Post Go back to editing

Current source "without" temp drift


I'd like to make a High precision current source with excellent temp drift (10 to the power of -6 or at best better than that). Current: about 0.5 Ampere

At the moment I have an AD624 combined with a Vishay resistor in mind. How can I go further? Is there a place for me to look up stuff like this?



  • Hello Matthias,

    Do you mean that you need a 0.5A current source which has less than 1ppm/C drift? That is very hard to achieve. It is a highly complex problem. Even if you have one or two very good components, you have to take into account and minimize (or control) every error source in your circuit (and pc board!). Then there is also the possibility to control the temperature so there is no temperature drift within a reasonably narrow operating temperature range. Can you provide more details of your end application? Are there additional requirements such as noise, bandwidth, etc?


  • Hello Gustavo,

    we would like to create a constant magnetic field, which we adjust once. It should be set by voltage or PC. Very low noise is pretty cool, but drift is more important. There are no other additional requirements, maybe not extremely expensive


  • Matthias,

    There are a few options and things to consider that might simplify your design. Some of those include a controlled environment and the ability of tracking the performance of your circuit.

    1. Is this a circuit for production or a single experiment?

    2. For how long does the circuit need to operate?

    3. Do you have a narrow operating temperature range? Is this going to be done in a controlled environment?

    4. Will you have the ability of tracking long-term stability of your circuit output with at least one more laboratory standard?

    5. Is it possible to relax your requirements to say, 5ppm/K?

    And a couple more questions about the source:

    6. When you say "current source" do you mean an actual source or a "current sink"?

    7. If you do mean a current source, what's the maximum voltage to which it needs to operate (a.k.a. compliance voltage)?

    8. I assume your load is quite constant. Is that correct?


  • This is going to be very challenging. If you use a reasonably large value resistor, e.g. 0.2 ohms (200 milliohms) then you will drop 100 mV across it. This means your sense amplifier must have a drift of less than 0.1uV. Opamps like the AD8628, TLC2652, AD8571 or LTC1050 have a typical drift of less than 0.01uV/C, so they may be suitable over a temperature range of 10 C. The trouble is that the resistor will be dissipating 50mW, and even a very good quality foil resistor has a coefficient of 5 ppm at full power. This means you will need to run the resistor at perhaps 2% of its rating.

    If you reduce the sense resistance, the resistor may be more stable but the contribution from the sense amplifier becomes more significant.

    Electrical joints also become very troublesome as thermocouple effects can swamp 0.1uV.

  • @Gustavo

    1. it's for experiment

    2. min. 2h, max 2 days

    3. it's on the field measurement, it must be close to stuff, which heats up to about 60 degree Celsius (in the lab at 20 degree), when we switch it on

    4. no, this should be the laboratory standard

    5. 5ppm/K isn't fine, but if better is to hard to achieve, it's maybe ok

    6-8 the source should drive an air-core coil with 9 Ohm (max. 4.5 Volt), the field around isn't changing


    see 5.


  • Matthias,

      Another thing that might help is whether or not it has to be 5 amps absolute, or just constant.  In other words,

    if 4.993 amps is o.k., as long as it doesn't change, then maybe the problem is a little easier.  The other consideration

    would be if you want it to be constant at some value you set, then sensing the field and opertating closed loop

    might work.  But then you have the temperature dirft of the sensing element and associated electronics.

    To give you some appreciation for the difficulty of your problem, if you just wanted a simple voltage reference that

    put out 5V at 1 mA with less than 1 ppm/deg C, that would be very expensive........


  • Matthias,

    We can do this exercise to try to make it as good as possible. Then we can see if it's good enough for your application.

    Since you said that your problem is to build just one, it would be a little easier to find components. If one component is not good enough, you can get a few and get the best of all. There are a few tricks that you can play. Also, if your operating temperature range is narrow, that will help too. Not all the drifts are linear; when you have a limited range and you operate around the flat portion of the drift, you can get better performance than specified.

    Do you feel comfortable answering a few extra questions?

    - Can you run this for a long warm up time (say 2hrs) and wait until the source stabilizes?

    - Do you have a settling time requirement? (e.g. from the time it turns on until it gets within 0.1% from the final value)

    - Do you have an oven and a decent meter that you can use to test a few components and to characterize the final system performance?


  • Hi,

    we require ovenless construction and the possibility of an immediate start of measuring - no heating... until it's ready.

    Is there any up to date application note, like "Dialable ampere current source with ovenless construction and ppm/h stability"" of 1976?

    If 1ppm/C is too  hard to achieve, it's ok.

    If you have any circuit/useful components in mind, please recommend.

    Thanks in advance,


  • This question has been assumed as answered either offline via email or with a multi-part answer. This question has now been closed out. If you have an inquiry related to this topic please post a new question in the applicable product forum.

    Thank you,
    EZ Admin