LM25010: Output voltage and FB voltage rising when load is applied

There is a mistake on the schematic. R106 is actually 820 ohms in my circuit. Ignore the marking that says 8.2k.

I began to change R107 before I saw your comment. My results are below. I will try again with ~0.1R, but the results with 2.7R and 0R were not great.

Real application for this circuit powers downstream PSUs, expected load <1A.

R107 = 6.04R

          Load = none, Vout = 17.05V

          Load = 330R, Vout = 18.2V

          Load = Application, Vout = 17.43V

R107 = 2.7R

          Load = none, Vout = 16.85V

          Load = 330R, Vout = 17.85V

          Load = Application, Vout = 17.3V

R107 = 0R (shorted)

          Load = none, Vout = 16.93V

          Load = 330R, Vout = 17.3V

          Load = Application, Vout = 18.05V

Ok, I got a good feeling 100mOhm will give you a good result. Please reply back with the results. I would also like to review a waveform of SW to confirm the PWM duty and frequency is constant.

100mOhm did not solve the issue. 

R107 = 0.1R

          Load = none, Vout = 16.92V

          Load = 330R, Vout = 17.24V

          Load = Application, Vout = 17.94V (output current measured at ~100mA)

The switching waveform (U9 pin 2) is below, with the 100mA load.

That is surprising.

Here is my math...maybe something was done incorrectly.

If you could, also measure FB. We should be seeing a 25mV increase in FB, cycle by cycle.

I am going to order an EVM to take a look at this on my end if we have continued issues.

It might also be a good idea to evaluate on a "Fresh" unit.

FB waveform with 100mA load. I do not have a 'fresh' full unit available, but I can swap the LM25010 on this unit to a fresh one.

Hi Brett, thanks for the additional waveform. We are seeing upwards of 1V excursion on VFB which still explains the poor line/load reg.

Per my calculations, we should be seeing something on the order of 25mV. With that said, I think we are overlooking something.

I would say one final datapoint would be decreasing R107 to 5mOhm,10mOhm,20mOhm and doing the same measurement as you have done.

I take it you dont have a current probe, so we cant measure the actual inductor current ripple.

This experiment will allow us to see how VFB scales and determine actual inductor current ripple based on corresponding FB ripple to give us a possible explanation to why we are finding our calculations not agreeing with bench data.

Also, during this experiment, please evaluate at full, 1A load current so we can operate in CCM, which will make the FB voltage rise/decay easier to see.

I do have a current probe - what conditions would you want me to measure the inductor current ripple under?

I will make some measurements with an array of R107 values.

That's good news. You can measure inductor current by tombstoning the inductor on SW and putting a wire between the floating inductor terminal and its landing pattern which is now open and grabbing onto that wire with the probe.

I would evaluate at 32Vin and app min Vin with a load current of 330R and 1A.

I just had an interesting observation. I replaced the IC this morning and the results with the 330ohm and 100mA loads were the same as previously.

However, with a load of 19.6ohms (~850mA), the voltage output was ~16.7V, which was promising.

I then tried to repeat the test and the IC had lost its ability to regulate, even under no load. At no load, it is at 17.7V, and with the 850mA load, it is at 17.4V. 

That is good to know. Had you the chance on testing the different Rramp array?

I am preparing to do that test now.

This morning, I replaced the IC again and I am finding that it is behaving well with a resistive load. Across a range of loads (10 tests linearly from 0ma to 930mA), the output voltage is within 2% of 16.8V. 

When I connect my intended application load (downstream power supplies), it loses the ability to regulate and jumps to 17.8V.

When I connect my intended application load and a separate 100mA load (160ohm resistor), it regulates to 17.3V.

To test if my application load is damaged, I supplied it 16.8V from a bench supply and it drew 100mA without issue. I am not sure what to make of this.

Hi Brett,

If you have 20mV or greater FB ripple (ie: 100mOhm Rramp) then your design is OK to proceed with.

I am glad you found that the downstream load was the cause of the issue. It could be the capactive ripple was squelching the injected voltage ramp.