LT4250 Hot Swap power shor circuit test

Hi,

LT4250 has a constant 500usec timer duration. With 20m ohm sense resistor , the current will be limited to 2.5A (nominal). The MOSFET that you are using does not have enough SOA to withstand 3A (taking max sense voltage tolerance of 60mV) for 500usec with a large Vds such as 71V as per your OV setting. 

I will recommending using a higher SOA MOSFET such as the PSMN3R7-100BSE or the IRF530 .

Best Regards,
Ritesh

Dear riteshl,

I change the MOSFET to IRF3710STRLPBF  and using 2mR (CSNL1206FT2L00) resistor. In this situation, we use 48V power supply.

When we power on the supply, after some Milliseconds the Drain-Source of Q9 short and the hot swap controller do not protect the board.

Isn't that the same MOSFET that you were using earlier?

Also schematic shows 20m ohm sense resistor and you mentioned 2 m mohm. For 2 mohm , current limit can be 30A (60mV sense threshold max). This MOSFET neither has SOA and also it cannot dissipate such high I^2R loss. A typical rule of thumb is to maintain DC dissipation less than 2W for a D2PAK MOSFET. Hence for 30A , you require overall Rdson of 2.2mohm or less. You might want to put 2 PSMN3R7100BSE FETs in parallel and retry the test.

Regards,
Ritesh

Thanks riteshl.

Your answer is helpful.

I need to add a restart control for hot swap circuit.

Do you have any document about it?

HI ,

You can add the automatic restart circuit as shown below. The behavior of the circuit is as follows:

Before any short circuit etc , GATE pin is high and Q3 is turned on , pulling node 2 to VEE. Resistor R8 turns off Q2. After a short occurs, the GATE pin is pulled low and Q3 turns off. Node 2 starts to charge C4 and Q2 turns on , pulling UV pin low and resetting the circuit breaker. As soon as C4 is fully charged, R8 turns off Q2, UV goes high and gate starts to ramp back up. Q3 turns on and pulls node 2 back to VEE. D1 clamps node 3 one diode drop below VEE. The duty cycle is set to 10% to avoid Q1 overheating.

Let me know if you have any concerns.

Regards,
Ritesh