Are you intending this to run from the +5V Arduino supply? If so, the op-amps need to run on 5V [a lot of older ones didn't]. The outputs need to get to the rails [old bipolar op-amps mostly didn't], otherwise you'll restrict the range of the ADC input that you can use. You'll also probably want the inputs to work to the rails too (I said probably because we don't know the circuit yet) [again a lot of older bipolar op-amps were poor at this - doesn't mean they were bad op-amps, just inappropriate for what you want here]. If the scope is going to be fairly precise, you don't want the input bias currents and the current and voltage offsets to be too large, though that isn't too much of an issue here [traditionally, FET and CMOS op-amps were good for low bias currents but often poor for offsets - but that has changed a lot in the last decade or two - better automation means they can be trimmed at manufacture]. You also need enough bandwidth from the op-amp so that it can work properly over the range of frequencies you will be looking at with the scope. The quick measure of bandwidth is the GBWP (gain-bandwidth product) - that's the frequency where the gain of the op-amp falls to unity; it needs to be much higher than the highest frequency you are working with (how much higher will depend on the circuit configurations you use, but a GBWP of 5-10MHz is the right sort of area for 100kHz of bandwidth at the scope input as long as you're not asking for too much gain from the amplifier).

Presumably, you don't want to be paying $20 for it either.

That will reduce the number of possible contenders quite considerably. If you also throw in the requirement that it be a conventional package rather than surface mount (to make your prototypng easier) then you're probably down to just a few devices from each manufacturer (they'll all have something that will do this because they're agressively competing with each other for your chip business.)

Here's one possible contender. If you download the datasheet and look through it you'll see how it's specced and you should be able to locate most of what I've just gone through amongst the tables.

MCP6024 - Amplifiers and Linear - Linear Op Amps

Other manufacturers you could look at include TI (Texas Instrumenats), AD (Analog Devices), ON Semiconductor (don't know what the ON stands for!), and LT (Linear Technology).

What do you want to do next?

Either you could try doing something for yourself or we could give you a circuit to experiment with. If you've already done some op-amp circuits, you might want to try piecing something together for yourself as a learning exercise (and for the challenge), if not you might prefer to have an example circuit with an explanation to get you started and give you something to play with.

Hi Jon

Thanks for the detailed reply.

It helps allot.

Power supply:

I can run it from the arduino power and I can also do it from outside fr higher voltages.

As you mentioned, the old BJTs are not rail to rail, I actually encountered this issue

I have tried to make a probe from lm538n, and I got clipping (on both edges) when I used 5v power supply, and allot of noise. here is the capture:

This is what I actually wants from the probe:

1. Big R_IN for sensing the signal without changing it more than needed.
2. Bigger than 5v
   1. I want to be able to do 1/5, 1/10 division in some way. so I could sample 15v as input signal without tempering the arduino
   2. Does it means I need a specific slew rate requirement?
3. GBWP
   1. Well, I do remember that one from school
   2. If I plan to do division (as I mention in #2), does it means I can have smaller GBWP since I need less than unity?
4. Breadboard pins as you mention can also help

Also I will have a look on the datasheet you provided and try to come with other thoughts

You talked about "input bias currents and the current and voltage offsets", could you pleas elaborate more? I am not actually following you.

About experiments:

I had in mind doing Peter's power bench supply circuit.

I guess I could use other designs as well.

Actually I am very interested entering this world of building circuits from op-amps but couldn't find a tutorial with nice learning curve.

Thanks

Idan

Sorry about that, I thought you were a "maker newbie" still in the "noob phase".

The resources DAB pointed you at would be a good refresher of what you did in your degree. (DAB's much better at this support/mentoring stuff than I am - he quickly spotted what you needed.) There's lots of material on manufacturers websites - it's very worth while reading through their application notes. It's often much better than you'd find in a textbook.

Gain of a tenth is no problem - look at a differential amplifier configuration if you want to offset at the same time.

I think the slew rate largely derives from the same internal compensation capacitor that gets to define the GBWP, so the two tend to go hand in hand. But yes, calculate what you need and check it against the figures in the datasheet.

For best accuracy you want to be working where there's a good amount of open-loop gain, so though you could relax the GBWP spec a bit, in practice I think I'd still go for for a few MHz. If you look at a datasheet, it will normally give you a plot of the open-loop gain against frequency.

The bias current is the current in or out of the input terminal. For a bipolar device it's mostly the base current with a bit of leakage from any protection devices, for a FET input it's largely just the leakage of the protection. You can get some op-amps with very low input currents, down in the 2 or 3fA area, if you fancy building yourself an electrometer.

The offsets are the difference between the two inputs. Although the inputs match very well when fabricated on a single chip because the transistors are (obviously) all made in the same conditions and have very similar properties, there will still be small differences. For what you're doing here it won't matter much - particularly for a 'scope where everyone is used to adjusting the Y offset by hand - but if you were looking at more precise instrumentation, like the kind of thing a couple of the professionals here are evidently doing in their day jobs, it would play a real part in the design, so you want to get used to looking at the figures and comparing with other devices you know/ have used.

Sorry about that, I thought you were a "maker newbie" still in the "noob phase".

The resources DAB pointed you at would be a good refresher of what you did in your degree. (DAB's much better at this support/mentoring stuff than I am - he quickly spotted what you needed.) There's lots of material on manufacturers websites - it's very worth while reading through their application notes. It's often much better than you'd find in a textbook.

Gain of a tenth is no problem - look at a differential amplifier configuration if you want to offset at the same time.

I think the slew rate largely derives from the same internal compensation capacitor that gets to define the GBWP, so the two tend to go hand in hand. But yes, calculate what you need and check it against the figures in the datasheet.

For best accuracy you want to be working where there's a good amount of open-loop gain, so though you could relax the GBWP spec a bit, in practice I think I'd still go for for a few MHz. If you look at a datasheet, it will normally give you a plot of the open-loop gain against frequency.

The bias current is the current in or out of the input terminal. For a bipolar device it's mostly the base current with a bit of leakage from any protection devices, for a FET input it's largely just the leakage of the protection. You can get some op-amps with very low input currents, down in the 2 or 3fA area, if you fancy building yourself an electrometer.

The offsets are the difference between the two inputs. Although the inputs match very well when fabricated on a single chip because the transistors are (obviously) all made in the same conditions and have very similar properties, there will still be small differences. For what you're doing here it won't matter much - particularly for a 'scope where everyone is used to adjusting the Y offset by hand - but if you were looking at more precise instrumentation, like the kind of thing a couple of the professionals here are evidently doing in their day jobs, it would play a real part in the design, so you want to get used to looking at the figures and comparing with other devices you know/ have used.

Hi

Since I only did analog in the university and in my day job I am doing digital I still considering myself as noob

I think I will order your suggested opamp. But I am not sure I will get it in time for this competition.

I might open a different blog when getting it.

Regards

Idan

Yes, everyone is at first, but having done your course you should be able to make sense of datasheets and application notes, which a beginner couldn't be expected to do, so it makes sense to direct you to that material rather than explain things in lots of detail.

One other thing, 'rail-to-rail' is a marketing term and even though such op-amps are very good, they still fall fractionally short of the rails (often by a few tens of mV) - it's also very dependent on how the output is loaded (light load will get you closer). The figures are all in the datasheet. I'm assuming that won't matter to you, for what this is, but if it does then you'd need to run the op-amp on supplies that exceed the 0-5V of the Arduino.

If you're digitising, then in theory you need to be sure you don't have any energy at the input above half the sampling rate or you'll get aliasing when you reconstruct the waveform from the samples. So, depending on the kind of signals you want to look at, you might want a filter there. It gets complicated quickly, though, because a filter will start to change the phase below the cut-off, particularly if the cut-off is steep. Anyway, something else for you to read up on.

At the stage you're at, simulating some of this stuff would be a useful thing to do. Microchip have a model for that part - I tried it quickly and it worked ok in TINA-TI.

Hi Jon

Simulating the circuit might be a great idea.

I downloaded TINA TI from the official website and I have version 9.3.150.328

I followed the instructions on how to import a spice model and when I tried to compile the model I got an error:

Have a clue whats wrong? you mentioned you got it worked like a charm.

How to simulate the load

I need to have a behavioral model of the load.

Do you have a hint how to simulate the PAD of the ADC pin in the arduino? (A0 pin)

Thanks!

Idan

You don't need to be compiling anything.

Make a copy of the .txt file and change the extension on the copy to .lib.

Open the New Macro Wizard from the Tools menu.

Enter the Macro name and select 'From file' and point it at the .lib

click 'Next>'

You should see this

If you want you could accept that shape, but I prefer a traditional symbol so

select 'Load shape from library'

untick 'Show suggested shapes only'

use the down arrow to the right of the symbol displayed to drop the list down, move down to <Amplifier5G> and select it.

You'll then have this

click 'Next>'

You'll then get this screen for doing the pin assignments. For some reason it manages the output pin for itself but not the others.

You then drag the unconnected pins which are flashing to the pins you want them attached to on the symbol.

You can tell which pin numbers go where from the netlist shown in the window underneath.

click 'Next>' again

That will then bring up save dialog box and you can save it wherever you want.

Finally you'll get this message box and can insert a single instance onto the circuit diagram.

From then on you can insert additional devices using 'Insert Macro' from the 'Insert' menu.

Unfortunately, that standard symbol is upside down, so I mirror then rotate right twice to get the + supply pin at the top. If you're good with software, you might want to try reworking the symbol.

And this is what it looks like on the circuit

Do you have a hint how to simulate the PAD of the ADC pin in the arduino? (A0 pin)

Figure 24-8 in the Atmel datasheet for the processor gives an equivalent circuit for the input circuitry if you really want to model it.