TPS25980: 7 does not start properly

Hi Kenji-san,

375Ohms RLIM set the ILIM to be around 4A which is equal to your actual current. You should set the current limit with some buffer like 10% or 20% above your nominal operating conditions.

KENJI FUTATSUGI said:

After that, the relay turns on and the 24V input stabilizes, but the TPS259807 keeps restarting. (Retry is set to infinite)

Do you mean that input voltage of eFuse is stabilized but the output is not 24V and the startup fails. This is happening due to very high power dissipation during the startup. You have 4A continuous current load and the output caps will also have some inrush current and during startup, Vin -Vout will be large. The device will perform thermal shutdown to protect the FET from such high power dissipation. changing the ILIM and dvdt cap will help here. Can you tell me the output cap you have in your system.

Best Regards,
Arush

Mr.Arush

Thanks for your reply

>> Do you mean that the input voltage of the eFuse is stabilized but the output is not 24V and the startup fails.

That's right.

I calculated using TPS25980xx_Design_Calculator_RevA.xlsm and tried ILIM and DVDT.

I tried DVDT with 750pF, 3000pF, 4700pF, and 0.22uF, but they all kept restarting.

The output capacitor is 0.1uF.

Best Regards

Are you simulating this or do you have this hardware. If you have the hardware, I suggest to test on it for final verification.

Mr.Arush

Thanks for your reply

The calculation was done by inputting Cout as 0.1uF and Rload as 6Ω.

After that I tried various things, and in conclusion I was able to get the device to start up by controlling the EN signal from a separate power source, rather than dividing it down from the 24V input voltage.

It seems that the presence of the inrush current prevention circuit is the cause of the eFUSE UVLO triggering several times.

In the attached waveform, yellow is the original 24V, blue is the 24V connected to the eFUSE which is the latter stage of the inrush current prevention circuit, red is the eFUSE output, and green is the original 24V and the current waveform.
The resistance of the inrush current prevention circuit is bypassed after about 36ms, so the blue voltage becomes 24V.

The first image shows the eFUSE output unloaded. 

The second image shows the device with 6Ω connected to the eFUSE output, and after several UVLO cycles it fails to start.

Could it be that the repeated ULVO state causes the internal temperature to rise, resulting in a thermal shutdown?

Additional information:

The reason we chose to control the EN signal from a separate power supply is because we want to operate the EN signal
to activate the eFUSE some time after the inrush current protection circuitry of the 24V input supply has been bypassed.

Hi Kenji-san,

When you don't have any load, UVLO is also not triggering and device is also performing the startup.

KENJI FUTATSUGI said:

The second image shows the device with 6Ω connected to the eFUSE output, and after several UVLO cycles it fails to start.

when the VIN is dropping due to inrush limiting resistor, UVLO will trigger and device will turn off and VIN will recover and device will come out of UVLO. I think you are seeing this and this is expected when controlling UVLO with resistive divider from Vin.

Can you confirm that the issue is the device is not starting up when Vin is stabilized to high value.

In the region ii, the input is stable but device doesn't looks like it is retrying. I understand that you have mentioned this above but can you please double-confirm if the NRETRY is set for infinite retries.

Also you mentioned that with external EN signal, device is able to perform startup. The UVLO won't affect the startup and because your Vin is lower, the device will also have lower power dissipation. 

Also did you try the startup with higher ILIM. We don't want device to hit current limit all the time.

Best Regards,
Arush

Mr.Arush
Thanks for your reply

I have already tried a setting to retry infinitely, and it continues to retry even when the input power enters the stable region.

I thought that it would start up once it entered the stable region and UVLO no longer occurred, but for some reason it kept restarting,
so I had no choice but to take the measure of delaying control of the EN signal.

I have not changed the ILIM setting and performed an evaluation.
However, since the specifications require the output to be turned off at 4A, I do not want to set it too high,
but I would like to try it if there is a possibility of improvement.

Hi Kenji-san,

KENJI FUTATSUGI said:

However, since the specifications require the output to be turned off at 4A, I do not want to set it too high,

How much is your operating current and load current during startup.

Best Regards,
Arush

Mr.Arush

The operating voltage at start-up is less than 10A.
The load voltage must be able to handle up to 4A, but this is not an issue during start-up.

Best Regards

Hi Kenji-san,

I am still unable to fully understand your system/concerns/issue. Do you think a meeting early next week will help here.

Best regards,
Arush

Mr.Arush

I'm sorry it's been a while.

What is clear at this point is that if you control EN and delay the startup timing so that UVLO does not occur at startup, the device will start up normally.

So it's unclear why UVLO occurrence at startup causes an abnormality, so what information is missing to solve this? If there is any other information I need to submit, please let me know and I will provide it.

Best Regards

Kenji-san,

Thank you for your patience.

I went through full thread above. I am summarizing it.

You are using a current limiting resistor which turns off in 50ms. You are keeping 4A as current limit and startup load is 6Ohms (24V operating voltage so 4A load). We know that the issue is not due to current limit resistor as in the initial startup when the current limiting resistor is present, the Vin is dropping and causing the UVLO but afterwards when the resistor is disconnected, we still see device failing to startup.

There is no issue when there is no load. I am expecting similar results for lighter loads like 1A or 2A. I see in the waveform that during failed startup, 2A is the maximum current so the max power dissipation during that time will be around 2A*(24-10) = ~25W-30W which should trigger the startup thermal shutdown in 4-5ms which is consistent with what we are seeing. 

In this scenario, first thing I would suggest is keeping dvdt floating for fastest startup time. I know that you have tried with dvdt values as low at 750pF and that looks like little marginal case. better to check with dvdt floating case.

Best Regards,
Arush