Good day, Our design is currently as follows: The MOSFETs we are using are: SQJQ184E-T1_GE3 After multiple over-current and over-voltage tests we end up with MOSFETs which have blown. We are drawing 10A and increased the input voltage to 33V which triggers the over-voltage protection. The circuit switches the channel off correctly. After the retry timer is complete the LTC4364 tries to switch the channel on again. With the 10A still being drawn on the output it manages to get to 10 V. At 10V the output voltage suddenly rises sharply. Inspection shows that the MOSFET blew. During the turn-on the LTC4364 required 0.5 mS to switch the MOSFET on before it blew. There were 33V across the MOSFET with an output resistance of 3 Ohm. 10 V were reached before the MOSFET blew. 10A * Vmosfet(33->23V) * rise time(0,5s) = 392 W I realise in this case that the 600W of the SQJQ184E degenerates rapidly as it warms up and hits its 200 W limit at 125 C. What are our current options? Is there a way to make the turn-on time faster? Which MOSFET would you recommend if we were to replace the SQJQ184E? We are very space-constrained. Should we look at a MOSFET with smaller input capacitance? The maximum expected current through the channel is 60 A, this will only occur once the MOSFET is switched completely on. The maximum turn-on current is 20 A.

MOSFET Selection LTC4364

Bubbles27

on Aug 22, 2024

Category: Hardware

Product Number: LTC4364

Good day,

Our design is currently as follows:
The MOSFETs we are using are: SQJQ184E-T1_GE3
After multiple over-current and over-voltage tests we end up with MOSFETs which have blown.
We are drawing 10A and increased the input voltage to 33V which triggers the over-voltage protection.
The circuit switches the channel off correctly.
After the retry timer is complete the LTC4364 tries to switch the channel on again. With the 10A still being drawn on the output it manages to get to 10 V.
At 10V the output voltage suddenly rises sharply. Inspection shows that the MOSFET blew.
During the turn-on the LTC4364 required 0.5 mS to switch the MOSFET on before it blew.
There were 33V across the MOSFET with an output resistance of 3 Ohm.
10 V were reached before the MOSFET blew.
10A * Vmosfet(33->23V) * rise time(0,5s) = 392 W
I realise in this case that the 600W of the SQJQ184E degenerates rapidly as it warms up and hits its 200 W limit at 125 C.

What are our current options?

Is there a way to make the turn-on time faster?
Which MOSFET would you recommend if we were to replace the SQJQ184E? We are very space-constrained. Should we look at a MOSFET with smaller input capacitance?
The maximum expected current through the channel is 60 A, this will only occur once the MOSFET is switched completely on.
The maximum turn-on current is 20 A.

Edit Notes

Updated Title
[edited by: Bubbles27 at 5:42 PM (GMT -4) on 22 Aug 2024]

Parents

Ashapiro

on Aug 27, 2024 9:43 PM
0
Bubbles27 said:
Is there a way to make the turn-on time faster?

Speeding up start-up time and increasing inrush is not the answer to blown FETs.
Once you select the inrush current and assume max Vin, check if the HGATE FET has the Safe Operating Area (SOA) performance to survive start-up.

Bubbles27 said:
Which MOSFET would you recommend if we were to replace the SQJQ184E? We are very space-constrained. Should we look at a MOSFET with smaller input capacitance?

Small Ciss isn't important in hotswap applications, since we intentionally add Cgate for inrush control.
You want FETs optimized for SOA performance.
Here's an example of such a FET: PSMN3R7-100BSE

Bubbles27 said:
The maximum expected current through the channel is 60 A, this will only occur once the MOSFET is switched completely on.

The maximum turn-on current is 20 A.

Make sure the load is off until Cout is charged (Inrush=0A).
Otherwise, the load current is added to the SOA stress.
Use the ENOUT pin to signal the downstream load that start-up is complete.
Cancel
Up 0 Down

Verify Answer

Cancel
Bubbles27
on Aug 28, 2024 2:40 AM in reply to Ashapiro
0
Thank you for the response.

The SOA of the PSMN3R7-100BSE does not look better than the SQJQ184E

Ashapiro said:
Make sure the load is off until Cout is charged (Inrush=0A).
Otherwise, the load current is added to the SOA stress.
Yes that is what I feared. We were hoping to save on having to protect an output to a DAL-B device which we expect not to fail but having the possibility that it short circuits causes this problem.
Ashapiro said:
Use the ENOUT pin to signal the downstream load that start-up is complete.
Yes, we will reconsider our power architecture.
Cancel
Up 0 Down

Verify Answer

Cancel
Ashapiro

on Aug 28, 2024 3:27 PM in reply to Bubbles27
0
Regarding the SOA graphs, I'll refer you to this article about the Spirito Effect from the great Dan Eddleman.
https://www.analog.com/en/resources/technical-articles/mosfet-safe-operating-area-and-hot-swap-circuits.html

I'm skeptical because the SQJQ184E-T1_GE3 SOA curves are perfectly straight and in parallel.
There's no Spirito region, so I wouldn't trust its advertised SOA performance.
Cancel
Up +1 Down

Verify Answer

Cancel
Bubbles27
on Aug 29, 2024 12:28 AM in reply to Ashapiro
0
Thank you Ashapiro, this is really helpful. I will look into this article. Which MOSFET manufacturer would you say is trustworthy?
Cancel
Up 0 Down

Verify Answer

Cancel
Ashapiro

on Aug 29, 2024 3:22 PM in reply to Bubbles27
0
Nexperia
Alpha & Omega Semiconductor
Cancel
Up 0 Down

Verify Answer

Cancel