Op Amp output.

[dan3460]

I'm working on the circuit below, it is a simple circuit designed to take a triangular signal coming between 0-3V and amplify it to 0-18V. I'm using a LM358 with single supply. When I calculated the circuit I did it for the 3V and the numbers shown gives me the required output, the peak of the wave is at around 18V and the bottom near 0V. My problem is when the incoming signal is only 0-1V, the output has the DC offset added to the signal, in other words the amplifications is ok but it is offset by around 7V.
I thought on adding a cap at the end to strip out the DC portion but it also makes part of the signal negative with respect to ground. I need to make the output to start always as close to ground possible. I'm probably adding a couple of transistors in a push-pull configuration to increase the current that this can handle.

[jwm_]

You just want to multiply by N preserving the dc component.

So drop the cap on input, that drops th dc component you want to keep. Then for your non-inverting inputs, you don't want to DC bias them up with your resistor divider, since the DC bias comes from your signal., send them to ground like normal. so you are multiplying by  -sqrt(N) * -sqrt(N) = N.

You can also probably do it as a single non-inverting op amp, and if you have a op amp that can go to ground on input and output you won't even need to supply a negative voltage.

[dan3460]

Thanks for the quick answer.
I'm using a LM358, I think the output can be very close to rails. I'm using the two op amps just because there are in the package. I understand the about taking the cap and driving the + input to ground, i'll try it this evening.

[SeanB]

LM358 can only provide 0V on output with a pull down, so you will need a resistor from each output to ground to bias it into class A and enable it to pull down to 0V. 10k would probably be about right, one from each output to 0V.

At least with the input voltage range you have selected you will not run out of input voltage range.

Look on YT for Alan Wolke W2AEW for his recent video on using an opamp on a single supply voltage.

[dan3460]

Thanks for the info.

[Zero999]

Why are you using two inverting amplifiers when a single non-inverting amplifier will do?



Using standard E24 resistor values:
Gain of 18:
V_IN = 1
V_OUT = 18

Therefore:
R1 = 10k
R2 = 2k
https://en.wikipedia.org/wiki/Operational_amplifier_applications#Non-inverting_amplifier

The LM358 will need a 24V power supply to reliably output 18V. If more output current is required, the spare op-amp can be connected with it's output in parallel via a low value current sharing resistors.


https://e2e.ti.com/blogs_/b/analogwire/archive/2013/03/26/versatile-tools-to-simplify-your-designs

[dan3460]

Just because it was there in the package.

[Zero999]

Just because it was there in the package.

But it makes things worse/harder not better/easier.

To use two inverting amplifiers, both voltages need to be relative to the same point, so you would have needed a negative power supply. Biasing it so the reference is ¹/₃ V_CC doesn't work because all voltages (both the input and output) became relative to ¹/₃ V_CC.

Two non-inverting amplifiers cascaded would be better but it would still introduce additional errors due to resistor matching and bias currents.

The best use of the spare amplifier is to connect it as shown above to increase the output current or configure it as a unity gain stage with its input connected to the other amplifier's output so it can be left as a spare.

[dan3460]

point taken.

Thanks

[jwm_]

If you already have dual supply rails, the double inverting isn't a bad way to go. You will get less distortion in general because your common mode voltages are in the center of the range.

 Plus you have a lot more choices in op amps since you are not restricted to ones that can go down to the bottom rail and accept input all the way to it. In general, inverting configurations are better unless you can't use one for some reason.  It's probably not worth installing a negative rail for, but if you already have one, go for inverting.

[dan3460]

ok, hero you were right. The setup proposed accomplish what I wanted to do, also in my final product I will put a parallel op amp to bust current.

Thanks again,

[FlyingHacker]

The LM358 will need a 24V power supply to reliably output 18V. If more output current is required, the spare op-amp can be connected with it's output in parallel via a low value current sharing resistors.


https://e2e.ti.com/blogs_/b/analogwire/archive/2013/03/26/versatile-tools-to-simplify-your-designs

How does something like this compare to adding a unity gain follower after the circuit but before the load? The method above would be less prone to oscillation, or what? Always looking to learn more. Any info on the differences and similarities and use cases would be very helpful. Thanks.

[Zero999]

How does something like this compare to adding a unity gain follower after the circuit but before the load?

It's totally different. A unity gain follower has a high input and low output impedance and can be used to reduce the loading on a signal source.

In the configuration above, both op-amps provide current to the load so the output current capacity is doubled.

[FlyingHacker]

How does something like this compare to adding a unity gain follower after the circuit but before the load?

It's totally different. A unity gain follower has a high input and low output impedance and can be used to reduce the loading on a signal source.

In the configuration above, both op-amps provide current to the load so the output current capacity is doubled.

I see. So in the configuration above you can put out more current than a single op amp can put out? (double, I guess?)

Could you also use a follower with an op amp model capable of handling more current to do the same thing? Or am I thinking of this wrong?

[Zero999]

How does something like this compare to adding a unity gain follower after the circuit but before the load?

It's totally different. A unity gain follower has a high input and low output impedance and can be used to reduce the loading on a signal source.

In the configuration above, both op-amps provide current to the load so the output current capacity is doubled.

I see. So in the configuration above you can put out more current than a single op amp can put out? (double, I guess?)

Could you also use a follower with an op amp model capable of handling more current to do the same thing? Or am I thinking of this wrong?

That's that's a possibility, although in that case it would be better to select a buffer with a much higher output current and put it in the feedback loop, rather than trying to make it share current with the op-amp.

[dan3460]

The circuit that I'm building is a transistor curve tracer. I took one project that I saw on the internet, build it and I was disappointed with its performance. So I experimented using my signal generator and got the curves that I was expecting. So, now I'm working my way backwards making the circuit a better performer. I have been collecting antique radios for a long time but only for a year or two starting to work with semiconductors, so there is a bunch of things that I'm learning.
Anyway, on the original circuit the op amp was followed by a pair of transistors in a push/pull configuration to support more current.
I will be testing the configuration that hero999 proposed and see if the current will be enough to do what I want to do, if not I'll try adding a couple of transistors.

[FlyingHacker]

Could you also use a follower with an op amp model capable of handling more current to do the same thing? Or am I thinking of this wrong?

That's a possibility, although in that case it would be better to select a buffer with a much higher output current and put it in the feedback loop, rather than trying to make it share current with the op-amp.

Right. It would be silly to use two (especially different) op amps for this. I guess I just wanted to make sure you could do it with a follower with the proper current handling. That was always a "building block" in my mind for when I needed some additional current and did not want to load the previous output.