Maximum speed of LM358 dual op amp ic

Thread Starter

-live wire-

Joined Dec 22, 2017
959
I got a bunch of LM358 ICs to experiment with op amps and see how they behave in real life. I have learned a lot about them but have not done too much with them in real life. I have been able to successfully use it as an amplifier and comparator.

I set it up so there was a current shunt being amplified and compared to a potentiometer. It is going to be a short circuit protection board. By ading a mosfet with gate resistor, controlled by the output of the comparator, I made an oscillator. While it was on a breadboard and the mosfet only had a gate resistor, it only oscillated at 44kHz.

So that go me wondering. What’s the maximum frequency it is recommended you use the LM358 at for good operation? I am looking for a rule of thumb type of thing, I know it can vary. Can you use it to compare or amplify a variable shunt voltage being switched at a few 100 kHz? What about a few 10s of kHz? I have a specific application in mind. Is the speed limit from parasitic capacities, and if so, which ones?
 

MrAl

Joined Jun 17, 2014
11,388
I got a bunch of LM358 ICs to experiment with op amps and see how they behave in real life. I have learned a lot about them but have not done too much with them in real life. I have been able to successfully use it as an amplifier and comparator.

I set it up so there was a current shunt being amplified and compared to a potentiometer. It is going to be a short circuit protection board. By ading a mosfet with gate resistor, controlled by the output of the comparator, I made an oscillator. While it was on a breadboard and the mosfet only had a gate resistor, it only oscillated at 44kHz.

So that go me wondering. What’s the maximum frequency it is recommended you use the LM358 at for good operation? I am looking for a rule of thumb type of thing, I know it can vary. Can you use it to compare or amplify a variable shunt voltage being switched at a few 100 kHz? What about a few 10s of kHz? I have a specific application in mind. Is the speed limit from parasitic capacities, and if so, which ones?
Hello there,

The actual max working speed depends on the bandwidth and the slew rate.
One way to do this is to calculate the bandwidth based on the gain bandwidth product on the data sheet, then check the slew rate. When you do both of those, you have what is sometimes called the "power bandwidth".

The slew rate comes into play when you need to have the amplifier follow a certain kind of waveform, such as a sine wave. The output peak to peak voltage then comes into play.
For example, say you have an output signal that goes from -2v to +2v and that works OK with the waveform you are using. That means that anything UNDER that will work too, such as -1v to +1v, but if you try to go higher like to -2.2v to +2.2v the slew rate may start to distort the output waveform and so it wont work right.
If you happen to be using a rectangular wave then you should know that the slew rate always affects it in some way or another. The time to ram up and down is finite and non zero. If this time period is too long it will make your rectangular wave looks like a triangle wave so you need to check this for each design.

If you post a circuit we can take a look.
 

dl324

Joined Mar 30, 2015
16,839
I am looking for a rule of thumb type of thing, I know it can vary.
It's usually in the datasheet.

The gain bandwidth (GBW) product for LM358 is 1MHz (for small signal). For large signal amplitude, use this:
upload_2018-8-13_6-49-50.png

You also need to consider slew rate:
upload_2018-8-13_6-50-25.png
Must be faster than 2ΠfV.
 
Last edited:

OBW0549

Joined Mar 2, 2015
3,566
So that go me wondering. What’s the maximum frequency it is recommended you use the LM358 at for good operation?
That information and much, much more is all contained in the data sheet. Read the data sheet. If you don't know what some of the terms mean and how they relate to the performance of a given circuit, ask.

I am looking for a rule of thumb type of thing, I know it can vary.
Can't give you a rule of thumb, as the answer would be situational.

Can you use it to compare or amplify a variable shunt voltage being switched at a few 100 kHz?
That would lead to major, MAJOR disappointment.

What about a few 10s of kHz?
That would lead to disappointment also, though not quite as bad as 100 kHz.

Is the speed limit from parasitic capacities, and if so, which ones?
No, parasitic capacitances have little to do with it except in very high-frequency applications.

Op amps have two inherent characteristics (both specified in the data sheet) that effect speed and frequency response: gain-bandwidth product and slew rate. Read up on them at the links.
 

Thread Starter

-live wire-

Joined Dec 22, 2017
959
That information and much, much more is all contained in the data sheet. Read the data sheet. If you don't know what some of the terms mean and how they relate to the performance of a given circuit, ask.


Can't give you a rule of thumb, as the answer would be situational.


That would lead to major, MAJOR disappointment.


That would lead to disappointment also, though not quite as bad as 100 kHz.


No, parasitic capacitances have little to do with it except in very high-frequency applications.

Op amps have two inherent characteristics (both specified in the data sheet) that effect speed and frequency response: gain-bandwidth product and slew rate. Read up on them at the links.
I think I understand them now. I wish I knew about them before buying those op amps. But I can still probably use a peak holder to achieve the desired result.
 

MrAl

Joined Jun 17, 2014
11,388
I still don’t fully understand what the gain bandwidth product is, and how it would affect a circuit. So it limits the gain at a certain frequency? And what factors
Hi,

The gain bandwidth product is a specification for a device that tells how fast it can work at a certain gain. It's just two specs multiplied together so you can calculate the max speed at a certain gain or the max gain at a a certain frequency.
For example, if you have a GBW product of 1000 then you can use it in a circuit that works at 10Hz with a gain of 100 because 10*100=1000, or you can use it in a circuit that works at 100Hz at a gain of 10 because 100*10=1000, or a circuit that works at 1000Hz with again of 1 because 1000*1=1000. So the product of the circuit gain times the frequency must be less than or equal to teh gain bandwidth GBW product.

This is also conditioned by the slew rate however, because for low distortion the slew rate must also be high enough to allow a sine wave at the chosen frequency to pass without being distorted. Since the fastest slew rate for a sine is at it's zero crossing, we have to make sure the slew rate is fast enough for the frequency we chose near it's zero crossing and that turns out to be:
slewrate>=2*pi*f*A

where A is the amplitude and f is frequency in Hertz.

So we see that the max frequency is determined by the gain of the circuit AND the slew rate AND the amplitude of the output of the device. That's five things you must know about the device and your application circuit:
1. intended gain of the circuit.
2. intended frequency of the circuit.
3. intended output signal amplitude.
4. GBW of the device.
5. slew rate of the device.

Example:
We have an op amp with GBW of 1000000 and slew rate of 500000v/sec and intended gain of 100 and intended frequency of 5000Hz and intended output of 10vpp.
First, the GBW of the circuit is 100*5000 which equals 500000 and that is less than 1000000 so this passes the GBW product test.
Now for the slew rate test, we must have:
sr>=2*pi*f*A
and so we have:
500000>=2*pi*5000*10
or:
500000>=314159

so it passes the slew rate test. This design should work reasonably.

Now if the intended frequency was 10000Hz, it would have failed the slew rate test because this following statement is not true:
500000>=628318

since the right is greater than the left. That design would have high distortion.
 
Last edited:

Thread Starter

-live wire-

Joined Dec 22, 2017
959
Hi,

The gain bandwidth product is a specification for a device that tells how fast it can work at a certain gain. It's just two specs multiplied together so you can calculate the max speed at a certain gain or the max gain at a a certain frequency.
For example, if you have a GBW product of 1000 then you can use it in a circuit that works at 10Hz with a gain of 100 because 10*100=1000, or you can use it in a circuit that works at 100Hz at a gain of 10 because 100*10=1000, or a circuit that works at 1000Hz with again of 1 because 1000*1=1000. So the product of the circuit gain times the frequency must be less than or equal to teh gain bandwidth GBW product.

This is also conditioned by the slew rate however, because for low distortion the slew rate must also be high enough to allow a sine wave at the chosen frequency to pass without being distorted. Since the fastest slew rate for a sine is at it's zero crossing, we have to make sure the slew rate is fast enough for the frequency we chose near it's zero crossing and that turns out to be:
slewrate>=2*pi*f*A

where A is the amplitude and f is frequency in Hertz.

So we see that the max frequency is determined by the gain of the circuit AND the slew rate AND the amplitude of the output of the device. That's five things you must know about the device and your application circuit:
1. intended gain of the circuit.
2. intended frequency of the circuit.
3. intended output signal amplitude.
4. GBW of the device.
5. slew rate of the device.

Example:
We have an op amp with GBW of 1000000 and slew rate of 500000v/sec and intended gain of 100 and intended frequency of 5000Hz and intended output of 10vpp.
First, the GBW of the circuit is 100*5000 which equals 500000 and that is less than 1000000 so this passes the GBW product test.
Now for the slew rate test, we must have:
sr>=2*pi*f*A
and so we have:
500000>=2*pi*5000*10
or:
500000>=314159

so it passes the slew rate test. This design should work reasonably.

Now if the intended frequency was 10000Hz, it would have failed the slew rate test because this following statement is not true:
500000>=628318

since the right is greater than the left. That design would have high distortion.
Thanks, but I watched a very good YouTube video that cleared up a lot. My finger slipped and I accidentally posted that before I could edit it.
 

MrAl

Joined Jun 17, 2014
11,388
I find LM358 to be sufficient for many applications. I have hundreds of them and they're my go to opamp unless I need something "better".
Hi,

I also second that nomination for the LM358 for the "Best Op Amp In The Universe" award :)

I have used it so many times i cant count.
For those that dont like the cross over distortion i hear about now and then, there is a way to get rid of that.
It is a general purpose op amp so yes there will be times when we need something a little faster or something with super low noise.

National Semiconductor has touted the slew rate as 0.5v/us for years but Texas Instruments has somehow managed to raise that a little. Or should i say that NS applied a conservative rating while TI was the one doing all the touting.
 

MrAl

Joined Jun 17, 2014
11,388
IIRC, mentioned in the very datasheet. And, IIRC again, a resistor between output and - V rail.
Hi,

Yes thanks for mentioning that.
For some time people complained about this and i think it was because back then it was not on the data sheet. I dont remember how i learned about it, but if we look at the internal schematic we can see how to keep the output biased for proper class operation so it never goes thought the zero crossing cross point.
 
Top