74

Simple enough question. Why not use a 741 op-amp in a target circuit or anyone's target circuit? What are the reasons not to use it? What might be the reasons to still choose this part?

Marcus Müller
  • 88,280
  • 5
  • 131
  • 237
Andy aka
  • 434,556
  • 28
  • 351
  • 777

8 Answers8

93

There are many good reasons not to use the 1968-vintage LM741: -

  • Minimum recommended power supply rails are +/- 10 volts
    • Modern op-amps have power supplies that can be as low as 1.8 volts.
  • Input voltage range is typically from -Vs + 2 volt to +Vs - 2 volt
    • Modern op-amps can be chosen that are rail-to-rail
  • Input offset voltage is typically 1 mV (5 mV maximum)
    • Modern op-amps can easily be as low as a few micro volts and have low drift.
  • Input offset current is typically 20 nA (200 nA maximum)
    • Modern op-amps are commonly available that are less than 100 pA
  • Input bias current is typically 80 nA (500 nA maximum)
    • Modern op-amps are commonly less than 1 nA
  • Input resistance is typically 2 MΩ (300 kΩ minimum)
    • Modern input resistance starts at hundreds of MΩ
  • Typical output voltage swing is -Vs + 1 volt to +Vs - 1 volt
    • Many cheap rail-to-rail op-amps get to their supplies within a few mV
  • Guaranteed output voltage swing is -Vs + 3 volt to +Vs - 3 volt
  • Supply current is typically 1.7 mA (2.8 mA maximum)
    • Modern op-amps with this current consumption are ten times faster and better in many other ways too.
  • Noise is 60 nV/sqrt(Hz) for LM348 (quad version of 741)
  • GBWP is 1 MHz with a slew rate of 0.5 V/us

The LM741A is slightly better but still a dinosaur in most areas.

Things of importance that the 741 data sheet does not appear to list (and that may depend on the age and manufacturer): -

  • Input offset voltage drift versus temperature
  • Input offset current drift versus temperature
  • Common mode rejection ratio versus frequency
  • Output resistance (closed or open loop)
  • Phase margin
  • likeliness of latchup (and gain reversal)

I can't think of any valid reasons to use the 741 other than "that's all I will ever have or own". Common reasons why they are still used in actual devices appear to be: -

  • Someone had a design that they didn't want to change from the 70s
  • Someone had millions of them lying around and wanted to put them to use
  • Someone actually determined that all the parameters are fine for their design, and at that moment the 741 was the cheapest to acquire and in millions of units it saved a few thousand dollars in total.

I've been an electronics designer since 1980 and I have never used or specified a 741 in any design I've been associated with. Maybe I'm missing out on something?

PlasmaHH
  • 6,498
  • 5
  • 38
  • 49
Andy aka
  • 434,556
  • 28
  • 351
  • 777
  • Comments are not for extended discussion; this conversation has been [moved to chat](http://chat.stackexchange.com/rooms/58573/discussion-on-answer-by-andy-aka-reasons-not-to-use-a-741-op-amp). – Dave Tweed May 11 '17 at 12:29
  • Please name 2 or 3 of these better alternative for 741 – Gabriel May 06 '19 at 12:58
  • 10
    @Rigel make a list of random opamps chosen from your favourite supplier and blind folded, put a pin in the list. The opamp nearest the pin hole wins and will almost certainly be better than the 741. In fact, all of them will be better in one or another respect. – Andy aka May 06 '19 at 13:36
  • Actually, I just did it :) - with Arduino in mind.... https://electronics.stackexchange.com/questions/316716/what-is-a-good-alternative-for-the-ua741-opamp-in-hobby-projects/437171#437171 – Gabriel May 06 '19 at 13:37
  • Just mentioning that Practical Wireless magazine ran an article on building a hi-fi amplifier, the "Texan", using 741s (and 748s which IIRC are just two 741s in one case) back in the [May 1972 issue](https://www.americanradiohistory.com/Archive-Practical/Wireless/70s/PW-1972-05.pdf), pg 48. There was an update in the July 1975 issue for better shielding and a toroidal power transformer, but still using the same 741 op-amp. My Dad built it as a kit, and it was our "family hi-fi" for many years. It sounded great to me when I was about 6 years old :) – Reversed Engineer Jul 20 '19 at 14:50
  • @ReversedEngineer did you suffer from tinnitus back then lol – Andy aka Jul 20 '19 at 19:42
  • :) It actually sounded pretty great if I remember correctly. Must ask my Dad what happened to that amplifier. Maybe it's still in a cupboard somewhere and can be fired up again for a more objective test. – Reversed Engineer Jul 25 '19 at 09:20
  • 1
    @ReversedEngineer in all honesty, a bad, but well-compensated opamp that you don't try to use in a high-gain operation can sound OK to good. But the 741 is really the opamp worst case, and you need to be *very* generous with supply voltages, headroom, limit bandwidth a lot ... and that bandwidth limiting together with a nonlinear compression might have given music a "warm" quality that doesn't make for good reproduction but often "feels like home". Anyway, nobody would consider the 741 a hifi component, really. – Marcus Müller Sep 18 '20 at 15:13
  • @MarcusMüller "bandwidth limiting"? The published frequency repose of the Texan amplifier I mentioned above, is "Frequency response (16 + 16 watts into 8Ω): -1dB 7Hz to 22kHz; -3dB less than 5Hz to 35kHz", as measured using the following test equipment: Radford Low Distortion Oscillator, Hewlett Packard 3590A/3593A Wave Analyser; Hewlett Packard True R.M.S. Voltmeter; Hewlett Packard 3300A/3305A Function Generator. Pacific Measurements 1010 A.C. Digital Voltmeter; and Hewlett Packard 181A Oscilloscope. Any "warm quality": in the sound was due to the original recording, not due to the 741! – Reversed Engineer Mar 18 '21 at 07:09
  • @MarcusMüller "very generous with supply voltages: - fair enough, they are using +/- 15 Volts to the 741s – Reversed Engineer Mar 18 '21 at 07:13
  • "well-compensated opamp"? Yes! A whole paragraph in the article (mentioned above) is dedication to the design of the compensation. "nonlinear compression"? The power supply seems pretty solid to me, so no clipping if that's what you mean? Harmonic distortion is quoted at "15 watts into 15Ω: 0.05% at 1kHz; 20 watts into 8Ω: 0.09% at 1kHz", so no "nonlinear compression". Seems well within the threshold of what accepted as "hi-fi". – Reversed Engineer Mar 18 '21 at 07:24
  • Huh, since when is 0.1% into 8 ohm a HiFi spec? – mmmm Jul 19 '21 at 02:46
  • @mmmm who was that comment addressed to? – Andy aka Jul 19 '21 at 07:03
  • Andy, that was addressed to @reversedengineer – mmmm Jul 19 '21 at 07:11
  • @mmmm It's [well](https://www.audioholics.com/loudspeaker-design/audibility-of-distortion-at-bass/total-harmonic-distortion-thd) [known](https://www.diyaudio.com/forums/chip-amps/164484-thd-considered-hifi.html) in the audiophile world that [THD is not a good indicator](https://www.avforums.com/threads/how-much-thd-total-harmonic-distortion-is-acceptable.943658/) of hi-fi at all! It turns out that most loudspeakers speakers [add about 1.5% THD](https://forums.audioholics.com/forums/threads/what-is-an-acceptable-level-of-harmonic-distortion.63126) anyways. – Reversed Engineer Jul 20 '21 at 08:32
  • Distortion figures are mostly a marketing gimmick, like shampoo guaranteed to contain less than 0.1% COVID-19 by mass. Negative feedback was added to amplifiers to make distortion figures look good, but many valve amplifiers don't use have any negative feedback to control the distortion figure at all, and yet are considered to be among the best sounding in the world. For example, this [modern valve amp](https://www.rogers-hifi.uk/rogers-e20a-ii-integrated-valve-amplifier#:~:E20a/ii%20Specifications) advertises a distortion figure of 3%, and is definitely "hi-fi". – Reversed Engineer Jul 20 '21 at 08:55
  • Hm we might have pretty different ideas of what sounds good, and if something is "well known in the audiophile world", then that's a bit like "biology facts well known in the vaccination skeptic movement", if you catch my drift. (That was an operating-point joke referring to tubes.) – mmmm Jul 20 '21 at 23:08
  • I once used one (TI's μA741 though) because it was the only thing available to me in time capable of weathering (well, just barely) the 35V supply from an older device. – Sixtyfive Sep 11 '21 at 20:57
  • 1
    I think the uA741 was made by Fairchild not TI but, who cares cause the device is crappy whoever made it. – Andy aka Sep 11 '21 at 23:13
40

If there weren't loads of textbooks dating back decades that used the 741 as an example, I'd be surprised if many people knew of it now. It'd have passed into history, like the OC71 which was the BC108 of the mid/late 60s.

One of the reasons I think it persisted for so long beyond the mid-80s, when it was well superseded, was because of its many bad characteristics. It's a good example for teaching students the characteristics to be concerned about, those weaknesses that highlight the inner workings of the op-amp. It has the significantly-large input offset voltages and input currents, needs the offset null, shows the bandwidth/gain product so vividly. Few/no op-amps of 20 years past have brandished so many of these so prominently. In modern devices, many of these characteristics are negligible or practically non-existent.

I've been designing for 30 years, starting with mid-volume equipment in the mid-80s. In my experience, the 741 wasn't anyone's choice then or since - there were always much better and cheaper parts to use. I imagine maybe I've worked on a circuit with one on in that time but I genuinely can't recall it.

TonyM
  • 21,742
  • 4
  • 39
  • 62
  • 7
    I agree - I've been employed as an EE since 1980 and never once have I chosen a 741. Chosen plenty of LM324s back then and TL08x op-amps but never the 741. – Andy aka May 10 '17 at 12:02
  • I guess a 709 (pretty obscure part these days, and idiosyncratic by today's standards) would be an OC71, a 741 a BC108 (pretty obsolete as a literal part number but still exists in very similar forms)... – rackandboneman May 10 '17 at 14:28
  • 11
    It is very common to see 741s listed at hobbyist vendors like adafruit or sparkfun or radioshack. It seems that they are one of the default "jellybean" components that hobbyists are expected to use and learn simply because of the large number of example projects that one can find on the intertubes. These vendors are somehow able to sell them at up to 8x cost. I suspect that most of the people using 741s now are hobbyists, and they simply don't know any other op-amp (since they possibly won't appreciate the complex subtleties involved in picking a better one). –  May 10 '17 at 16:09
34

Another reason not to use a 741 is that under certain conditions it can go into a latch-up state, where the output saturates and sticks to one of the supply rails until it is powered down. I can't find a reference specifically about the 741, but this page describes something similar: https://www.allaboutcircuits.com/textbook/semiconductors/chpt-8/op-amp-practical-considerations/ (search for the word "latch-up" on the page).

Just out of school, I built a thermostat for our fish-tank, using a 741 with a small amount of positive feedback to act as a comparator to switch the tank heater with a relay. Twice the 741 latched up and killed all our fish.

  • 2
    [This page](https://electronics.stackexchange.com/questions/62084/reasons-for-741-op-amp-latching-up) also mentions the 741 latching up – Reversed Engineer May 10 '17 at 15:26
  • 2
    The latch-up in that kind of op amps was frequently caused by the inversion of the sign of the gain of the first differential stage when the common mode voltage exceeded the specified limits: the feedback turns from negative to positive and the amplifier turns into a Schmitt trigger. – Massimo Ortolano May 11 '17 at 20:37
  • 10
    Now in the box office: "741: The Fish Killer" –  May 12 '17 at 15:00
  • 2
    :) Yes - poor fish. It happened only twice over several years in the 80s, but I still felt sorry for the fish, and eventually plugged in a more modern op-amp (once I was working as an engineer and could afford something i didn't find at the bottom of my Dad's drawers :) – Reversed Engineer May 12 '17 at 15:14
  • 2
    You didn't add a "butter thermofuse" which melts at 25°C (or similar). – Janka May 15 '17 at 12:13
  • 1
    This is called *op-amp phase reversal*. If you google for it, you'll get mostly relevant results. The phase reversal is something people who do analog control theory labs learn about rather quickly :) – Kuba hasn't forgotten Monica Jun 08 '22 at 20:13
  • @Kubahasn'tforgottenMonica Thanks for this - I Googled it and got a lot of relevant info. – Reversed Engineer Jun 10 '22 at 09:14
22

No one has mentioned the social signalling aspect in that a circuit using a 741 will probably work with most any opamp, at least for a relaxed definition of "work". So if you need higher freq or different voltage limitations given that the original design was very vanilla, substituting a somewhat different amp will almost certainly work. This can be extremely educational, if you're trying to build an ultra low noise low pass input then build a boring low pass input with a 741 to wrap your brain around the "low pass" part of the project, then figure out the modifications needed (if any...) to substitute in a low noise amp to divide and conqueror a two part project like that. Lets say you have a very special ultra high input impedance circuit, that is unfortunately static sensitive, you could debug at least portions of the design with a mostly indestructible and very cheap 741, then swap in the ultra high impedance (probably expensive?) op amp at the very end. One reason to still use 741s is noobs and experts think its very predictable. Folks with experience in the middle tend not to like the 741 because they've probably latched a few up or found a way for a project to fail due to lack of rail to rail output. But they're mostly trustworthy and better understood than the rest of the project, most of the time.

Be wary of substitution if the design spec'd some obscure ultra high freq part or an ultra high output current or ultra low noise or chopper stabilized. If you slap in a 741 or possibly any other amp than the specified one, the design is almost certain not going to work.

The safest way to professionally figure out the above paragraphs with an actual design is to figure out the Venn diagram of the two components specs, then figure out how those specs relates to your individual project, then decide if the performance difference is irrelevant or a bit deal. This works as an engineering strategy for almost any component substitution problem you'll ever run into.

Vince Mulhollon
  • 382
  • 2
  • 2
  • 8
    That's an *excellent* point about the 741 being used as an "insert your favorite op amp here" placeholder in a design. – ThreePhaseEel May 10 '17 at 23:59
  • This approach works well if you have a "classical" analog design where you got +/- 15V supplies and such. Even using slightly more modern parts like LM324 requires quite a bit of thought when the supply goes towards the minimum recommended operating voltage. Here's an example of a design that goes to some length to work around the limitations of LM358 used in low voltage regime - down to 4V: https://electronics.stackexchange.com/a/609484/10810 – Kuba hasn't forgotten Monica Jun 08 '22 at 23:28
11

What are the reasons not to use it?

High supply voltage requirement, high bias current, not rail-to-rail operation, limited current drive, slow, etc.

What might be the reasons to still choose this part?

For many applications, it gets the job done, respectable DC performance, fast enough, wide availability, cheap, limited current drive, slow, etc.

High performance isn't always needed, and sometimes low performance has its advantages.

Peter Mortensen
  • 1,676
  • 3
  • 17
  • 23
dannyf
  • 4,272
  • 1
  • 7
  • 9
  • 2
    "sometimes, low performance has its advantages" goes against the grain. It's like saying "this circuit won't work if the transistor's ß is too high". – Finbarr May 10 '17 at 10:41
  • It seems fairly expensive these days – Scott Seidman May 10 '17 at 10:54
  • 2
    @Finbarr yes and no. If you're driving some low frequency signal that low bandwidth is rather good at ignoring logic switching noise. – Trevor_G May 10 '17 at 11:03
  • 3
    Had a counter circuit where if there was a TI IC in place, it would count each input pulse as 2 to 3 pulses, Any other brand would count 1. Call to TI -"We don't release them unless they're 3 times over minspec!" Bingo! The narrow pulse caused the PC track to act like a transmission line and the pulse was reflected back and forth - the TI chip counted them, but others didn't. In the early PC days, mass storage meant tape recorders. Cheap and cheerful - good. Professional quality tried too hard to make musical sense of the data and failed. Sometimes, not-so-good is better... – Magoo May 10 '17 at 12:00
  • 15
    A well-known MIT analog circuits professor had a saying: "Those who ask for more bandwidth than they need deserve what they get." Words to live by. – WhatRoughBeast May 10 '17 at 12:12
  • 2
    But those that depend upon a "typical" figure instead of a hard spec to get that low bandwidth instead of designing for it also deserve what they get. – Scott Seidman May 10 '17 at 12:32
  • 7
    @Finbarr There is a classic low parts count two transistor flasher cct that relies on the beta product of the two transistors being below what is needed to allow a high resistance timing resistor to latch the circuit on. Use ancient low beta parts and it works. Use transistors with Beta of say 300 each and it latches. Ask me how I know :-). – Russell McMahon May 10 '17 at 12:59
  • 1
    I'm not disputing that it happens, @RussellMcMahon - just that it's not good design! – Finbarr May 10 '17 at 13:24
  • 4
    "this circuit won't work if the transistor's ß is too high" and there are plenty of examples of "this circuit" :) – rackandboneman May 10 '17 at 14:26
  • 7
    High speed and high slew rate opamps are downright evil from the perspective of a beginner and layout engineers. They are more like error amplifiers that show off ones ignorance to the world, proudly so. I think there is a lot of wisdom in introducing beginners to an easy opamp like the 741 that tolerates design errors. – dannyf May 10 '17 at 20:57
  • 2
    Using a 741 *is* a design error! – Ian Bland May 11 '17 at 08:31
  • 4
    @IanBland And the 741 tolerates it! – Christian May 11 '17 at 09:44
  • 1
    @Finbarr We are not really in disagreement - I'm really just pointing out that for the circuit in question, to "design" it properly transistor betas MUST be allowed for and, if not, a point is reached with increasing beta product where it latches on. Examples are [here](https://s-media-cache-ak0.pinimg.com/736x/bb/6d/c6/bb6dc650818a42618d0ec848cd7d9c21.jpg) and [here](http://josepino.com/articles/transistor/flasher_other.gif) and [here](https://www.google.co.nz/search?q=2+transistor+flasher&num=20&tbm=isch&tbo=u&source=univ&sa=X&ved=0ahUKEwiw0_TE6-fTAhVNO7wKHcXoBJoQsAQIJA&biw=1344&bih=697) – Russell McMahon May 11 '17 at 12:35
  • 4
    You shouldn't usually use ultra high speed opamps if you don't need the speed but there are plenty of cheap, slow opamps that perform better than the 741. Also the 741 is really slow. If you get into trouble with a modern ~20 MHz GBW opamp the solution is not to get a 741 but learn what you are doing wrong. – Evan May 14 '17 at 22:08
  • 1
    The pulse-rejection of UA741, in the minus VDD power pin, was approximately ZERO. Taught me to consider the design of VDD tree, caps plus resistors plus more caps and maybe more resistors and caps, for high ENOB, high SNR circuits. – analogsystemsrf May 28 '17 at 22:31
  • 1
    The 741 tends to tolerate poor layout, and for this reason is popular for solderless breadboard prototyping (just try this with a 715-type!). Provided you know the drawbacks, the low bandwidth of a 741 can be an advantage. – user131342 Feb 03 '18 at 10:54
3

Before we harangue too fiercely newbies who specify a 741 opamp in their questions, it's become apparent from reading many of the 'homework' questions that it's the opamp used in their courses, whether highschool or graduate level.

I can only imagine that much of this teaching material was updated when opamps arrived on the scene, and has not been updated since.

Neil_UK
  • 158,152
  • 3
  • 173
  • 387
  • 2
    The 741 is handy for didactic purposes because it's such a good example of just how badly nonideal an op amp can be. Whether this is the *intended* purpose of using it in education, I can't say. – Hearth Dec 08 '20 at 06:11
  • 1
    @Hearth If it was being held up as bad, then surely they would come here with the better alternatives they'd been shown. No, I fear that it's presented in a 'typical opamp circuit'. When you have +/- 15 V supplies in the lab and only teaching specs to meet, it's not so much of an issue. Besides, what would you use as an alternative? My favourite for one particular job was OP275, but that choice did depend on specs. Do you go R2R, or fast, or low power, or low noise. The solution may be to use a 'black box' opamp with no part number, but it's not their problem that needs solving. – Neil_UK Dec 08 '20 at 06:23
  • 1
    Lol yeah. Same with 8085, 8086 etc. They are still in the curriculum (atleast here in India). – Mitu Raj Dec 08 '20 at 09:04
3

Use the 741? It depends. If the 741 meets your requirements, and it's your cheapest/most available multi-sourced solution, by all means use it. A legacy design that uses it can keep using it if it didn't have any identified problems in the field.

That said...

New designs? In this year 2022, the 741's well-known limitations will disqualify it from most applications: newer systems tend to be lower voltage and lower power, making the 741 is a poor choice. To meet these needs as well as others there are excellent, easy-to-buy choices that do things the 741 could never dream of. Neat tricks like 'over the top' high-side current sensing, working at 3.3V with a rail-to-rail output swing, having uW standby power, or having nearly-zero input bias current.

In fact it will often be the case that the advantages a modern op-amp brings to your design can reduce your overall system cost, even if the op-amp itself is more expensive than the 741.

The downside? Multiple sourcing: choosing a specific op-amp requires some research and qualification work. That said, 741s from different vendors can also vary in characteristics, so you'd still need to qualify each vendors' part for your Approved Vendor List (AVL). So really, you're not saving any legwork by choosing a 'multiple source' 741 vs. choosing several sources of modern parts with similar characteristics.

Existing designs? Designs that show the 741 as the op-amp may be the designer's way of saying "hey, my design is so tolerant that it can use this old POS op-amp". (Is that really true? Hard to know.) On the other hand, if the designer is counting on a specific characteristic of the 741 (either intentionally or accidentally), beware. Changing the op-amp could cause unexpected behavior.

Either way, when upgrading the 741 op-amp to a more modern type, carefully analyze the circuit.

  • Check with the designer if possible to gauge (1) their intent in choosing the 741 and (2) if they knew what they were doing.
  • Simulate it using a good sim like LTSpice, comparing the 741 and its proposed upgrade/replacement.
  • Make a prototype with the upgraded part and test the snot out of it to fully qualify the new device.

This is standard sustaining-engineering stuff, really.

hacktastical
  • 49,832
  • 2
  • 47
  • 138
0

There are very few designs that rely on things we used to call bad in the 741. Low GBP is used in some filters, for example.

TonyM
  • 21,742
  • 4
  • 39
  • 62
atomd
  • 21
  • 3
  • 4
    Then it is a poor filter. – JRE Apr 06 '19 at 10:17
  • Plus, if you really wanted an opamp with intentionally limited bandwidth for noise, power or robustness reasons: dozens of cheap better opamps to choose from. – mmmm Jul 19 '21 at 02:51