Hi all and thanks in advance for your help.

I am using an op amp MCP607 connected as a voltage follower (with negative feedback). Vdd is 5.0V and Vss is connected to ground.

When I pass 1V to (+)input I get 1V on the output, which is normal.
But If i pass 3.5V to (+)input I get 3.53V on the output, - well probably ok,
and if I pass 4.5V to (+)input I get 5V on the output. This is very strange to me.

This opamp has a 120dB Open loop gain and a 250uV input offset voltage.
Having read the article on op amps , I would expect to have a slightly lower voltage, say with a couple of mV, but not 0.5V.

What could be wrong?

 

Well the max input voltage is equal Vdd-1.1V = 3.9V.
Read datasheet "Common mode input voltage"
http://ww1.microchip.com/downloads/en/devicedoc/11177d.pdf

 

so up to 3.9V the input voltage should be followed exactly, right ?

Then why 3.5V is followed as 3.53V ?

 

Do you have any load on the output of the opamp besides the noninverting input connection?

Do you have a 0.1uF bypass cap from the Vdd terminal to GND?

With no load and a bypass cap wired as a voltage follower, your output voltage should be within 250uV of the input voltage - as long as you're within the specified common mode voltage range.

If not, your opamp may have been damaged or may be defective.

 

thanks , you turned out to be right my op amp was out of its common mode.

What op amp would you recommend for a good voltage follower to drive
ADC measurements ?

 

Something like an LT6003 would be good.

Basically, you're looking for rail-to-rail inputs and outputs, low offset, and low drift that is stable at unity gain.

 

I agree. TI has a number of very good CMOS single-supply, rail-to-rail op-amps with a higher common-mode range than 3.9V on a 5V supply. Try looking at the TLV341 (click the attached link for the datasheet)http://focus.ti.com/lit/ds/symlink/tlv341.pdf . It has common-mode range up to 0.5V less than the supply, and rail-to-rail outputs. But here is what I really want to say: analog layout is critical, even in a run-of-the-mill unity-gain follower circuit. Here are a few things you must do to ensure trouble-free operation:a) Don't layout the circuit on a press-in type prototyping board. These are fine for low frequency digital and discrete low-gain analog, but bad for most analog, especially high-gain op-amps. b) Do layout on a vectorboard with pre-drilled holes and a copper, circuit board type prototype pattern. C) Do bypass the op-amp IC directly across the supply pins as close to the pins as you can get with a 0.1 uF and perhaps a 0.01uF in parallel with that. When I say direct to the pins, I mean it literally. The best way is to cut the cap leads very short and solder them on the bottom-side of the proto Vectorboard right on the supply and ground pins themselves. D) Do ensure you have a fairly clean power supply. If the vercotboard is more than 4-6 inches away from your power supply circuit (not including the power supply cable and/or connector), add a 10uF to 100uF across the power supply input on the Vectorboard. E) Place the feedback connection from the op-amp output to minus input very close to those pins and keep the wiring short. If using feedback resistors from the output to negative input, place the resistor with as short a lead as possible to the minus input side, and the other end going to the amp output can be longer and is far less sensitive to noise and external pickup. F) closely read the datasheet and especially it's applications section for any special requirements, like adding an output cap to the amp circuit to keep the circuit stable at unity gain. Make sure sure are within all the recommended limits of the IC (do not use the Absolute Maximum Ratings, as these are generally for transient conditions and not for continuous use, or be prepared for unexplained failures). It's a good idea to use a series resistor on the non-inverting (PLUS) input in your unity-gain follower. Not too high a value, perhaps 10K to 100K to protect the amp against input transient overvoltages. G) Although I don't know what your load on the amp is, make sure that the load is within the capabilities of the op-amp or else you may need a buffer amp to boost the output current with an NPN emitter follower or similar, although there is a price to pay with losing the rail-to-rail output on the positive side (you'll lose about 0.7V in the follower ckt). Make sure the op-amp slew rate is fast enough for the type of input you're using. If it's a slowly changing DC input, no need to worry about it, but if you're using fast square waves, you must worry about it. Make sure the amps GBW (gain-bandwidth product) is fast enough for the frequency of signals (or rep rate) of your input. H) Be careful of any unused inputs. If you're using a dual or quad op-amp and one side is unused, tie both unused inputs to ground. The TLV341A I recommended has a shutdown pin. Make sure to hardwire it to the proper point to disable that input so that noise doesn't falsely shut down your amp circuit. I hope this helps.
Good luck
Kamran Kazem

 

I will try to look for tlv342 but in the mean time I am playing with mcp6142 op amp.
Common mode from Vss-0.3V to Vdd+0.3V, input offset 10mV, rail-to-rail output.

So when my +input has 4.82V, on the output I get 4.92V. This is 100mV instead of 10mV. Is that normal?

The op amp is connected as a voltage follower,Vss to ground, Vdd=5V, no bypass cap, no load on the output pin.

 

with a gain of 10 it would be

 

From Microchip's MCP6142 page (link: http://www.microchip.com/wwwproducts/Devices.aspx?dDocName=en010453 )

MCP6142 - 600 nA, Non-Unity Gain Rail-to-Rail Input/Output Op Amps (MCP6141/2/3/4) (5/14/2009)
The MCP6142 is a dual 600nA op amp offering rail to rail input & output over the 1.4 to 5.5V operating range. This amplifier has a typical GBWP of 100kHz with a maximum quiescent current of 1 microampere

Rail to Rail input/output swing G>10 stable

So, without even looking at the datasheet, you should have known that you would need a gain > 10 for it to be stable.

You cannot operate that opamp at unity gain.

The 10mV offset really is quite high.

It does not meet two important specifications that I gave you.

I would not have selected that opamp.

The devil's in the details.

[eta]
I'll add a couple more items to the list:
1) Don't use a micropower opamp for your buffer/voltage follower.
2) use an opamp that has at least 1MHz bandwidth.
You want the opamp to bring the ADC input very rapidly to the voltage that is at the noninverting input of the opamp, and be able to hold that voltage steady while the ADC is sampling it, which means a low impedance output from the opamp.

A 10nF cap at the ADC input will help to keep the voltage stable while it is performing it's sampling, but you need an opamp that can respond rapidly enough, and supply enough current to keep that cap charged to the proper level.

If you have to drop one of my suggested specifications, then ditch the "low offset" specification, as you can compensate for the offset by subtraction in your ADC sampling routine.

 

Sarge you are so much more eloquent than I.

OP, Im guessing you have this opamp laying around and are looking for something to do with it? Is this to educate yourself with the workings of opamp or is there a specific circuit you wish to build?

 

Sarge you are so much more eloquent than I.

Eloquent?

I'm just researching, and then stating the facts as I know them.

Really, a big part of electronics is just knowing how to look stuff up, correctly interpreting the stuff you find, and getting more specific in your criteria until you get just the right item for what you want to do.

OP, Im guessing you have this opamp laying around and are looking for something to do with it? Is this to educate yourself with the workings of opamp or is there a specific circuit you wish to build?

Our OP needs to buffer a high impedance voltage source to provide a low impedance input to an ADC.

 

Well, it seems I bought the wrong op amp again

To be honest, half of the things in the specification I do not understand.

GBWP, this is the maximum frequency at which the amplification will be stable ?

Right. The devil's in the details!

 

How about mcp6022 ?

 

How about mcp6022 ?

That should work OK.

 

Hi,

I finally got my hands on mcp6022 and it does not work as expected

The op amp is connected as a voltage follower, Vdd = 5.0V stablised from a 5V voltage regulator, +input is 1.39V and output is 1.42V ?

This is a difference of 30mV ? Shouldn't the output be 1.39V ?

According to specs this is a unity gain stable op amp, 500uV voltage input offset and common mode range Vss-0.3V to Vdd+0.3V.

If input voltage is higher then difference btwn +input and output gets higher.

 

The op amp is connected as a voltage follower, Vdd = 5.0V stablised from a 5V voltage regulator, +input is 1.39V and output is 1.42V ?

Where does this voltage comes from?

Are you measuring input voltage and output voltage with one meter or two?

Could the impedance of your meter loaded down the input voltage, with its usual 10MΩ input resistance?

 

The +input voltage comes from the 5V regulator through a voltage divider.
The power supply is 12V , 500mA.

I am using one meter only.

Could the impedance of your meter loaded down the input voltage, with its usual 10MΩ input resistance?

Not that it is not possible , but I doubt it.

some more measurements:
when +input is 1.03V -> output is 1.04V
when +input is 3.81V -> output is 3.9V

very strange!

 

Do you have both a 1uF and a 0.1uF capacitor across the supply pins of the opamp?

If you look at the test schematics, that is what they are using.

The 1uF can be an aluminum electrolytic.

The 0.1uF should be a metalized poly film or ceramic cap.

 

Not that it is not possible , but I doubt it.

You don't doubt, you find out and eliminate the possibilities.

What is the values of your voltage divider?

It doesn't take a genius to work out if their values are like those in the following simulation, then it will be exactly the type of measurement results you are getting.

Use a proper voltage source or voltage divider made up of resistors in the low KΩ. e.g. a 3K9 and 1K resistor to get 1.02V and test again.