Problem: When current is enabled, my load detection indicator LED stays on whether the load is attached or not.

My constant current power supply, circuit attached, is working just great at breadboard stage EXCEPT for the most simple task, detecting the load.

In the diagram, everything to the right of the load is simply for detecting the presence of the load, and lighting up an LED indicator. The circuit as shown works great when the main current flow is disabled; no load connected=LED off, load connected=LED on. But enabling the current causes the LED to light constantly regardless of whether the load is connected or not.

I can't seem to find the magic resistor values that fix this. If I raise the load detection comparator reference voltage from ~0.24v to something much higher, eg. 2.5v, the LED flickers when the load is on and stays on constantly when it is removed. I believe it may be flickering a bit even when the reference voltage is 0.24v. How can that be? It's not the power supply itself, since other LEDs lit by the same power are not affected.

Also, I've tried using the optional resistors shown to protect the comparator's inverting inputs. They don't seem to matter one way or the other. Is there a reason to leave them in place or can I simply omit them?

[Update] To be precise, R22 is currently 0.47Ω and the current flow when enabled is ~0.5A.

 

Try reversing the plus and minus leg inputs of the comparator(IC3-4) and then try out some larger values for R4 in the voltage divider.


Does your load draw a steady current or does it operate in surges(pulses)?

 

Try reversing the plus and minus leg inputs of the comparator(IC3-4) and then try out some larger values for R4 in the voltage divider.

Does your load draw a steady current or does it operate in surges(pulses)?

Load is static for sure, purely resistive.

I don't think reversing the poles can work overall. Keep in mind that it works now when the current is disabled. The voltage at the bottom of the load is actually above the 5v rail of the comparator, so there's no reference voltage that would turn that on if the poles were reversed.

 

...try out some larger values for R4 [actually R24] in the voltage divider.

That got me thinking. I had tried that, but not systematically. All I had to do, to do it right, was measure the voltage at the bottom of the load, when the load is not connected but the current is enabled. That's the set of conditions where I was having the problem. Sure enough, the voltage was ~550mV, putting it way over the reference and lighting the indicator. So I boosted the reference voltage just over that (R24 = 16K) and off went the LED. Now the indicator works properly whether the current is enabled or not.

Thanks for the nudge!

 

Still wondering - do the inverting inputs of the LM339 need current limiting resistors? The schematic for the 339 shows the input connects directly to the base and collector of an NPN transistor, and thus I can see where current limiting might be wise. Is that right?

Same question for the LM358. I'm pretty sure it does NOT need limiting. Right?