Hello,
We are using TPS259827ONRGER in our product and discovered some problems recently when it started with high output capacitance. The following is schematics of eFuse in our design. An MCU receives Power Good signal and sends ENABLE signal to the eFuse. Current limit is set at 7.92A, output slew rate is set its fastest value by making Cdv/dt not populated on the board. Voltage of VIN and VOUT of eFuse can be 5V, 12V and 24V. The start-up problem is discovered when voltage is 24V.
Output power from eFuse in our product is used as power supply to other products from users. The following image shows an example of the application. An individual load can have 220uF input capacitor and around 0.5A in stable condition (probably the start-up current is lower, but there is still some load current during start-up). Users may connect 8 loads, or even more, in parallel to the output of eFuse in our product, which will create a huge output capacitance to the eFuse. In reality, users could use different load with even higher input capacitance, which is something out of our control.
According to section 9.2.2.5 of TPS25982 datasheet, PD(STARTUP) = PD(INRUSH) + PD(LOAD). In order to make PD(INRUSH) lower, a relatively big Cdv/dt should be used to slow the output slew rate. But the PD(LOAD) is only decided by the voltage and load current. In order to make the start-up time shorter than "Time to Thermal Shutdown" in Figure 53, a small Cdv/dt (even without Cdv/dt) should be used to make output slew rate as fast as possible. So, it seems we need to trade off between PD(INRUSH) and PD(LOAD). We decided to not use Cdv/dt to achieve highest output slew rate. Some attempts to slow the slew rate failed probably because of the load current during start-up.
Our test with 4 loads (as mentioned above) shows a 14.7A inrush current, which makes the PD(INRUSH) to around 176W. Here is the waveform of the test.
We tried several approaches to solve the problem, unfortunately, non of them are perfect yet.
(1) We tried to modify the circuit so that eFuse works under indefinite auto-retry mode with minimum retry interval as mentioned in the application report SLVAED9. But the retry is not fast enough. According to the datasheet, the retry timer only starts once device cools down.
(2) We tried to make MCU create multiple fast ENABLE pulses to eFuse. It could successfully bring up the output voltage to 24V with 4 loads connected. However, it still failed when more loads were connected.
(3) We tried to add an inductor in series at the power input of each load in order to limit the inrush current. The inductor is PA4344.223NLT (22uH/12A), but it seems the inductor can hardly lower the inrush current, the measured inrush current was still above 14A. In this test, we just pulled EN signal HIGH once, not multiple EN pulses. Here are block diagram of this test and measured waveform.
(4) We replaced the inductor with a NTC - B57234S0100M051 (10 Ohm). It worked with 4 loads connected, but again it failed when more loads were connected. Oddly enough, no matter how many loads were connected, the inrush current was reduced to around 5.5A, but the eFuse could only bring up 4 loads at most. Here are the measured waveforms with 4 loads and 8 loads respectively.
(5) We also tried another approach with a combination of resistor and diode to limit the inrush current as shown below, but without any luck.
After all these tests, we want to know:
(1) Is there any information we missed when using this eFuse chip?
(2) Is there any other ways we can use to make eFuse work with high output capacitance.
Thank you!
