3

I'm trying to build a circuit that limits the current flowing to the load at a specific level. To do this, I'm using a current sensing IC (INA138) and comparators to provide the necessary logic. However, when I try to use more than one comparator in the logic, the simulation bugs out and is never able to load. EDIT: by "bug out" I mean that the program continues iterating infinitely upon attempting to simulate.

This is a simplified version of my circuit with one comparator: (it works fine if the load is set to a level where the current sensing doesn't need to kick in - this is why I'm using a load of 2R).

enter image description here

And this is with two comparators:

enter image description here

This is a simplified version of my circuit; I need to perform other tasks preferably with other comparators before the logic signal is sent to the NPN BJT to open the PFET, so the second comparator probably doesn't look like it's particularly necessary there (I could upload a version of my circuit that's more elaborate for that part but I'm not sure that level of detail would be clarifying).

Is it possible to do what I'm trying to do with multiple comparators feeding into each other? Or should I set up some other elaborate solution involving PFETs or something?

Output between the comparators: enter image description here

The INA138 is a non-standard LTSpice component. I've made it available at: https://easyupload.io/2e1l4j

EDIT: user @Kuba hasn't forgotten Monica suggested I used some kind of switch to shut off the load. Would this work?

Circuit: enter image description here

Would this work in principle so long as the capacitor is charged? I've designed it to discharge over 2 seconds but how would I ensure sufficient charge on the capacitor at circuit start?

p.chives
  • 105
  • 2
  • 9
  • 1
    Could you please provide details regarding *"the simulation bugs out and is never able to load"*? Also, if the INA138 model is not native, please add it to the question. – devnull Apr 05 '22 at 13:33
  • 1
    @devnull - by "bugging" out I mean the simulation continues stepping with seemingly endless iterations. Updated question to include your suggestions. – p.chives Apr 05 '22 at 14:11
  • 2
    Does the simulation never get past time 0? If so, [set your power supplies to 0 at startup](https://www.cooperconsultingservice.com/ltspice-tips-power-simulation/), or force an initial condition to the hot end of R5 (Google for how to do that). Spice simulations start by trying to find the DC operating point; that comparator may be preventing one from being there. Let us know if that works. – TimWescott Apr 05 '22 at 15:21
  • 1
    @TimWescott - thanks, I set the power supplies to 0 at simulation start and produced some weird oscillating output which might explain my problem. I've posted a screenshot of the voltage between the two comparators during simulation. – p.chives Apr 05 '22 at 16:01
  • Regarding the latest update, if you mean in the simulation, you could check this: https://electronics.stackexchange.com/questions/237337/how-to-define-the-initial-charge-of-a-capacitor-in-spice – devnull Apr 05 '22 at 17:47

1 Answers1

2

The “weird oscillating output” is because it looks like the circuit is an oscillator with positive feedback. You designed it to be that way. The simulation is correct.

It also has nothing to do with presence of multiple comparators, just that you happened to design a circuit that oscillates when you add the 2nd comparator :)

There’s no stable DC operating point. Thus, if you precede the transient simulation with a DC solution, as is the default, it’s not supposed to ever succeed – that would be a gnarly bug. So it “iterates” forever, by design.

Just think about what you want it to do. A load goes over the current limit. So you shut the load down. What happens to the current then? It goes back to 0. Hey, no more overload. So you turn the load back on. And the cycle repeats.

You need to do something else, for example:

  1. Latch the shutdown state until power off.

  2. Instead of shutting the load, regulate the current linearly, but monitor the mosfet temperature. If it gets too hot, shutdown until power off.

  3. Latch the shutdown until some reset action is performed, like disconnecting the load. You could use a simple impedance sensor to detect presence of a load. Push VCC via a 5M resistor to the load, and if it’s below VCC-10mV, treat the load as plugged in. Reset the shutdown latch when the load is removed, or a “Reset” button is pressed, etc.

Kuba hasn't forgotten Monica
  • 32,734
  • 1
  • 38
  • 103