Best way to detect what power source is connected?

[KiwiSteve]

Hi,

I'm designing a circuit that can accept 3 sources of power:

• Solar
• DC
• AC

1 or more power sources could be connected at the same time. I've attached a simplified diagram of how I think the key parts should be arranged.

What I'm trying to figure out is the best way to determine what source is providing the most amount of power. Can I simply measure the voltage and current at a point before each ideal diode?

Thanks,
Steve

 

[Harald Kapp]

what source is providing the most amount of power.

With an ideal diode (using a MOSFET and controller), only one source will provide power. The control circuit will shit off those sources that have a lower voltage than the output. It should be easy to take the information which ideal diode is 'on' from the respective control circuits.
Which controller do you use for the ideal diode circuit?

 

[KiwiSteve]

Thanks Harold,

I'm fairly new to all this stuff and I'm learning as I go. I'm thinking of using the LTC4357 (http://cds.linear.com/docs/en/datasheet/4357fd.pdf) as the ideal diode. I haven't chosed the MOSFET part yet. I don't believe this diode controller part provides any intrinsic status, or perhaps I'm not understanding your comment "It should be easy to take the information which ideal diode is 'on' from the respective control circuits."

But, if I understand correctly, by simply reading the current before each diode I can determine which diode is (or isn't) active? I haven't breadboarded anything yet but I'm looking forward to trying things out soon.

Much appreciated!

Steve

 

[Harald Kapp]

The LTC4357 may not have an explicit 'status' output, but you can (ab)use the gate control signal as status signal. High gate voltage turns the MOSFET on, so it is an indicator of an active transistor.
You will need to decouple the gate voltage with a fairly high resistance (a few 100 kΩ, the higher the better) to ensure vaild gate drive to the MOSFET.
Use e.g. a resistive voltage divider from the gate to ground such that the output voltage of the divider can turn on a small MOSFET which in turn provides the current for e.g. an indicator LED.

Measuring the current is also an option, but more difficult than simply supervising the gate voltage.

 

[KiwiSteve]

Thanks! I hadn't thought of that approach. I'll definitely try that as my first option, it may also save on some parts

I very much appreciate the support and collective wisdom from this forum.

Steve

 

[KiwiSteve]

The LTC4357 may not have an explicit 'status' output, but you can (ab)use the gate control signal as status signal. High gate voltage turns the MOSFET on, so it is an indicator of an active transistor.
You will need to decouple the gate voltage with a fairly high resistance (a few 100 kΩ, the higher the better) to ensure vaild gate drive to the MOSFET.
Use e.g. a resistive voltage divider from the gate to ground such that the output voltage of the divider can turn on a small MOSFET which in turn provides the current for e.g. an indicator LED.

Measuring the current is also an option, but more difficult than simply supervising the gate voltage.

Harald,

You mentioned that this approach is an "ab(use)" of using the MOSFETs gate. I'm curious, would this be considered a hardware "hack"? Would this approach be frowned upon in a professional design? I plan to try it out soon, but I'm also wanting to learn and apply best practices. As I said, I'm just curious and wanting to learn.

Thanks,
Steve

 

[(*steve*)]

It's not abusing the gate, it's abusing the signal to the gate.

The gate drive signal is there to switch the MOSFET, not to power a load. However, assuming a fairly low impedance drive to the gate, it may work.

Edit: "hardware back:? Possibly.

 

[tedstruk]

multi-stack gate.
you might build a switched voltage detector, that fires relays depending on the power plant. start at the high input, if its there, shut off the others. If there are 2 inputs, fire a 2nd control circuit for the 2 circuits, and then if there are 3 inputs, relay on to the 3rd controller.

 

[KiwiSteve]

It's not abusing the gate, it's abusing the signal to the gate.

The gate drive signal is there to switch the MOSFET, not to power a load. However, assuming a fairly low impedance drive to the gate, it may work.

Edit: "hardware back:? Possibly.

Thanks Steve,

Yes, I see that it's taking advantage of the gate signal that was not intended to be used for another purpose, but could this approach still be considered acceptable? I'd be interested to see a schematic that made this "dual signal" idea work.

Steve

 

[(*steve*)]

There is no reason why you couldn't design a circuit to work in this manner. It's just in this case that it wasn't and the suggestion is made on the basis that the drive to the MOSFET gate is fairly "stiff".

We don't know what the gate drive in your circuit is like, so it may not work. If we knew how it was designed we could determine if this would work in your case or recommend any required changes.

 

[KiwiSteve]

My idea is to use the application notes and part references in http://cds.linear.com/docs/en/datasheet/4357fd.pdf to have a 3-way OR-ing ideal diode setup for a max 36V @ 10A. The MOSFET specified is the Fairchild FDB3632 (https://www.fairchildsemi.com/datasheets/FD/FDB3632.pdf). Eventually, I need to send a digital signal to a microprocessor to indicate which power source is providing the power hence the need to "read" the active MOSFET and provide this status.

Please forgive the newbie question, but when you say "the drive to the MOSFET gate is fairly "stiff"" do you mean that the voltage or current should remain steady as it drives the gate?

Based on the choice of driver and FET, how would you suggest I proceed? I'm hoping to order the parts from Digikey later this week and start breadboarding something.

Thanks,
Steve

 

[(*steve*)]

"stiff" means it can supply current without the voltage dropping.

In this case the chip you plan to use has a charge pump to provide gate drive to the MOSFET. That will be about as stiff as a piece of cooked spaghetti.

My initial thought was that you could probably use the gate signal to drive another n channel MOSFET which has its drain connected to the incoming voltage source, switching a LED to ground. However, this will probably not work for several reasons.

I'll see if I can come up with something.

P.s. you don't sound like a kiwi :-D

 

[Harald Kapp]

Based on the choice of driver and FET, how would you suggest I proceed? I'm hoping to order the parts from Digikey later this week and start breadboarding something.

There's an easy way: test it.
check the gate drive signal without additional load and compare it with a signal when a load (e.g. a resistive divider in the MΩ range is attached. You need to look at the switch signal (on/off) in particular.

 

[KiwiSteve]

"stiff" means it can supply current without the voltage dropping.

In this case the chip you plan to use has a charge pump to provide gate drive to the MOSFET. That will be about as stiff as a piece of cooked spaghetti.

My initial thought was that you could probably use the gate signal to drive another n channel MOSFET which has its drain connected to the incoming voltage source, switching a LED to ground. However, this will probably not work for several reasons.

I'll see if I can come up with something.

P.s. you don't sound like a kiwi :-D

Thanks again.

I'm actually a "hybrid" Kiwi. I moved to New Zealand from UK and became a naturalized citizen. I now live in USA.

 

[KiwiSteve]

There's an easy way: test it.
check the gate drive signal without additional load and compare it with a signal when a load (e.g. a resistive divider in the MΩ range is attached. You need to look at the switch signal (on/off) in particular.

Thanks. Yes, I'm fairly new to all this stuff (particularly analog electronics). I'll do as you say once the parts arrive.

Steve

 

[Bluejets]

What I'm trying to figure out is the best way to determine what source is providing the most amount of power. Can I simply measure the voltage and current at a point before each ideal diode?

I don't suppose 3 * hall effect ct's and some type of comparitor would be any help?

Such as Allegro ACS758..???
Tried to put pdf file but it said file was too large.

 

[tedstruk]

another way is to send all the power input to the same buss and control it all from there. (which is what you appear to want) Frankly I think its a bad idea, power busses always wind up melted unless they are loaded with breakers. Honest. don't fire the lot together without relays and controls. its just not experimental.
I would relay the input to a transformer that can handle it

 

[KiwiSteve]

I don't suppose 3 * hall effect ct's and some type of comparitor would be any help?

Such as Allegro ACS758..???
Tried to put pdf file but it said file was too large.

Thanks,

Yes, I think this an excellent idea, and in fact this was the direction I was going initially. If the MOSFET gate idea doesn't work I will probably fall back to this approach. My motivation is to reduce the number of parts and cost of the overall design. I like the Allegro parts very much and am quite familiar with them.

Steve

 

[KiwiSteve]

multi-stack gate.
you might build a switched voltage detector, that fires relays depending on the power plant. start at the high input, if its there, shut off the others. If there are 2 inputs, fire a 2nd control circuit for the 2 circuits, and then if there are 3 inputs, relay on to the 3rd controller.

Thanks for the suggestion. Honestly though I think this will consume too much board space and create unnecessary noise. Also, I'm concerned about the possibility of eventual mechanical failure, particularly since the application will be used in a high-stress (knocks, bumps, vibration) environment. For these reasons I'm probably going to avoid using relays.

Steve

 

[KiwiSteve]

another way is to send all the power input to the same buss and control it all from there. (which is what you appear to want) Frankly I think its a bad idea, power busses always wind up melted unless they are loaded with breakers. Honest. don't fire the lot together without relays and controls. its just not experimental.
I would relay the input to a transformer that can handle it

I'm planning on using resettable fuses on each input. Do you think another one of the main bus is also required?

Steve