I'm building a product which is powered by PoE/PoE+ and has an output protected by a TPS25947 eFuse.

I need to be able to switch the current limit on this eFuse between 500mA and 1A (at 12V) based on the PoE class that was negotiated. I already have an output indicating the PoE type, I just need a way to switch the current limit on the eFuse.

The current limit is controlled using a resistor from ILM to ground - it appears the way the current limit works is that it "mirrors" the output current onto the ILM pin at a specific ratio (182uA/A) and if the voltage exceeds an internal threshold (~0.6V?) it trips the over-current protection.

I'd need to switch between a 3.3k (3334 / 1A ~= 3.3k) and 6.8k (3334 / 0.5A ~= 6.8k) resistor between ILM and ground.

Is an analog switch my best option? I'd need a 12V switch as ILIM can almost reach 12V (I've seen ~10V on my scope) for a few microseconds if the output of the eFuse is shorted (which is something I want to protect against).

 

Is an analog switch my best option?

Using transistors.

 

I'd need to switch between a 3.3k (3334 / 1A ~= 3.3k) and 6.8k (3334 / 0.5A ~= 6.8k) resistor between ILM and ground.

Is an analog switch my best option?

You could use a small N-MOSFET, such as a 2N7000, BS170, etc.
Connect a 6.65kΩ resistor from ILM to ground, and a 6.65k resistor from ILM to the MOSFET drain with the MOSFET source to ground.
Thus with the MOSFET OFF the resistance will be 6.65kΩ, and with the MOSFET ON the equivalent parallel resistance will be 3.33kΩ.

Alternately you could connect the two 3.32KΩ resistors in series from ILM to ground, with the MOSFET drain connected to the junction of the two resistors, and the MOSFET source to ground.
Now the resistance will be 6.64kΩ when the MOSFET is OFF and 3.32kΩ when the MOSFET is ON.

See no reason to prefer one configuration over the other.

 

Version 2 using transistors.

 

Version 2 using transistors.
View attachment 348212

I prefer using a MOSFET for the transistor which looks like a low-value pure resistor when ON, since It's not clear how much a BJT's inherent saturation on-voltage will affect the current-limit point.

 

Bench test shows saturation voltages of 0.004 for a BS170 and 0.012 for a 2N3904 with a 3.3K load at 12 volts.

 

I did originally consider using a MOSFET switching a parallel resistor and even started drawing it into the schematic, but I thought the 0.6V Vds would be an issue for some reason.

I think I saw "Vgs = Vds" in the datasheet without realizing it was the test parameters used to find Vgs(th).

Thank you both! I think I am going to go with the parallel resistor and MOSFET (probably using an AO3400 which is overkill but I am already using in my design).

This simulation shows it nicely: https://www.falstad.com/circuit/cir...yg0WmIhSlU0iKUjYAGcPl8xLhLHdrCAOABDAA2YIYbCAA

It has a 182uA source (simulating 1A through the eFuse) - when the MOSFET is off the current limit is 500mA (shown by the 1.208V reading on the voltmeter - double the 0.6V overcurrent threshold) and when it's on the current limit is 1A (shown by the 607mV reading).

 

Thank you both! I think I am going to go with the parallel resistor and MOSFET (probably using an AO3400 which is overkill but I am already using in my design).

I would change the 10K resistor in series with the switch to 100 ohms.
Improves the Rdson a little better.

 

Other wise the voltage at the gate with a 5 volt source is only 2.5 volts

Actually, the AO3400 is a logic-level type, so has an on-resistance of only 52mΩ max at a Vgs of 2.5V.

 

I know that but might as well use the most of the 5 volt signal IMHO.
I had edited my post earlier reflecting that change.

 

Yeah not sure why I used a 10k there - I've been using a 200 ohm gate resistor in all of my other designs using this MOSFET!

 

You only need one switch. Put the 6800 ohm resistor in full time, and then for the increased current limit connect a second resistor in parallel with it. Much simpler, fewer pats, and no change-over glitches. To control the current accurately you will need at least 1% tolerance resistors.