looking for a dual and/or triple channel ideal diode (/ power ORing) controller

Hello Frederic,

LM74700-Q1 is linear regulation controller which completely blocks reverse current from one channel to another.

There wont be any cross conduction from one power supply to another power supply.

If two power supplies voltages are different then one power supply with higher voltage will supply the full load current.

If the two voltage rails are close enough, then both will share the load current. (no cross conduction).

But I believe based on the implementation of "break before make" scheme, you are asking about power MUX or a priority power MUX.

This would require two back-back connected N-Channel MOSFETs in each of the 3 power supply paths.

We recommend to use LM74800-Q1 which support power MUX for 25A applications using external N-FETs.

This is discussed in more detail in the application note (Design 6, Figure 15) : https://www.ti.com/lit/an/slvaes2/slvaes2.pdf?ts=1594921629574

There are other "break before make" implementation that is discussed in:

1. 

2. 

The above application note talks about using eFuse or Load switches to implement power MUX.

The same concept is reused in the priority power MUX implementation in Design 6 of https://www.ti.com/lit/an/slvaes2/slvaes2.pdf?ts=1594921629574

Let us know if you require further support on the power MUX implementation.

Regards,

Kari.

Hello Kari,

Thanks for your feedback.

I agree: To control the voltage of the MOSFET gate, LM74700-Q1 has a linear regulation.

LM74700-Q1 has also 2 fast comparators:   fast turn ON  + fast turn OFF

The problem is that the linear regulation is quite slow (/low current) and my MOSFET is a high current MOSFET.

Therefore, it takes either

- a significant amount of time    (linear regulation + "high" gate capacitance)

or

- a significant amount of reverse current (fast comparator => -11mV / 2mOhm = -5.5A)     (-11mV is the typical threshold for fast turn OFF) to turn OFF the MOSFET.

During this time there a a "cross conduction current" flowing from one "ideal diode" to the other "deal diode".

My application cannot accept this behavior.

Do you agree with my analysis?

Do you have a solution for this?

Ideally, I would like to avoid back to back MOSFETs (and shut resistors) because:

#1    the function needed is only "ideal diode"

#2    2 MOSFET = more space + more voltage drop + more heat

I will also look at the documents you have provided.

Thanks

Frederic

Hi,

For slow change in power supply, both the linear regulation and reverse comparator act together to reduce the peak reverse current.

The peak amplitude and time for which the reverse current flows depends on how the two power supplies vary with each other.

For slow changes in power supply voltage rails with another supply, the peak reverse current is not -11mV/2mohm (-5.5A). When the difference in power supply changes, the linear ORing scheme is also reducing the gate voltage, this leads to increase in RdsON and RdsON increases as power supply difference increases. Note that while the RdsON reduces, power supply sitll provides current in forward direction. Thus eventually the RdsON is increased high enough to block the reverse current completely. 

But if the inputs are changing faster, then maximum peak reverse current is -11mv/rdsON for 1us, because linear regulation will not react and only the reverse comparator would react quickly within 1us to turn off the MOSFET.

We have other options for ORing where the reverse current threshold is programmable. TPS2412 is a linear ORing controller with maximum rating 16V and suitable for high current applications. Reverse current can be programmed as close as -2mV or -1.5mV. 

Note that TPS2412 doesnot support reverse polarity input protection. Let us know if you require reverse polarity protection, as we have other options like LM74800-Q1 which has -4.5mV reverse comparator threshold and supports reverse polarity protection.

Regards

Kari.

Hello Kari,

To answer your questions, I will try to complete the requirements:

- 2 and 3 channels ideal diode / power ORing
- VDC max = +14.5V
- IDC max = 25A
- N channel MOSFETs (preferred)
- fast turn ON (storage capacitor for the charge pump)
- 105 or 125degC operating temperature       85degC (acceptable??????)
- reverse input voltage protection (ideally)
- no back to back MOSFET (preferred)

- no/minimum transient reverse current 

(Transient reverse current causes serious problems in my application.

For example, if you inject current in the output of a synchronous boost, the boost will apply its output voltage to its input

thus causing serious problems

)

About "But if the inputs are changing faster, then maximum peak reverse current is -11mv/rdsON for 1us, because linear regulation will not react and only the reverse comparator would react quickly within 1us to turn off the MOSFET."


I agree with everything you said:

"slow" input voltage transitions are not a big problem

"fast" input voltage transitions are not a big problem

The problem is "medium" speed input voltage transitions.

Both TPS2412 and LM74800-Q1 seems good suggestions to decrease the transient reverse current (cross conduction between the 2 ideal diodes).

However, there seems to be a risk of having the fast turn OFF threshold closer to zero. The risk is instabilities (due to noise or other things).

Also, I was thinking that a dual (/tripple) ideal diode controller could completely eliminate the transient reverse current

by making sure that only one MOSFET is ON  => break before make

Could you comment on this?

Thanks,

Frederic

Hello Frederic,

Thank you for sharing more about the application use case, this helps us to get back to you quickly.

Medium transitions such cold crank are 12V/10ms or less than that. During such scenario LM74800-Q1 works effectively to prevent reverse current even under severe cold crank situations. But if have a specific frequency range or ramp rate, we can check that out as well.

I agree having two dual ideal diode controller could eliminate the transient reverse current as both try to regulate the gate voltage, RdsON on each of the MOSFETs would increase resulting in overall increase in RdsON and leading complete block of reverse current.

But this would add to power dissipation due to two series MOSFETs.

Regards,

Kari.

Hello Kari,

About

"

I agree having two dual ideal diode controller could eliminate the transient reverse current as both try to regulate the gate voltage, RdsON on each of the MOSFETs would increase resulting in overall increase in RdsON and leading complete block of reverse current.

But this would add to power dissipation due to two series MOSFETs.

"

I was actually thinking about 1 controller for 2 ideal diodes => 1 MOSFET per channel

in order to eliminate the transient reverse current.

but it seems that TI does not have a solution like this.

Is this correct?

Thanks,

Frederic

Hello Kari,

Thanks for the confirmation.

Do you agree that a dual (ideal diode) controller would (or at least could) completely eliminate (or at least strongly improve) the transient reverse current (/ cross conduction from one diode to the other diode) ?

Thanks,

Frederic