Need op amp for constant current source

Discussion in 'The Projects Forum' started by someonesdad, Feb 11, 2011.

  1. someonesdad

    Thread Starter Senior Member

    Jul 7, 2009
    1,585
    141
    Attached is an older application circuit from Maxim using obsolete parts. I'd like to build a few of these (I'll give some away to my relatives) current sources. I have substitute parts that would work (I'll use an LM385 for the voltage reference), but I don't know a good substitute for the op amp.

    Can you experienced folks recommend a good micropower op amp that will work well in this application that works from a single rail supply? I'd also like it to be cheap as hobby dollars are in short supply at the moment. Oh, I'd also want it in an 8 pin DIP package so these old eyes can see the dang thing. :p

    Thanks!
     
  2. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Well, about the cheapest you'll find ($0.40 USD or less) in an 8-pin DIP is an LM358 dual opamp. They're available just about everywhere, fairly low bandwidth @ 700kHz, fairly low power, but they have a pretty decent Vcc range (~32v for single supply operation) and the inputs/outputs go within millivolts of the negative rail; Vout is about Vcc-2v with a light load.

    Many of the newer opamps are just available in SMT packages, so kind of tough to use with breadboarding - great if you're making PCB's (no drilling necessary - yay!) but a bit of a challenge if your vision just isn't what it used to be.

    Also, much of the RRIO (rail-to-rail input/output) opamp crowd has a limited Vcc range.
     
  3. Wendy

    Moderator

    Mar 24, 2008
    20,766
    2,536
    That is very similar to electronic load I've been meaning to build forever.

    Electronic Loads

    I'm not sure how much can transfer over, but it is worth looking at.
     
  4. someonesdad

    Thread Starter Senior Member

    Jul 7, 2009
    1,585
    141
    Thanks, SgtWookie, that's just what I needed to know. My Someheimer's had kicked in and I had forgotten about those LM358's (I have some LM3904's somewhere and they're nearly the same chip; I think I have a few left over from another project). I'm moving all my junk to another room over the next few days, so I'll likely run across the stash while organizing.

    Boy, you're right about the SMT stuff. I found a very nice LTC op amp and voltage reference all in one and it was only a couple of bucks. But it's only in surface mount and my soldering iron's tip is probably bigger than the whole IC. :p

    I think even a sub-audio bandwidth would be OK for this application, as I plan to just use this in a little box with my DMM to measure low resistances. I only plan to make it a 1 A constant current source, so with my Fluke DMM that reads to 0.1 mV, it will let me resolve resistance to 0.1 mΩ. Nice when you're away from the bench instruments.
     
  5. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Yeah, those LM358's are cheap, plentiful, and meet your 8-pin DIP spec. I think you meant LM2904 instead of LM3904; the LM2904 is the automotive spec version; Vcc's de-rated for those and the input offset spec is a bit worse (wider temp range accounts for that).

    Sure, you might find something better, but newer opamps tend to have a pretty short marketing cycle, and the good old DIP package seems to be going the way of the dinosaurs. Board space, drilling, and DIP insertion is a lot more expensive to manufacture than the SMT/SMD stuff, which is why they're going away.
     
  6. someonesdad

    Thread Starter Senior Member

    Jul 7, 2009
    1,585
    141
    An update to this project. I built the circuit shown in the first post on a prototyping board and it worked great, just as advertised. Interestingly, since I had to sort through lots of stuff in the move to another room, I found a little box from Maxim that had two each of the required parts in it (but those parts are obsolete).

    I also have a beautiful old 10 turn Beckman 500 kΩ pot (40 mm in diameter) and it's very linear, as I have a large turns-counting dial on it and have had it in a box as an adjustable resistor for years. This thing will be perfect to turn the fixed current source into an adjustable current source. Changing the shunt resistor to 10 kΩ let me adjust the current easily over the 0 to 10 μA range full scale. I have a 7 position rotary switch that will let me have full scale currents from 1 μA to 1 A by decades.

    This will give me a precision current source, something I've wanted since I used an HP precision current source back in the 1980's. What's nice is that I have all the parts for it, including a new die cast aluminum box. It will use some internal D batteries to supply the current or can be switched to use an external source. The IRF540 is rated up to 100 volts, so there's lots of room for compliance. This will be a nice tool for the lab bench when it's done! Of course, I'll show pictures and brag about it when it's done.

    You'll notice, of course, the feeping creaturism -- it originally started out as a constant 1 A current source... :p
     
  7. steveb

    Senior Member

    Jul 3, 2008
    2,433
    469
    Nice work!


    So, the burning question now becomes, ... what is the process for becoming one of your relatives?
     
  8. someonesdad

    Thread Starter Senior Member

    Jul 7, 2009
    1,585
    141
    Well, first put $10000 in small, unmarked bills into a paper bag... :p

    Or, my youngest daughter is still unmarried...
     
  9. someonesdad

    Thread Starter Senior Member

    Jul 7, 2009
    1,585
    141
    Alas, Mother Nature always has a way of stomping on your "Oh, it's finished" statements. :p

    I characterized the circuit today and was a bit chagrined to find that there's a consistent drop of 2.5% in the output current over time. It almost certainly is due to things heating up, as it stabilizes after about 5 minutes. This means it would still work OK as Maxim originally intended in their application note 106, but not as a precision current source, which I'd expect to be stable to within a few tenths of a percent. Thus, if I want the precision current source, it's back to the drawing board. Maybe some of you EEs would have a good suggestion of an easy way to fix things...
     
  10. Adjuster

    Well-Known Member

    Dec 26, 2010
    2,147
    300
    To sample your output current accurately you would ideally want to use four - terminal resistors, sensed differentially.

    Using the multiple tap output arrangement of the schematic you have posted (which uses simple resistors) would require the circuit layout to be very well arranged to avoid error and drift introduced by copper trace resistance, especially for the 1A range. I can imagine that the thin tracks on a prototype board could give a lot of trouble as they warmed up.

    Of course, your problem may be due to something completely different, but this is something that you might want to check.

     
  11. someonesdad

    Thread Starter Senior Member

    Jul 7, 2009
    1,585
    141
    Good points, Adjuster, but the prototype circuit was built on one of those plastic prototyping blocks and I used about 20 gauge wire for the 1 A circuit part. I disassembled the circuit already and I'm not planning on building it again, at least on the prototyping board I used. If I do build it again, it will be on perf board with point-to-point wiring, as I don't make PC boards.

    I assume a 4 terminal resistor is just one that has a convenience Kelvin connection, right? To take advantage of the differential measurement (a good idea) means adding more components and I'm already pressed for room.

    I have an Altoids mint box with a toggle switch mounted on it along with two banana jacks and the TO-220 FET screwed to the bottom of the tin. I plan to use it with this circuit on a small perf board. That will leave barely enough room for a 9 volt battery and one AA battery to provide the 1 amp current.

    Based on the measurements today, the precision current source project is back to the design phase to find a design that works solidly. :( I'll go take a look at the stuff Bill posted.
     
Loading...