LM5012-Q1: VOut Dropping drastically when Applying a high enough current

Hello, 

The feedback and ripple injection network impedances are quite high here (3MΩ upper feedback resistor for example). Try scaling down the impedances by 10x to improve noise immunity.

--

Tim

Also, 47pF is a really low value for the coupling cap (to FB) and may be getting discharged and thus affecting regulation. Try increasing this to > 100pF.

We'll try reducing the impedance first, see where this will lead us 

Thanks, Luca

Hi Timothy, 

We changed the value of the Impedance by changing Rfbt to 470 kOhm and Rfbb to 15kOhm, the voltage in regulated finely at 39.4V  until we hit 0.5 A. At that point,  the bus began to drop until we reached again 1.4A and the bus went all the way to 4V. We'll try to change the capacitance for a 100 pF one 

We just changed the capacitance and it did not help 

Do you think it could come for a defectuous batch of component ? We have two PCBs that were built at the same time. We might go for a change of component for LM5013-Q1, but we are exploring every possibility before doing that 

Send on a completed quickstart calculator for this design, so we can check the ripple injection components, etc. Also, verify the waveform of the ripple injection at FB.

Here is the suggested design by webench 

I don't think there is any quick calc tool for the LM5012 Q1

Here is the Vout profile with a 0.5 A load , the mean is at 38.5V

Here is the Vout profile with a 1.4 A Load, the mean is at 5V

The upper plot shows some instability. Try a higher coupling cap value to debug this, e.g. 1nF (the higher cap just slows the transient response a bit). I have seen with high Vout when there is a large voltage across the coupling cap (Vout - Vref) that it's quite easy to have a slight discharge that affects the FB voltage and associated ripple information.

Could it be that we hit the current limit of the IC because of the inductance ? 

I doubt it as this is a 2.5A a rated converter with peak current liimt set above that, but check the inductor current to verify.

Here is a compilation of FB and SW respectively in order : 0A, 0.2A, 0.5A, 1.2A, 1.4A :

We tried with a 2.3 nF cap and it didnt help at all 

Our inductance is already shielded. I think we will try to change it for an higher inductance (e.g. 47 uH)

the Output cap is ceramic but we'll try to put a 1uF cap in parallel

Here is as a reference the PCB trace of the Inductance:

Do you think the trace between the output cap (C5) and the inductance is not adapted ? 

for the FB impedance we might go outside the recommended values if we continue to reduce it. We'll try changing the inductance beforehand. 

We try adding a 1 uF in parallel , it did not change the behaviour. We also tried with a 10 uF and no luck either 

C5 is indeed a ceramic capacitor already. Our inductor is already shielded also