Op Amps: DC Offset

DC Offsets are undesired DC voltages that appear at the output of an amplifier. They lead to ambiguity and inaccuracy of DC output signals. We investigate how to compute their worst case magnitude and how to mitigate their effects.
References: Operational Amplifiers and Linear Integrated Circuits: Theory and Application; Chapter 5, section 5.
My free texts and lab manuals are available for download at my college web site www.mvcc.edu/j... and at my personal site www.dissidents.com
Inexpensive print and kindle copies are available at Amazon www.amazon.com...
If you like my texts and videos, and would like to help defray the costs of making and maintaining them, consider making a modest donation at www.buymeacoff...

[@simonyoungglostog]

Great. I'm looking forward to getting to this in the Text. I'm still on Bode plots.

[@stephaniebeals5936]

Hello! I have a question. If i WANT my op amp to produce a DC offset of a certain kind. Can I just reverse engineer this? Would I still need the Roffset resitor?

[@ElectronicswithProfessorFiore]

If you want a DC offset you probably don't need Roffset. If you need a large offset (like a few volts), you might look into summing in a DC level (see "summing amplifier").

[@deviantmultimedia9497]

In the case of an audio circuit with multiple op amps cascaded in a unity gain filter configuration, is it necessary to remove DC from the signal in every stage or just the last op amp of the filter chain? For example, a unity gain low pass filter with no series capacitors.

[@ElectronicswithProfessorFiore]

You could do it at the final output. Generally the offsets won't be so large as to limit the dynamic range of the individual stages so you don't have to do it at each one. OTOH, if there is a potentiometer in the signal chain, you'll want to get rid of the DC before it. The reason is that with a less than ideal pot (i.e., dirty), any DC voltage will create level shifts and transients when the pot is adjusted. This will be heard as a scratchy noise.

[@deviantmultimedia9497]

@@ElectronicswithProfessorFiore OK great. That saves me from a lot of math. Could I possibly use single-amp op amps as buffers before entering the power amplifier and utilize their offset adjust pins? The datasheet is telling me not to. I guess because they only intended for the pins to be used to adjust the offset of that op amp. Do I have to listen to that? They also say that the op amp is internally trimmed and doesn't need adjusting. Can I listen to THAT? I'm just trying to avoid using a capacitor on the input of the power amp and I definitely don't want to design a DC servo circuit. But at the same time I want to eliminate the chance of DC reaching the output section. Is this a "wish in one hand and sh*t in the other type of situation?

[@ElectronicswithProfessorFiore]

@@deviantmultimedia9497 Not sure what you mean by "the datasheet telling (you) not to". Offset adjust pins are made for adjustment. The downside of manually nulling an op amp is the cost of a precision trim pot (don't use a cheap 3/4 turn, get a 20 or 25 turn) and the fact that you have to adjust it (time/labor). Over time it might go out of cal, too. Most recent generation op amps have very low DC offsets to begin with (assuming you're using a typical design- see details in the text). We're talking order of magnitude reductions compared to first and second gen devices like a 741 or LF351. But if even that is too high, a coupling cap is not evil. Just get a nice film type (polyester is a good performer for the price, polypropylene is nicer but more $). The Fc is just 1/(2 pi C R) where R is the input impedance to the next stage (the output Z of the op amp stage can be ignored due to its small size). You can also place the output stage within the op amp's feedback loop. Lots of power amplifiers have been designed this way. There is some discussion of this in the text (chapter 6).

[@deviantmultimedia9497]

Excellent advice, man. And your pdf rocks! Thank you for sharing it. When I said the thing about the datasheet telling me not to I was referring to this "This adjustment should be used only to null the offset of the operational amplifier, not to adjust system offset or offset produced by the signal source." from TI's OPAx134 datasheet. But now I don't think that applies to me since I do have the input AC coupled with a 10uF PP film cap. Well, there's a differential summing stage first (to receive a balanced audio signal) and then I have the cap on the output of that. That being said, you made a good point about modern op amps having negligible DC offset. Also, yea those trim pots are not cheap. I'm not opposed to overkill at all, but after adding up the extra cost and board space I can see this becoming "the retarded type of overkill and not the smart, beneficial type of overkill." As a wise accountant may have said. Thanks for the advice and the pdf book. Cheers!

[@dutchdykefinger]

i feel too mentally challenged to grasp the theorethical aspects of circuits as hobbyist programmer with a love for audio i'm all too comfortable with PI and the concept of calculating phase, dabbled into some fourier additive synth stuff too. i've struggled with some mathematical concepts before, and found that some teachers can make 1 comparison/analogue with something to explain the concept, and i can get what i struggled with for years, just 1 sentence that draws the parallel of things i've always never really grasped but just went by whatever some research was a common formula the maths aren't even that hard in most of electronics, and there's plenty of calculators out there filling in the 3rd using 2 values, so that's not really the problem sure, parallel non-equal value resistors is probably one of the more annoying ones to calculate and that's still mild. i'm just a bit freaked out by uncertainties i have with basic concepts of electronics, i've done my due diligence of spending days worth of hours on just researching those things online, i have a lot of the vocabulary, i know what components do for the most of it and know how to identify them, but i lack the experience to just trust the process lol and i already built a perfectly fine functioning LED COB growlight (~120-186 watts dim range) with 8 cobs, 2 parallel strands of 4 cobs in series, and the bitch dims just fine and has no problems of thermal runaway without fuses or big current limiting resistors on the parallel side, although that's mostly commercial pre-made products that are known to be quality and have good tolerances, also, they are driven far under their nominal specifications for that extra bit of efficacy, +a lot of efficacy by the fact they now have less hotspots, and have better light distribution on the short distance, by having more of them and driving them softer (they're 36 watts nominal, i drive them at ~24 watts running full tilt, so it's about 65% of their nominal drive at worst (or best, depending on your viewpoint) clawing me back around 8-12% of efficacy on output/watt, likely to be more on the high side of that seeing as the thermals are low and keeps the thermal runaway and added resistance in check. then againt cobs are very rugged and can just keep functioning fine at 80 degrees C without blowing out, they don't really cancel themselves until 120C i believe even, these aren't pedestrian crybaby gotta be the perfect constant current led modules, they're 12x12 (or 18x18) arrays of power leds in a single "flip-chip" module with far better thermal properties. there's a lot of efficiency to be had keeping them under 40, you can overdrive them 100% (twice teh nominal spec) and if you can handle the thermals to stay under 40-45C, the diminished efficiency is more than acceptable, high-bay lighting solutions for low-bay purposes, nice and overspecced, just the way i like it but the gains you get from counteracting the hard losses of inverse square law by driving them soft, then bringing it up closer is where the real profit is,. you're basically playing some divide by 2, then multiply by 4 in terms of photons that reach the target, the exponential nature of that relation with inverse square law goes off the charts on very small distances, long as you can guarantee the distribution. also reduces the amount of high quality heat dissipation solutions required, as heat distribution is handled by physical distribution of the cobs over an aluminium sheet (it's passively cooled) :) anyway i have this strong tendency to reinvent the wheel just to try and understand it, once i do, i understand things are based on concepts that haven't been improved since liek the 60s (like fft based on fouriers theorem that still stands since the late 1800s orso), and i'm better to take a pre-made solution by people that are far more well versed in the particular subject basically i have this insane imposter syndrome because i feel like a loser i don't grasp every little detail of it, maybe i'm being a bit hard on myself BUT I HAVE TO KNOW and it's maddening that wanting to know things only ends you up with MORE QUESTIONS, curiosity is as much a curse as it is a blessing lol here's the problem though, i don't have an oscilloscope to really vet my results so i'm now going to cut up some mini jack cable to set up one of my USB audio interfaces as a low frequency oscilloscope, it is audio i'm working with after all, and go from there :D

[@ElectronicswithProfessorFiore]

If it's of any consolation, I have always told my students "Every good question deserves a good answer, and every good answer leads to a new question." There are some free apps out there that turn your computer into a 'scope (via the audio hardware). Kind of limited, but it's a start.