Soft start circuit problem

Question

I am currently working on a soft-start circuit. The picture below is my schematic. It worked fine in simulation and even in a real circuit. But after several times, turn on and off the power source (My power DC source: V=24V, 290W), the MOSFET dies for no apparent reason (Vgs = Vds = Vgd = 0 V). I even touched my finger on top of the MOSFET and it was just warm the whole time.

I use Chroma 63204 Programmable DC Electronic Load in CR mode. I don't know why the mosfet died so quickly. Any idea would be appreciated!

I forgot to take photos of my scope screen but when I measured Vg,Vs,Vd by scope, it looked the same as the simulation result

Answer 1

Consider this important diagram from the data sheet when running in linear mode: -

Say you set a gate voltage that produced a drain current of 5 amps. At a 25°C ambient, the device would start warming straight away. Say it warmed to 75°C, it would then be taking 8 amps. Do you see where this ends up?

This is a poorly documented MOSFET given that it doesn't tell you what happens at 150 °C (absolute maximum junction temperature) but, you can put money on the fact that that it would be trying to draw hundreds of amps if it could. Of course it can't because it will burn internally.

This is the main problem of linear applications in MOSFETs that are not intended for linear applications. It's called the Spirito effect by the way and is a well-documented phenomena of running MOSFETs in linear applications.

In most cases, the thermal runaway can happen in less than 1 ms and not hardly produce any heat to the touch. Choose a better MOSFET that is designed for linear applications.

Other stack exchange Q/A on this matter: -

• Spirito effect #1

• Spirito effect #2

• Spirito effect #3

• The lie they told me about MOSFETs when I was at college

• Burning a power MOSFET

Answer 2

First observation, the hat1072h is or is about to get discontinued.

You do not state the current and what timings is the soft-start at which you test your circuit, but supposedly at around 5A? (R6 4.8ohm load).

If we take a mid-state point, 24V source, and the MOS limiting the current, loading 2.5A so it gives about 12V drop across the MOS, at 2.5A (ignoring C2).

Looking at the Maximum safe area of the hat1072h we can see that at 12Vds the maximum allowed current is about 2-3A, and 1A at Vds 20V, which seems dangerously close.

At the 19V Vds, 5V on the load at 4.9ohm it's about 1A, which is also right at the limits.

I've been warned by "old experienced analog guys" to not trust the DC characteristics given for MOSFETs on datasheets unless they are specifically designed for linear mode. Thus when operating in the linear mode, I take 2x -3x margin over the specs. Take this advice as is.

You might not feel the MOS heating up because it perhaps fails before the package gets hot.