18V from the positive to negative rails, and the output will be 9V... And the negative rail (-9V) + 9V = 0V. :) So always using the divider is a nice, generic way of doing it that will always work no matter what, but in practice you usually already have a mid-scale reference voltage kicking around to bias the working opamp, or you just tie it directly to ground for a symmetrical dual supply.
-
It'll be fine to connect the inverting input to 0V, since the LM324 is a single supply part.
-
Why do you want to maximize power consumption?
-
Why do you want to maximize power consumption?
Yep - having read that Maxim page I see the issue. Ok, a voltage divider to set the + inputs to mid range it is.
Thanks
-
Good video from Texas Instruments about terminating unused op amps.
http://www.digikey.com/videos/en/v/Engineer-It-How-to-terminate-an-unused-op-amp/2606239929001 (http://www.digikey.com/videos/en/v/Engineer-It-How-to-terminate-an-unused-op-amp/2606239929001)
-
I have watched the video and read the links above and it's good advice but too general. I disagree with the point about always avoiding saturating the output, because some op-amps will actually draw less power when the output is saturated!
In the case of the LM324, it should use less power when the output is saturated at the positive rail, than when it's at half the supply voltage because the current sink enabling it to reach the negative rail will be switched off. I don't have my breadboard handy but I'll do a simulation if you're interested.
Op-amps with a class A output stage such as the CA5260 will draw less power when the output stage is saturated to either supply rail because one of the output transistors will be turned off.
http://www.intersil.com/content/dam/Intersil/documents/ca52/ca5260-a.pdf (http://www.intersil.com/content/dam/Intersil/documents/ca52/ca5260-a.pdf)
In some cases it may be alright to connect the inverting and non-inverting inputs to opposite supply rails. The OPA2227 drew an enormous current with its inputs connected to opposite rails because there are back-to-back diodes between the inputs. The NE5532 would also draw excessive current with the input connected to either rail but not all op-amps will do that. For example the MCP602 will be perfectly fine with its inputs connected to opposite supply rails.
-
Why not just use an LM321? That way no loose ends to worry about.
-
Why not just use an LM321? That way no loose ends to worry about.
True, but the 324 is half the price and a pair of 0805 resistors is much cheaper. :)
-
I think it'll be ok to ground the input with an LM324 because it's a single supply part. According to the datasheet it has an input common mode voltage range including ground.
This will likely drive the output into saturation which could be a problem in a low power design.
Some operational amplifiers can be "disconnected" by pulling the inputs to one supply or the other.
-
Hero's right that some specific opamps might have a configuration that's more optimal than tying it in the middle somewhere, but doing it like this should always be safe and reasonable. I'm trying to think of a situation where that's not true. Possibly someone can come up with one where you need to be more careful.
-
I did a simulation. I was right that when the output LM324 is saturated to the positive rail, it will use less power, than when it's at half the supply rail or anywhere in between. I was incorrect about why though. It has nothing to do with the 50µA current sink on the output. The main reason is, when the output saturates at the positive rail, Q12 will turn off, thus lowering the current by 100µA. Q1 to Q4 and Q11 turning off will also reduce the power consumption slightly but not by much.
According to the simulation, maximum current draw occurs when the output is at 0V.
The model I used is based on the schematic on the LM324 data sheet. It probably isn't 100% accurate but it should be close enough. Don't bother using the TI SPICE model, as the supply current draw won't be correctly modelled.
http://www.ti.com/lit/ds/symlink/lm2902-n.pdf (http://www.ti.com/lit/ds/symlink/lm2902-n.pdf)
(https://www.eevblog.com/forum/beginners/unused-opamps-within-a-multi-opamp-package/?action=dlattach;attach=271762;image)
-
One thing to watch out for on dual and quad operational amplifiers which a simulation will not reveal is that sometimes the individual operational amplifiers share bias circuits so driving one into saturation, exceeding its input common mode range, or other unusual operation may disrupt the others. I think some RC4136s (sort of a quad 741) had this problem.
The universal solution for a unity gain operational amplifier is to configure it as a buffer and hook the non-inverting input to some stable voltage well within its common mode input range. I am not sure what the best solution for a decompensated operational amplifier would be but they rarely come in dual or quad packages.
If you are doing a micropower design, then it pays to find out exactly what conditions minimize power draw.
-
One thing to watch out for on dual and quad operational amplifiers which a simulation will not reveal is that sometimes the individual operational amplifiers share bias circuits so driving one into saturation, exceeding its input common mode range, or other unusual operation may disrupt the others. I think some RC4136s (sort of a quad 741) had this problem.
I didn't consider that. I'm aware that the amplifiers in the LM324 do share some of the biasing circuitry so that could apply here. I suppose the only way to know for sure is to build a circuit and test it.
-
A voltage follower pinned mid rails. You cannot go wrong with this method.
-
If unused then use as a unity gain buffer for the analogue signal on another opamp, or as a mid rail reference buffer.
Another use, for those opamps that are not unity gain stable, is to split a gain stage over 2 opamps instead, improves bandwidth when you really need a high gain high bandwidth amplifier, but at the expense of slightly increased noise. If nothing else connect input to another opamp input ( providing input bias currents are not a worry as your circuit impedances are low enough that double bias current has little effect) and add a simple gain stage with high value resistors ( reduce output stage current, you do not really care if the unused output is going to be slew rate limited anyway) to give a gain of 2.
Another use is just add a low battery detector to drive a low current LED, or a power on indicator using the opamp as a low frequency relaxation oscillator to flash it slowly.
-
+1, if you have an extra opamp, just use it. Possible uses are an extra buffer, mid rail buffer or an active guard ring driver (output to the guard ring or to the shield of the input cable). Or, if your circuit has single-ended input, an extra opamp could make it differential.
If you already have all of these and there is still an extra opamp, you don't have a convenient place to connect the input to and you don't really care about the power consumption, just drive this opamp from your signal and don't use the output.
-
G'day,
I'm putting together a design using an LM324 op amp. The circuit only uses 1 of the 4 op amps in the package. I've searched but can't find much guidance on the question of what to do with the unused ones. Leave the inputs and outputs floating? Gut feel* tells me to set them up as unity gain non inverters and ground the input.
A couple of years ago I compared the supply current of several LM358 and LM324, connected as voltage followers, with + inputs connected to half Vsupply vs. GND (single voltage supply)
GND connection resulted in about 20% less current
both LM358 and LM324 were from ST
I do not know if ICs from other manufacturers are different, in any case I suggest you to check yourself with a breadboard...
-
For the LM324, you can just float unused pins, unless you have a lot of HF noise floating around.
Otherwise just ground the positive inputs, this will saturate in the lowest power dissipation mode.
The LM324 has a lot of bias current so little chance of it doing silly things.