0

I'm working on a electronic load and below schematic is the simplified version of it.

schematic

simulate this circuit – Schematic created using CircuitLab

The problem is the thermal drift in resistors and op amp but what makes several mA drift is the shunt, although it stabilizes after initial drift but it affects the first set current value.

I have tried almost all kind of resistors as a shunt but less or more they all have some drift with temperature.

Is there anyway to compensate for thermal drift and what are our options?

ElectronSurf
  • 2,185
  • 3
  • 24
  • 61
  • Over what temperature range are you wanting to stablise? Have you looked at NTC + normal resistor combination, where you organise it to cancel out (or minimise) the thermal effects, rather like [this](https://en.wikipedia.org/wiki/Gridiron_pendulum) – jonathanjo Apr 04 '23 at 09:42
  • Give an example of the actual current drift i.e. so-many mA or amps drifts to x mA or amps over a time period of y seconds. Numbers are important as are data sheets, op-amp power connections and expected performance. In fact, why don't you show the proper circuit instead of that pointless simplified circuit? – Andy aka Apr 04 '23 at 09:48
  • @jonathanjo I did actually read an article about connecting a thermistor to the shunt but the problem is that the design is all analog and there's no micro-controller to read thermistor value and change the set current. – ElectronSurf Apr 04 '23 at 09:48
  • @ElectronSurf ... no I did mean you use the NTC as a resistor, you pair it with a normal resistor with positive coefficient (in series, say) to approximate a no-thermal-drift resistor. No microprocessor, no measurement. – jonathanjo Apr 04 '23 at 09:52
  • @Andyaka Hi Andy, the change is almost linear; with ~5℃ increase ~1mA drift. so I do have 2 to 2.8mA of -initial- drift that I'd like to avoid. I'll update the schematic. – ElectronSurf Apr 04 '23 at 09:56
  • @jonathanjo The shunt value is 20mΩ, How do I pair it with NTCs that are several hundred ohms? can you please elaborate a little. – ElectronSurf Apr 04 '23 at 09:57
  • 2
    It's vital that you show how the resistors connect together to see that you haven't created problems with volt-drops down wires. You still haven't shown the power supply connections to the op-amp. I suggest that you ditch the CircuitLab tool and post your own schematic. – Andy aka Apr 04 '23 at 10:07
  • 5
    The sense resistor is dissipating over 400mW if the schematic is accurate. Maybe you can reduce the dissipation by decreasing the sense voltage and perhaps put a few (physically separated) resistors in parallel for the sense resistor temperature change is reduced. By cutting the voltage to 1/3 and using 3 well separated resistors in parallel you can get a reduction of close to 25:1 without even improving the resistor performance. – Spehro Pefhany Apr 04 '23 at 10:53
  • 2
    @ElectronSurf Please, post here whatever is needed to give you a meaningful answer. This is not a general electronic forum. The question must stand on its own feet. Links to external resources are ok as long as they give collateral details, but the bulk of relevant information should be posted here. Voting to close until the relevant information is posted. – LorenzoDonati4Ukraine-OnStrike Apr 04 '23 at 11:58

1 Answers1

1

It all depends the stability you are targetting.

Usually you don't need to compensate for the thermal drift of the shunt resistor.

The key element that make this possible are :

  1. Choose the smallest shunt resistor value you can that is up to the task. Resistors in the magnitude of 0.1 Ohm are typical. Therefore, with P = R * I^2, The power that the resistor has to dissipate will be small and it won't heat up that much.

  2. Use a resistor model that is meant to be a shunt resistor. It will have a very big package and good thermal conductivity to the PCB or a dedicated heatsink. That way, the dissipated power won't heat up the resistor too much. They may also use a material that has a thermal coefficient that is close to 0.

  3. You may use an OP amp that is designed to sense current shunts.

Blup1980
  • 6,120
  • 3
  • 26
  • 47