Advertisement

Blog

Do We Need Any New Op-Amps?

I was thinking the other day about op-amps. I got a “New Product” email blast from a major analog IC supplier that discussed a few new op-amps. One was a “4MHz, 7nV/√Hz, Low Offset and Drift High Precision Amplifier.” Another was a “Low Power, 1 nV/√Hz, G≥10 Stable, Rail-to-Rail Output Amplifier” (single and dual versions available).

Packaged op-amps have been available since 1952 (if we go all the way back to the tube-based Philbrick K2-W). Monolithic op-amps date back to the μA709, vintage 1965. The '709 was a dog by most standards — external compensation, a tendency to latch up, and so forth — but it was a whole lot better than designing an op-amp from discrete transistors. It also used a lot less power than the K2-W, which needed a power supply to operate the tube filaments and positive and negative high-voltage supplies for the plate circuits.

In the 61 years since the K2-W, we've seen many enhancements:

  • Internal compensation, starting with the μA741
  • Process tricks like super-beta and circuit tricks like bias current compensation to reduce input currents
  • More process tricks like adding JFET (and then MOSFET) input stages for lower event input current
  • Laser, Zener zap, and chopper stabilization to lower offset voltage
  • Faster processes to get more and more bandwidth
  • Various tricks to lower distortion for high crest-factor signals
  • Rail-to-rail inputs and outputs
  • Lower noise, supply voltages, and power consumption

Today we can get singles, duals, quads, and octals, and a friendly vendor will put even more amps on a chip if you really want them. Of course, advances in packaging have shrunk op-amps to teeny weeny things, and prices have continued to drop.

There was a time when you had to choose between fast op-amps or precise op-amps. The newer generations offer both, so you don't have to choose in most applications. Some, of course, are optimized for either speed or accuracy.

And yet, with all these choices, new op-amps are still coming out. As of today, the TI website claims 1,381 different op-amps. Do we need them all? Do we really need any new ones?

I think that, if no new op-amps came out from TI, ADI, LTC, Maxim, or any other broadline analog IC house for the next few years, most designers would still be able to find something that met their needs. They would be able to get their circuits and systems designed and into production just fine.

But I'm curious if there are still any unmet needs out there. If you had your choice of one new op-amp that would solve your latest design problem, what would it be?

28 comments on “Do We Need Any New Op-Amps?

  1. jkvasan
    April 23, 2013

    Doug,

    Do I need any additional features in an opamp?

    How about gain programmable opamp? Already available.

    Voltage controlled amplifiers? Already available.

    Isolated amplifiers? Available.

    Any other function would turn the opamp into something else which is available readily such as DAC, ADC, etc.

    I guess all combinations have been thought out by the chip vendors.

  2. SunitaT
    April 23, 2013

    Opamps play major role in analog system design. It is used is every
    circuit to perform countless analog functions. With the diverse range
    of op-amps offered by tens of manufacturers, choosing the right one is
    a tricky task. Even the experienced engineers find the task of
    selecting the best op amp for a given application daunting.

  3. Vishal Prajapati
    April 23, 2013

    @JK, you are right. There are almost all possible combinations available from different silicon vendors. The next step could be integrated functions for different applicaitons like adder, subtractor, differentiator and integrator. There could be every parameter programmable in the next gen opamps.

     

    There is one more challange is to lower cost of special opamps like isolated opams. They are still expensive.

  4. jkvasan
    April 23, 2013

    @
    VP,

    I was just wondering if I would be considered greedy if I ask for optoisolated analog input for the opamps. Though the mind says it is a tall order, heart wishes it.

  5. Vishal Prajapati
    April 23, 2013

    @JK, I heard one of my senior in my previous company had implemented it using discrete components. He linearised the Opto isolator LED intensity with respect to input analog signal and on the other side linearized transister biasing made it possible to retrieve the actual analog signal successfully though with less accuraccy.

     

    But I think there is another way to do this is to generate the PWM signals according to input analog signal and apply it to Opto LED. On the other side LPF will recover the envelop of the original singal. This is just a thought not sure how accurate can it be?

  6. amrutah
    April 23, 2013

    @Doug, Thanks for the blog post and for a minute it made me think do we really need new op-amps.

         As the applications are increasing, op-amps needs to be either faster (than the existing ones), low power, incorporate improvement in  tricks like powermanagement, offset cnacellation, noise reduction, reliability, ESD handling.  Increase in the military, space and biomedical devices will call for 6sigma to 9sigma to reliability which would need op-amps.

      I think though the existing opamps are good and designers have developed many of the opamps, but relaibility and other such requirements will drive for more new opamps.

  7. jkvasan
    April 23, 2013

    @VP,

    I had used both methods.

    The analog method can only be used to, may be, detection of a threshold crossing. It may not work for the whole measurement range. However, if your range is small, you could centre it around the linear portion of the curve and get a neat output. You may have to subtract the offset voltage you may have provided to shift the lower end to the linear region.

    The second method works very well. I have used it for isolating the ECG pre-amplifier from the microcontroller.

  8. Bill_Jaffa
    April 23, 2013

    Doug, we'll see new op amps from vendors until one of them comes out with the “ideal” op amp: perfect specs, zero power dissipation, and zero cost!

    Seriously, you make a very good point. Today's op amps are just so good, and so reasonably priced and easy to use, that almost all users are OK enough. Yes, there may be a few on-the-edge applications that really need the better performance or dissipation—but not enough sockets out there to justify the design/fab/test costs for the vendor.

    Plus, in many cases, the user can get better performance specs by calibrating and storing the cal factors in the system's smart transducer, or the analog front end, or even the processor—that can take a lot of the error out of the signal chain, including tempco related. So there is almost no need for a device which is inherently better.

    And users are so confused about which op amp to get, there are so many tradeoffs, that adding one more to the mix may not even be noticed by them, unless it solves a really significant problem they have.

  9. D Feucht
    April 23, 2013

    Bill, your last point is well-taken in that I occasionally have an op-amp application where I cannot find a commercial part with the right mix of features. High speed and picoampere input bias current for fast sample-and-holds with a 36 V supply range is one example.

    What I think we tend to do is to find solutions to problems given the available parts, even though some of them are suboptimal. I have my list of about 20 preferred op-amps (kept in stock for prototyping, of course) from which I draw for new designs. However, I am glad that the available commercial offering is much larger, even if it does not always contain the part I want, if not need.

  10. Andy Pienkowski
    April 24, 2013

    I'm interested in new Op-Amps always.  I'm always looking for better devices to use to help me debug RF systems for which I am continually running into slew rate and current drive limitation capability at 13MHz – 30MHz with what is available today.

    I would love to have something which allows me to make a small, moderate power amplifiers quickly at 10s of MHz frequencies to get sizable currents through inductors without the limitations of having to have a 50R output – synonymous with commercial RF blocks.  Since I am not much of a power amplifier designer myself at the moment, getting something “oven ready” from seasoned designers at places like TI or Maxim which, together with a couple of transistors, could be knocked up quickly to do measurements in a lab would be great.

    This isn't a commercial deign application for high volume manufacture but it is fun.

    Andy Piuenkowski

  11. Flagstaff Rich
    April 24, 2013

    Frankly, this is the wrong question.  Every engineering discipline you can name has achieved plateaus that were “good enough” until they weren't any more.  Tall ships ruled the seas and built empires until steam power and iron hulls changed the game.  Wood-and-canvas biplanes ruled the sky until all-metal monoplanes took over, and electronics is rich with examples.

    This ideal op-amp business is another plateau.  We get to decide what parts prevail but manufacturers need the freedom to pursue the Next Big Thing however they see fit.  Most will continues the quest for the perfect Buggy Whip, but some smart manufacturer is going to figure out what really comes next and throw us a rope to the next paradigm.  The exciting part is that together with academics we can help point the way.  3-D opto-quantum silicon devices?  Mitochondrial processors with DNA memory?  I can't wait to see!

    I had a EE Prof at Purdue who started as a junior engineer at Westinghouse in the 1920's.  His first job was as part of a team perfecting the copper oxide rectifier for battery chargers.  I would love to get his take on this!

     

    Rich Oliver, EDS

    Lowell Observatory

  12. Jonathan Allen
    April 24, 2013

    For those of us who do prototypes and one-of-a-kind circuits it would be very helpful  if newer op-amps were available in DIP packages, even if they cost extra. 

  13. treefarmer
    April 24, 2013

    @Johathan, I found a solution for prototyping ICs that only come in SOIC packages. I use Schmartboard products that break out the pins to 0.1″ headers. They boards come pre-soldered and the headers are provided. I plug the circuit into my solderless breadboard. The headers are long enough to place 1/8 W resistors and small capacitors under the board. Check it out.

    Jerry Cogswell

  14. Jonathan Allen
    April 24, 2013

    Thank you very much  I will have to check that out. 

  15. Bob @ JVD Inc.
    April 24, 2013

    The IC Master lists over 58,000 Op Amps available from which to select the one that best fits your needs. You may not need a new Op Amp, but you may need a solution that includes a low noise version, a high gain version, a differential version and maybe some power management on the same chip. That part won't be among your 58,000 choices. In the Analog ASIC world, “New” Op Amp solutions are being created every day and that will continue long after we're all toes-up.

  16. Bill_Jaffa
    April 24, 2013

    Hmmmm…I wonder how many microcontrollers are listed? I'll bet it is almost as many, when you look at the major families, and their variants. So choosing the right one is a problem there, too–except with a micro, you choose performance and features, and only as much memory, I/O etc as you need. So some of those micros are the same except, for example, for memory size and type.

  17. Davidled
    April 24, 2013

    Selecting microcontroller is one of biggest challenge tasks. Unfortunately, decision is made by not engineer, but engineer manger and purchasing manger. If product is underestimated, they complain the engineer. Engineer just designs the selected microcontroller in the condition of any limited memory, size and speed with the limited development tool.

  18. Davidled
    April 24, 2013

    Please check TI website. http://www.ti.com.

    They have a few DIP Package of Opamps.

  19. Vishal Prajapati
    April 28, 2013

    @JK, Thanks for sharing your experience. When I will need to implement, I will try the second method first. That seems easier but a bit complex than designing a full analog design.

  20. Hughston
    April 29, 2013

    You would think that everything the designer could want is out there but not always. For example, there are lots of good instrumentation amps but then a few years ago Analog Devices came up with a very high temperature version IA in a small package for oil drilling equipment that was optimized for low impedance sensors.

    I need an op amp that drives a lot of current close to the rail. I can get DSL drivers from Linear Tech that do that and I use one for that purpose but I want one in an 8 pin SOIC package that can drive good peak current close to the rail.  The parts on the market are not clearly speced and I have to try various parts out. I don't need to dissipate much power; I just want to deliver brief peak drive.

    Most op amps are incremental improvements over older parts. Some of them are new versions of the old part in a new process. Process changes and improvements drive  a lot of the part obsolescence.

    Generic improvements like lower power, lower noise, lower offset, lower operating voltage, are always welcome. Stability is something you take for granted, but don't always get with your circut configuration. Another thing that seems to be new in the last few years is to offer parts with RF immunity. Some parts had it in the past, but probably by accident.

  21. Gina.Hann
    April 29, 2013

    In response to the question of “why do we need new op amps?”: While we certainly have our fair share of amps to choose from and a lot more that we can design (one of our engineers once estimated that there are over 10.3E8  possible amplifiers possible for us to design – not kidding!) the reason we design new ones has a lot to do with realizing some of the incremental changes needed in both performance and cost over time and process capability. Amplifier performance requirements change not just because of new process technology or new topologies (achieving zero-drift without the significant added errors of chopping frequencies in the past was unthinkable!) but also because the systems that you design using op amps change. Smaller, faster, more complex and even simpler all mean a change in the amplifier performance needed to meet the total system requirement. That 1nV/rtHz op amp for example could be the determining factor in whether or not your 16 or 18 bit SAR converter is able to realize it's 2LSB INL, and consequently whether or not that motor encoder you are designing will be accurate enough to maintain a smooth transition between steps. So, while we do have a lot of amplifiers to choose from, we also have a long way to go – we will keep making amplifiers as long as you keep engineering new stuff!

     

  22. Brad Albing
    April 29, 2013

    So, an early pioneer on solid state devices. Very cool. Would have been nice to get the perspective of someone from way back in the day.

  23. Brad Albing
    April 29, 2013

    You make some good points as to why we may still need more new op-amps. Thanks.

  24. Brad Albing
    April 29, 2013

    Sadly, so few would be used (i.e., purchased by just a few people to prototype a ckt. – or hobbyists) that no mfr will make them because they would need to charge so much (to recover their NRE) that you could not afford them. How much? Think Mil-grade or rad-hard devices.

  25. Brad Albing
    April 29, 2013

    Thanks Gina. Some good info that helps clarify things a bit.

  26. Brad Albing
    April 30, 2013

    I'm surprised that hasn't happened yet. I was discussing in another comment about how companies like Rockwell would love a part like that. Seems like the volume (annual usage) is there.

  27. Brad Albing
    April 30, 2013

    Andy – I'm sure you already know how this works. The apps guys or the FAEs would readily help if they knew there was an order to be had later for a bunch of parts. Otherwise, unless you just happen to pique the interest of the apps guy, not much will happen.

    Still, this should be a straightforward design – wide BW op-amp plus a couple of power FETs. I assume you've looked thru some apps notes already. Perhaps we will get a suggested circuit here.

  28. Brad Albing
    April 30, 2013

    I've used those too. pricey but effective.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.