L7987L Failure

[Shivalee]

Hi, we are using L7987L IC top generate -15V from 24V input. However, we see repetitive failure in the IC. I have attached the waveform also (Green- Input Voltage, Pink- Vout). I am also sharing the layout for the same. Switching frequency we have kept around 470 kHz.

[MHoll.2]

Can You post the "real" schematics so I can understand better (if possibile as PDF)?

Is the selection of a standard pn Diodo with 1A/1000V (recovery time is 75ns) ok? 

Do You see some negative spikes on Pin 13,14?

Have You tried to using a Schottky Diode? (if 27V is the max. input voltage You need a diode with Vbr > 42V --> 60V should be OK, this is also the max. voltage of the L7987L)

Martin

[Shivalee]

Hi Martin,

I cannot really post full schematic but I will provide snippet of the circuit. We are utilizing -15 V to run hall effect current sensor and few comparators. Overall loading is not more than 0.2 A.

Do You see some negative spikes on Pin 13,14?- So far we have not observed any negative spikes. Every time failure occurs, shorting is observed between pin 13 & 16 only (after removing inductor from board), which means that LS MOSFET is failing (pls correct me if I am wrong). 

Have You tried to using a Schottky Diode?- We have not tried it yet, but we are using same IC to generate +15 V from 24 V and have used same L, C & diode, and different compensation but it works just fine.

I have also attached waveform across the diode D37.(During healthy condition)

Yellow- Vin=24V, Red- Vout=-15V, Pink- D37.

From this much ringing across D37 we inferred that inductor is running in discontinuous mode (also note that this is a condition where load is minimum).

Also, please see waveform of failure.

 -Shivalee

[MHoll.2]

Yes having a short circuit between pin 13/14 and 16 could indicate an over voltage condition.
Using the L7987L as "normal" step down converter is less demanding on the DC/DC converter chip (max. voltage on the switch is Vin + Vdiode), while using the DC/DC as inverter will generate a much higher voltage on the switch (Vin + Vout + Vdiode)
I have noticed that You are using a Datarecoder, what is the bandwidth of the Chanel You are using to measure the Voltage over D37, could it be that you don’t capture a fast negative spike?

[Shivalee]

Hi,

I am using Yokogawa DL850E (BW is 2MHz).

From your suggestion, I have changed diode to schottky, IC failure is not observed but still output is not as smooth as desired. (Red is -15V, Pink across Diode). I believe this is compensation issue

[MHoll.2]

Happy to hear that you have no failure.

Yes probably you have to tune the compensation cirquit.

I would ad a (ceramic) low ESR capacitor in parallel to C103.

Checking against the application note in https://www.st.com/resource/en/datasheet/l7987l.pdf, You don't have the RC lowpass filter on pin 4 (VCC), please verify that Your input voltage is clean (against IC_GND and GND).

[Shivalee]

Sure, will check and update designs with your suggestions.

Thanks a ton, you have been a savior.