The circuit below shows a part of a circuit that I am designing for a PSU. This is the first time I do anything even remotely this complex. Video playlist for context. Direct Link to the Testing discussed below

The problem I have with the circuit below is that as soon as the circuit goes into Current Limit mode, it pushes V-Out up to about 8V, and I cannot recover from this, even if I set the current limit much higher than where it initially latched. The only way to recover from this is to un-plug the circuit and then reset.

Current Limit and Voltage Set Circuit

(Note: Original Schematic improved for better readability as requested by Rohat Kılıç and replete)

What I think is happening is that when we go into Current Limit mode, and Q1 turns on, OpAmp (IC3A) tries to compensate for its output being pulled to ground, and somehow drives 8V across... Actually no... If Q1 is on, surely the maximum voltage that it could show would be 0.7V, no? How is that OpAmp able to sustain more than 8V with its output shorted to ground?

Some context

  • I-Sense senses the Current over a shunt resistor and can be 0-3V.
  • I-Set comes from a DAC with an upper reference voltage of 3.0V
  • V-Set comes from a similar DAC with the same 3.0V reference voltage
  • V-Out is the input to an LT3080 Linear Voltage Regulator and could range between 0 and 30 Volts
  • CLIM is purely a digital input to a micro controller to let it know when we are in Current Limit mode.

Before we go into Current Limit mode, everything seems to be working as expected.

Note that I am testing this with no load but there is a 10mA current source connected to the output of the LT3080 as stated by its spec sheet to allow varying the voltage down to 0V.


@user55924: The Image below is what I see of Pin 1 of IC3A when the output is at about 1.7V. Is that expected? I would not have expected the pulse. Does this mean the V-Set signal is not smooth enough? Testing with a standard Arduino Mega PWM smoothed (somewhat) with a crude single stage RC filter. IC3B is off until it latches and then goes to VCC (34V in this case) as expected.

enter image description here

  • 4
    \$\begingroup\$ Any chance you could tidy up the schematic, so that designators aren't on top of each other, outputs don't have input symbols, rails electrically above ground aren't visually below ground etc.? It's a confusing jumble, I'm sure you'd attract high quality answers if it weren't such an effort to decode the schematic. \$\endgroup\$ – user133493 Apr 14 '17 at 7:42
  • \$\begingroup\$ Is R46's one end connected to V-OUT? Since there's no junction point, I couldn't be sure. Anyway, this is a feedback block, it's quite easy to see that IC3B's output (current limiting signal) goes to somewhere in the controller block but what does IC3A's output do? How is the output voltage regulated? \$\endgroup\$ – Rohat Kılıç Apr 14 '17 at 7:58
  • \$\begingroup\$ @RohatKılıç I have updated the schematic for your viewing pleasure. Any assistance would be much apreciated. IC3A drives the Linear Voltage regulators set pin. \$\endgroup\$ – Gineer Apr 14 '17 at 19:08
  • \$\begingroup\$ LM3080 is an OTA, not a regulator. Do you mean LT3080 ? To see which part is not working, measure pin 7 on IC3B (output). If this latches normally, the problem is in the feedback loop. And then measure pin 1 on IC3A, (output). \$\endgroup\$ – user55924 Apr 14 '17 at 19:29
  • 2
    \$\begingroup\$ @Gineer, I mean IC2B. I found a full circuit in your videos on YouTube. I am convinced that there is an error in the circuit. Which leads to the problem you described. \$\endgroup\$ – AltAir Apr 19 '17 at 10:15

What is happening is you are triggering the LT-3080 foldback current limit. Foldback current limiting means the the regulator latches itself into a low current state.

LT-3080 has foldback current limiting

The only way to leave the foldback state is to cycle power. The reason this happens with no load is because the LT-3080 current limit drops to 0 according to note 9 in the data sheet.

differential greater than 25V

Once this happens, there is not enough current through the sense resistors to keep IC3B in the current limit state and your vout is free to rise.

  • \$\begingroup\$ This is interesting. When I initially tested after making the changes suggested by AltAir above, it seemed to just go into that 0 current limit state (although nothing on the board felt hot (or even warm) tot he touch. A couple of things in your description that does confuse me is that here V-Control is at 34V, V-In is 18V. VOut is open circuit except for a DMM in DC Volt mode (I.E. High impedance) so VIn-Vout should never go close to the 26V suggested. I also have a current source connected to V-out to draw the minimum 10mA. I'll retest with a simple resistive load. \$\endgroup\$ – Gineer Apr 20 '17 at 21:46
  • \$\begingroup\$ Adding a load resistor seemed to have stabilised things. I now tested with a 270ohm load resistor and everything seems great. Jack Creasy's suggestion that the current source idea is probably not going to work, is correct, but it was ultimately your idea that sorted my problem out. Thanks you. On with the design... \$\endgroup\$ – Gineer Apr 23 '17 at 7:33

There are multiple troublesome elements in the design, it's hard to know where to start. There is no latch up in the circuit you provided, but when associated with your external power supply circuit (https://www.youtube.com/watch?v=Z_YtQgTnvkU), it is a poor overall design.

I am assuming your design intent here is to provide a CV/CC design, where the voltage is regulated until you reach the current limit, and then the current is regulated. Your design does not do this.
In your design, you attempt to drop the voltage to zero by shorting out the 'reference' output with Q1. But the time delay in shorting out the I(set) pin defeats most of you objectives.

  1. The LT3080 needs to be included within your feedback path. It currently is not since you are trying to set I(set) and with long delays when you have a 22 uF capacitor connected to it.
  2. When you reach your current limit you move completely out of any regulation feedback. This might work if the only overcurrent that ever occurs is a resistive short circuit. What about modeling your CV/CC approach when you have a string of LEDs attached and you want to regulate the current through them. Any decent CV/CC power supply can do this.
  3. Your current sensing is via an RC delay, which as noted in the other answer almost guarantees you will oscillate between CV and no CV.


  1. Feed your reference voltage V-SET via a 1K Ohm series resistor to IC3A.
  2. Connect you current limit detection, Q1, to IC3A(+). This will now reduce the voltage as the current limit is reached. To get to zero volts you would need to change Q1 to an N-Channel FET of course, or have a negative supply for Q1.
  3. Get rid of the 22 uF on I(set)
  4. Connect the feedback resistor R42/43 to the actual output of the LT3080's to include them in the feedback loop.
  5. While it was a good thought to use an LM334 to supply the minimum current for the LT3080's, it does not work down to zero volts. If you use this then setting voltages below 1 V won't work, and this is probably the reason for the non-linearity you see at low currents/voltages. If you expect the power supply CV/CC support to work down to mV and mA levels then you need to move your current detection to the outputs of the LT3080's and draw the minimum LT3080 required current before the detection point. A simple N-Channel Enhancement mode FET would do the job, though you need a negative supply to get regulation down to zero volts.
  • \$\begingroup\$ This great feedback and awesome advice which i will definitely use in the next revision. Thank you very much, but i don't think you are answering my original question, so I cannot give you the bounty. I have up voted your answer and I hope many others do to. \$\endgroup\$ – Gineer Apr 21 '17 at 22:05
  • \$\begingroup\$ I'm not surprised that you see oscillation, but I have not done an analysis of your whole schematic....it seems to change through your videos so I'm unsure of what you've actually done. However I can say that the power supply you have does not do what you want ... CV/CC .... whenever you trigger your OC sensor, you lose regulation. What you want to do is change from regulating the voltage to regulating the current ....you don't do that. \$\endgroup\$ – Jack Creasey Apr 21 '17 at 22:52
  • \$\begingroup\$ Yea, the videos show a bit of an iterative design learning process. This supply as it stands now is not supposed to be a CV/CC supply, but rather a CV/CL supply. I do have some digital sensors in the solution that might give me the ability to do CC, and I want to use this analogue part as a protection circuit. Thanks for the feedback. Will definitely use your input in the next revision. \$\endgroup\$ – Gineer Apr 23 '17 at 7:27
  • \$\begingroup\$ I was thinking that I could even get rid of the current sensing circuit (IC2A, IC2B and IC3B) and use the INA226 to sense the current digitally (or leave that circuit in, biased slightly higher as a safety). I will be driving the set voltage via a microcontroller anyway, so I should then be able to maintain a constant current using a software PID loop. Other suggestions will go into the next revision. \$\endgroup\$ – Gineer Apr 23 '17 at 16:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.