LM5176-Q1: LM5176 no switch on low side Gate2 of mosfet

Re-post for waveforms.

50Vin , Output Load 0.5A

50Vin , Output Load 0A

Hi Aska,

Could you replace the IC and retest?  The waveform appears incorrect with a good IC.  The circuit should be in the buck, and boost, alternative switching, but your SW1 and SW2 almost move together and this is not right. 

Thanks,

Youhao Xi, Applications Engineering

Dear Youhao:

Thanks for your reply.

Before your suggestion, I have replaced by new IC or new built board. But it keeps same problem.

Follow your suggestion, today I tried it again. It also doesn't work normally. (The LG2 of mosfet doesn't switching. Same as previous waveform.)

Could you please share another suggestion to me?

Appreciated you so much.

Aska

Hi Aska,

Your scope picture really confused me.  Like in the first picture, if LG2 (i.e. LDRV2) does not switch, SW2 will not move to low but stay as a straight line.  Please double check if you have the right connection as shown in your posted schematic, or if your probe was placed at the right circuit node.

By the way, is your output voltage at the regulation setting point?

Thanks,

Youhao

Dear Youhao:

Thanks for your input. I will do it, but I have researched it for a week. Now I power up another practical circuit (LM5170) for 3KW demand. 

The probe of LG4 is correct after I have tested it by swap probe.

If Fsw is 150kHz(R_RT is 56.2Kohm), the output voltage 40V@0A is not at my set point (48V). While the output load is increasing (eg. 0.5A), the Vout is drop to 35V.

If Fsw is 300kHz(R_RT is 27.4Kohm), the output voltage 25V@0A is not at my set point (48V). While the output load is increasing (eg. 0.5A), the Vout is drop to lower.

Thanks.

Aska

---BTW, I remember a strange thing when I research this board.---

While I tried cut off PCB trace for the ground connection between power gnd and IC gnd, I saw the LG2 signal is appear for a short time. But I didn't catch it by scope.

After this, the IC (LM5176) is damaged (SS pin impedance is only 88ohm).So I replaced it by new IC. But the DC source is still current limit during I power up it again.

Then I found the HG1 mosfet D-S is short. After replaced the mosfet, the phenomenon is same as pervious waveform (no LG2 switch) and Vout is only 39V.

After this event, I changed another new PCB board (no PCB trace cut-off) to keep research this problem. Always same phenomenon.

--------------------------------------------------------------------------------------

Hello Aska,

If there is no load on the output and your output voltage is not at the set output voltage, something is not working properly in the design. Would you please check if the output voltage is at the desired output voltage at different input voltages with no load.

BTW, as the output voltage is fairly below the required output voltage and the input voltage it is normal that the lower transistor on the boost side does not switch.

Is it correct that the circuit is simulated with pSpice? If this is the case it should work as well in your design. I would recommend to order an EVM, replacing the components with yours and check if it works on the EVM layout.

Dear Brigitte:

Thanks for your suggestion.

After checked it, the output voltage at no load can be regulated at the set point voltage. I am still looking for the wrong thing in the design.

Before this current schematic and PCB, I have tested it in previous PCB by rework wire. I am sure it is workable that Vout is at the set point voltage.

That is good idea that confirm the circuit in EVM. I will try get the EVM from TI sales.

BTW, I found a problem while I study my PCB.

The VINSNS pin2 is drop. Could you please provide some suggestions after you reviewed my test waveform?

Thank you so much.

Aska

TEST1: LM5176 turn-on while 50Vin power up.

TEST2 - I tried to rework the CS/CSG trace by jump wire.

Dear Brigitte:

Thanks for your suggestion.

After checked it, the output voltage at no load can be regulated at the set point voltage. I am still looking for the wrong thing in the design.

Before this current schematic and PCB, I have tested it in previous PCB by rework wire. I am sure it is workable that Vout is at the set point voltage.

That is good idea that confirm the circuit in EVM. I will try get the EVM from TI sales.

BTW, I found a problem while I study my PCB.

The VINSNS pin2 is drop. Could you please provide some suggestions after you reviewed my test waveform?

Thank you so much.

Aska

TEST1: LM5176 turn-on while 50Vin power up.

TEST2 - I tried to rework the CS/CSG trace by jump wire.

Hello Aska,

From VISNS to SW1, the voltage drop across the top transistor is measured. Can you please measure the current through the inductor when this happens? In addition, please review the connection of VISNS for possible injections or problems.

Dear Brigitte:

Appreciated any input about LM5176 from you.

About the waveform (inductor current during VISNS drop), please refer the previous waveform-2 (Zoom-in, Trigger VINSNS and scope pin2 VIN of IC).

Actually the inductor current is around zero during VINSNS drop.

Today, I have another result. If added a cap (0.1uF) in VINSNS pin3 of IC, the VINSNS drop is disappeared. But no LG2 switching. I guess it shall be important key.

BTW, same value circuit in previous version PCB board. LM5176 is working normally with LG2 switching.

I will keep looking for the PCB issue, but I hope you provide any direction or suggestion. Thank you very much.

Aska

0112Test1-Add a 0.1uF cap in VINSNS pin3 of LM5176

Result: No VINSNS drop.

But the LG2 is still no switching.

0112Test2- Test same value circuit in previous version PCB board.

If it is normal working, the LG2 is switching. (Base on same test setting.)

Of Couse, no VINSNS drop. And the output voltage at no load can be output desired set point.

Hello Aksa,

Please do not add a capacitor on the VISNS pin. The inductor current direction is sensed on this pin, so this will cause regulation issues if there is no load on the output or the output voltage has a transient from high to low load.

If the old circuit works, please check which components you changed between the old and new circuit. Even if you only changed a resistor, it is good to know.

If there is absolutely no difference between the old and new schematic, the problem is caused by the layout and I do not see any other possibility than changing the layout closer to the old one to remove the issue.

Dear Brigitte:

You are right that PCB layout shall be back to old, but I have no chance. The new PCB space is small than old one.

Good news is LM5176 converter can power up with normal LG2 switching and support 3A load successfully.

Bad news is LM5176 could be damaged during power off. It can power up normally. Next power up it is damaged. (The impedance of pin8 SS is below 100ohm)

I modified many PCB trace until cut-off the trace of SW1 and HG1.

I found they are routed in layer3 and cross over the pads of LM5176 (bottom layer).

I understood your concern that 0.1uF is in VINSNS. Thanks.

May I put 0.01uF cap in VINSNS? Evan I modified the PCB trace, the VINSNS drop is still there. If the cap is bigger than 0.01uF, the VINSNS drop can be minimum.

I keep looking any possible.

Appreciated any suggestions from you. 

Aska