OR-ing controller with priority

You have a diode-OR configuration, so the highest supply will be passed to output, regardless of GATE voltage (due to NFET body diode).
If you want to pass a lower voltage supply, the higher voltage supply needs a prioritizor config (back-to-back NFETs) to block body diode conduction when GATE-SOURCE is low.


A resistor divider on the SHDN# pin won't be very precise. There's too wide of a tolerance on its threshold.
The controller for the prioritizor should be a LTC4373, since it has a dedicated UV pin with tight tolerance.


Hope this helps,
-Aaron

Hi Aaron,

Thank you for the design explanations. I'll simulate the schematic. Should I recalculate the resistor dividers for 48V primary and 50V backup or leave them as is?

Nick

Adjust the divider values for your 48V application. 
The application circuit I posted was for a 28V system.

Hi Aaron,

I tried to simulate the schematic with some changed resistor values, but stuck in the nowhere and I don't understand where I'm going wrong. Please have a look at the sim file and advice what is wrong.

Thanks in advance.

Nick

LTC4373+LTC4372.zip

Hi Aaron,

I tried the simulation, but stuch with impossibility to resolve the issue with blocking the backup battery.

Please have a look at the sim picture ( I didn't find a way to send you the sim file ) and tell me what is wrong.

I didn't find a way to attach the sim file. Please advice how to do that in the future.

Thanks in advance

Nick

Hi Aaron, LTC4372 sim model doesn't work. I took it from ADI website. Whatever I try, the output is ON, regardles of the SHDN level.

From my first post:

Ashapiro said:

You have a diode-OR configuration, so the highest supply will be passed to output, regardless of GATE voltage (due to NFET body diode).

It doesn't matter what the controller does, the NFET body diode will conduct in the forward direction if Vin is higher than Vout.
What the ideal diode controller does is lower the forward voltage drop from 700mV to 30mV, which results in power savings.

Hi Ashapiro,

I modified the circuit, but found that I have about 3ms between switching from BAT to Primary and v.v.

Please advice is it normal and if not, how to fix it? The plot is in the zip file.

BR

Nick

LTC4373-4372-oring.zip

LTC4373+LTC4372.asc

Hi Ashapiro,

I modified the circuit, but found that I have about 3ms between switching from BAT to Primary and v.v.

Please advice is it normal and if not, how to fix it? The plot is in the zip file.

BR

Nick

LTC4373-4372-oring.zip

LTC4373+LTC4372.asc

Nick,
First thing first: you still have your voltage divider on the LTC4372 SHDN# pin instead of the LTC4373 pin :/
That aside, you have two 50V sources and a 50A load, with 10uF of output cap.
How realistic is this situation?
My rule-of-thumb is 20A of current handling per FET.

The heavy load causes Vout to droop during switchover between supplies. 
The quick fix is to add Cout to hold up Vout during switchover.
Since you have a 50A load, you would need mF of Cout to prevent droop. This isn't realistic, since you also want fast switching, and this would place a lot of SOA stress on the FETs.

At these current levels, avoid back-to-back FETs since you are doubling R_DS(ON). The most practical solution is to use a diode-OR config (single FET). You would get fast switchover without the SOA stress.

Hi Ashapiro,

Some explanations:

1. The two sources are different and they must have a priority - Battery ( 50.4V fully charged ) and Tethered Power supply ( 48V ). The priority power must be from the tethered power. So, when Tethered Power is connected, the system must disconnect the battery, to preserve its power for emergency reasons;

2. I know the NFETs are weak, and in the real PCB I'll use IAUT300N10S5N015ATMA1.

3. Today I simulated with similar NFET parameters with 1000uF capacitor, and the voltage drop was about 10V during switching, which is acceptable.

4. The duration of this voltage drop is still ~3ms. I think it is also acceptable.

5. I can't use single NFETs at least for the battery, because the body diode will conduct till the battery voltage is higher than the tethered power.

BR

Nick

P.S. Some simulations from today. They are with NFETs in parallel and 330uF, 1 Ohm load.

LTC4373+LTC4372-26-Nov-2024.zip

Hi Ashapiro,

Any updates?

Your schematic for ref:

I'm mainly concerned about the Safe Operating Area (SOA) of the switch FET.
SOA does not reliable share share when paralleling FETs, so you are limited to the SOA of a single FET.

One way to ease SOA requirements is to enable the load AFTER the output cap is charged.
You'll need a part with a Power Good pin. Look into the LT4363 and LTC4364.
It's the pin is called ENOUT, and it signals the downstream load to turn on when Vout is high.

50V/50A prioritizer is very demanding. If you try to switch too fast, you'll stress out the switch FETs with high inrush into Cout. If you go too slow, your output voltage will droop.
The easiest way to get fast performance and avoid SOA issues is to use single NFET (diode-OR config), and make sure the battery/backup supply is always lower than the main supply.

I reworked the schematic. Attached is the sim file + jpg.

Thoughts?

LTC4364+LTC4364-LTC2965-02-Dec-2024.asc