That is very easy. Just remove the capacitor on the negative supply and it will work fine. Why is that: when chopping there flows a small switching Charge (e.g. 1pC) through the negative supply. This is not a problem, but when you add your capacitor it will translate this charge to a constant error voltage. In case of split supply your supply tries to compensate and this amplifies the error. You should have analyzed it with an oscilloscope not a multimeter. For any part which is designed for single supply it is a very bad idea to place a cap on the negative supply.

Dave, manufacturers do recommend to balance opamp inputs against bias current. In the test circuit you need to ground (+) input via 0.99kOhm resistor.

Yes, that was my first thought too. I usually try to equal the input impedances to reduce offset. Also, using just a bypass cap between V+ and V- (instead of two to ground) usually reduces noise.

Keep em coming Dave!!!

Nice to see, that you got your DaveCAD license renewed ;)

These videos are great. I'm currently troubleshooting my circuit that has an op amp input offset error issue.

With a chopper amp, as there is internal switching, I'd think the correct way to decouple would be across the power pins, not to ground, as the latter could inject noise into your ground reference. My guess is if you have your cct as shown and then touch a cap between the power rails you'll see the offset drop.

I think you're seeing AC Coupled current from the chopper circuit to ride the supply rails and is converted to an offset voltage across the resistance of the bypass caps. It's then added/injected back into the non inverting input. Placing the bypass across the rails works against the ping pong (push pull) chopper currents to effectively zero them. The I/V conversion is done across the rails with no path to ground so is sent back to the rails in opposite fashion.

consider matching the impedances at the opamp inputs, since the (small) input currents could lead to different voltage drops that you then see as offsets...

I had a similar problem with a Microchip chopper amp, the offset voltage was way out of spec. I found that in split supply configuration both plus and minus supplies had to come up evenly relative to the Ov reference (-ve input) in order for the chopper stabilisation to work properly. I would agree with @mikeselectricstuff, the decoupling should be across the power supply pins of the op amp, and probably decouple both supplies to zero volts close to the start zero volt point too.

Great video! What does the application note or sample circuit in the datasheet show for decoupling?

All right! There's the challenge post-it in DaveCAD (say namaste to the Indian in the post-its cupboard for us.) You changed the decoupling caps before shaking the specsheet at the camera too; so clearly what cannot happen in the next ep. is for the substantial DC leakage of the 302-type caps and/or input differences between the rail-bound inputs and supply to weigh in! So: grisly, twisted, medium-current family infighting between input isolation stage bias supplies internal to the opamp FTW.

I could certainly see that center tap to ground decoupling arrangement jiggling the 0v rail.. That particular opamp seems to have quite a lot of "stuff" going on internally.

Perhaps in needs a small load on the output? Thanks for for the videos Dave I do enjoy them.

Dave my first thought was that the inputs on the opamp aren't seeing the same impedance. Normal practice would be to connect the non-inverting input to ground via a 1k resistor. Having said that the input bias currents are only 100pA max on the data sheet so that should only account for 0.1uV offset. Still might be worth investigating.

Standby for the next video...

Thank you very much

I think the input offset voltage varies with input common mode voltage and probably supply voltage. When using split supply, there was common mode voltage with respect to the negative supply and when you vary the supply voltage both the common mode voltage and supply voltage are changing, causing the input offset voltage to change. With single supply, the input common mode voltage remains zero and only the supply voltage is changing.

21:27 happy 420!! :D

Awesome!

You are just awesome.

Thanks

Is it due to seeback effect on the one of the OP Amp pins? although the temperature difference is negligble, but on microvolt level this could be accountable. Analog Devices suggests to add two resistors from non inverting input to the ground so that the seeback effect could be neutralised. Two plus two. This will also ensure that the bias current is balanced properly on these two pins.

The function that gives that sort of curve is called a Gaussian function. The curve is still called a bell curve because of the shape of the graph.

was instructive, traps on single supply vs stabilities of duals...

Bit of a long shot but maybe thermal noise from the resistors. Try heating up the 1k and see what happens. Maybe one of your supplies could be off a tad. Move your ground up and down a bit and see if you can zero out the offset. Might give you a clue on to what's going on.

My thoughts too

+1 to this. Shouldn't make a difference in theory, but it's a simple test to discard polarization currents

You're doing a great measurement fault. The offset voltage is the voltage needed at the input, to get an output of zero volts (wikipedia and so on). In your setup, the output voltage isn't zero and therefore the input voltage difference isn't the offset voltage. If the output voltage would be zero, your input voltage would also be zero. That means you only make the right measurement, if the output and the offset voltages are zero.

With a single supply the offset is nill. With a dual supply the offset is ~.3mV It has to be the reference point. The gnd is floating, with the non-inverting input and inverting input pull-down resister not getting along... ?????

Redraw this to make it more obvious. Move the 1K around so it's between the + and -. So + is to ground, - through the 1K to ground, and then whatever output that generates wrapped back around through 100K to swing the - input. Sort of doubting that's the offset voltage measurement circuit they're using to spec the device. So the 320mV would be what the input offset/output errors etc induce when it's then put into a non-inverting amplifier circuit..

Nice !!!

Maybe you could probe around with the scope to see if there is anything weird.

Don't know about Wollongong, sorry. UOS is mostly for managers-to-be or academics-to-be. UNSW is the best best for engineering and electronics.

Pause AdBlock, Reload for EEVblog! :)

Doesn't that analog datasheet say it is a single supply op-amp(in large font), it doesn't say it can do dual supply.

Thumbs up. That resistor should be 100k || 1k though. See page 1.60 (p 62 of the PDF) in ADI and Walt Jung's Op Amp Applications Handbook (available for free at analog dot com) There is also a discussion a few pages up about measuring input offset voltage. (1.53, PDF p55)

Can you explain why the output current is not shown or specified on chopper amp data sheets?

I would like to know if he is using carbon resistors (noisy) or metal film or oxide resistors (much quieter) at the inputs of his Op-Amps.

hey dave! Ive recently graduated year 12 and wanted to be an electronic engineer and ended up in a bridging course at the university of Wollongong for electrical engineering (Ive been told the course contained electronic). Have you heard anything about this course? do you have an opinion on any other unis (UOS, UNSW, etc)? I know my question is kind of specific so you might not have info on it but its worth a try seeing as you're in the area & industry :) thanks!

Maybe it's because of the input current of the inverting input. Try to put an 1k resistor between the non inverting input and ground!

Fundamentals Friday: Datasheets and Debugging skills

i see you renewed your davecad licence

Sweet! Time to pop 'em popcorns! :D

Hi Dave,

shouldn't the non-inverting input be grounded alone and not connected to the inverting input? (I'm a total newbie to electronics!!!) and wont the caps be passing voltage to the ground through to Vos

is AD8628 a current feedback amplifier.

There one in tafe called electronic engineer check that that one

I tried several times to measure the OFF SET voltage of an LM 341, but it never worked. We attend the teachers' classes and everything seems easy, but when it comes to measuring, hundreds of problems appear that even they don't know how to solve. From Brasil.

More Fundamentals!

I see you renewed the DaveCAD license again, or did Mike crack it?

Note that the worst case input bias current (over a full temperature range) is 1.5nA - that would give ~1.5uV across a 1kOhm resistor. Also, the bias current will likely depend on supply voltage (more leakage at higher voltage) so this could be the case indeed.

Must be inherent to the AD chip since the other manufactured chips do not have that problem.

Dave, there should be three cables from power supply when using split supply (+, - and 0v), I can only see two.. could it be that you forgot to connect the 0v cable from supply? If so, the 0v reference is being determined by the parasit DC current of the capacitors and that is the reason the output changes according to the capacitors you used..

9:00 this is a gauss curve and not a bell curve..

I'd be interested to know how you're getting your ground reference... if you're using a potential divider, then perhaps the potential divider providing your feedback is pulling your ground about? Try measuring -ve to 'ground' as you wind up the supply voltage? See if it's solid.

Common Mode Rejection Ratio

Isn't there some Johnson noise voltage across that 1k ohm?

Any finds in the old garbage room lately?

I think you're referring to 'jeez' as in 'wow' or 'are you serious?!'. It may have reference to the name of a certain character in the Bible, who's name is often used as profanity - but I'm not sure about that.

CMRR?

7 years ago...what did AD say?

Dave, what do you mean when you say "G"? We are really good at English in Sweden, but some expressions are hard to understand. I'm guessing that the "G" is some kind of expression of "what they mean, why they did so, or what is the point of this?" Well... as a matter of fact, it's better than the f-, s- or d-words they use in America (=

I think opa189 may be a better chopper.

Nope, that's not it.

Maybe this chip would work better with some salsa. ;)

Please everybody, let's stop promoting proprietary patent-encumbered tools and instead support gDaveCAD, the open source alternative to DaveCAD!

yay electronics

having 0V 5V versus -2,5V 2,5V is the same, you still have 5V between each point. only that instead of using 2,5V as center, you would use 0V, but it's still +2,5V from V- It's just a matter of reference point, having dual or single supply doesnt matter, only if the opamp cant do rail to rail or for zero crossing.

hi

We're building a line of kits to go along with Dave's awesome Fundamentals Friday videos. Dave has given us permission to do this but please don't ask him for tech support on them, he hates that :) The site is still under construction but if your interested please check us out at: InfiniteReachSC com

Yes you're absolutely right. 990.09900990099.....

please change your voice modulation.

talks too much