# What makes a precision op amp precise?

I'm making a dummy load which can draw 5V/3A maximum, but can also draw small loads (e.g. to drive an LED array of 20mA).

simulate this circuit – Schematic created using CircuitLab

I have selected a 0.5 Ohm shunt resistor after looking through a bunch of datasheets of mosfets.

The positive input is controlled by a coarse and fine adjustment pot, so it can be varied accordingly until 20mA is seen through a multimeter.

To get 20mA, I would need 10mV at the op amp inputs. From the datasheets I've looked at on precision vs. non-precision, the big difference I see is input offset voltage. In this application, is this parameter critical? I believe it isn't, because I can just adjust the pot, but I want to make sure.

Regular op amps such as an LM324 says it has a minimum common mode input voltage of 0V. Are there any benefits for using a precision op amp here? Will it increase the fine grain adjustment of the output voltage (if my input's resolution is very large)?

• The precision of a opamp is proportional to the number of marketing people on the project. Dec 24, 2014 at 15:23

The LM324 has a maximum offset voltage of 9mV (worst case, over temperature), according to the datasheet.

With your circuit, with 0V in, you could have a current of 9mV/500m$\Omega$/9mV = 18mA below which your pot would not be able to set the current. So it's not a very good design if you need to set it to less than 18mA. It's luck of the draw- the next op-amp (even in the same package) could be 9mV in the opposite polarity, so you'd just move the pot.

Maximum temperature drift of the LM324 is not specified (it's not intended for precision applications, after all), but it might easily be +/-10uV/°C, so if the board changes by (say) 70°C as the MOSFET gets hot, the current will change by 0.7mV or 1.4mA, so you'd have to readjust the pot. Of course the highest power dissipation occurs at high output currents, so the change is relatively small (1.4mA out of 2A is < 0.1%). A 20°C change in ambient temperature means a change of perhaps (no guarantees) of 0.4mA, which is several percent of a 15mA current. If you only care about 5%, and currents above 20mA, probably just okay.

Another difference between a cheap amplifier and a good one is the gain. The LM324 can be as bad as 25,000 gain (and it changes with temperature). A precision op-amp will have a gain in the millions. The difference will show up in how well it compensates for load or line changes (not a big deal in this case).

The bias current of the LM324 can be as bad as 0.5uA (typical 20nA) and it changes with temperature so if you had a high resistance pot, you could see it change with temperature.

The noise of the LM324 is a fairly miserable 35nV/sqrt(Hz), and it has nasty crossover distortion, neither of which affects you much in this case.

A couple of things (other than being extremely cheap) that the LM324 has that a typical precision op-amp may not have- wide supply range (especially on the high end), though it may not do so well at very low supply voltages, and it's single supply (input common mode range includes the minus supply) which you absolutely require for your circuit.

So there are plenty of reasons to use a decent op-amp if it's required by the specifications. Or you can get clever with the circuit- increase the sense resistor to get good accuracy for low currents, but to get wide dynamic range, a good amplifier (and other techniques such as good resistors and good layout) may be worth it. For just hacking around and if your current range not huge (minimum to maximum), an LM324 is certainly acceptable. There's no point in using a \$5 op-amp if a 1-cent one will do. On the other hand, there are some requirement for which the best ones are not good enough and one has to resort to discretes and other techniques.

By the way, your circuit may not be stable against oscillation. It can be fixed with some passive components, but loading op-amps with the equivalent of a large capacitance in series with a small resistance is inviting trouble.

• Stability concern is especially true with LM324 which has a phase margin of only ~45 degrees without capacitive loading. Dec 24, 2014 at 20:04
• you said "...On the other hand, there are some requirement for which the best ones are not good enough and one has to resort to discretes and other techniques.", can you give an example? May 10, 2015 at 18:50
• @quantum231 High voltage, high current, extreme low noise with high input impedance are all examples where monolithics sometimes don't have adequate performance. In all those cases, significant improvements can be had by using discretes or other techniques. An example of "other techniques", if input impedance is not important but low noise is, then you can get an N:1 improvement in noise by combining $N^2$ op-amps. May 10, 2015 at 18:59

A precision op-amp usually has a much smaller temperature dependancy of offset voltage therefore you can better rely on it over temperature and time. Ditto offset currents although in your circuit it's hard to say because you haven't stated your pot values on the circuit.

The main point of precision is repeatability. The more you can predict what a parameter will be the less you have to do to mitigate variations in that parameter. If you can use a more precise device in your circuit and thus do away with any manual adjustment or other value tweaking, then the extra cost of the precision component is easily offset by the reduction in production costs.

As you have a manual adjustment stage anyway, then precision is irrelevant.

I only wonder what should be the threshold voltage of the MOSFET at this power supply voltage of 5 V and a maximum current of 3 A through a current sense resistor of 0.5 Ω...