Hi everyone,

I'm pretty new to all of this so I apologize if this is a silly question, but I'm having trouble in picking a FET. When you're kind of new to electronics its a bit overwhelming. The basic idea is that I want to use two FETs and an ATtiny to make an uninterruptible power supply. I want it to run in the 12 volt range with two inputs and one output. The load should be around 100 watts (give or take). I want to use resistor dividers and the analog digital converter on the ATtiny to watch the power supplies. Then use logic lines to control the FET gates.

I have been looking at the Texas Instruments CSD16325Q5C
http://www.digikey.com/product-detail/en/CSD16325Q5C/296-25645-1-ND/2202571

I think they should work but that might just be a newbie fantasy. I'm hoping someone with some experience using FETs can let me know if I'm on the right track or not.

Attached is a rough schematic of what I'm thinking of doing

thanks in advance

 

If you reversed the MOSFETs, the internal body diodes would eliminate the need for the lower pair of diodes. At least I think so. As drawn, they will conduct through the body diode regardless of the gate status.

I think that TI choice looks fine. I'm assuming you don't need (or want) high frequency switching?

 

looks ok, the fet should run right around 150mW when the load is ~10A. the part case though is a "odd" case, will you be able to get that soldered in?

the gate charge looks small (20nC), which is good. i suspect the IC can handle that gate charge? are the gates being driven at any specific frequency, or this a simple fail-of-line-power cutover switch?

i would add a pull-down resistor on each gate to help drain off any charge when going to the off side.

others can correct me as needed......

 

wayneh
The frequency should hopefully be pretty low. In the area of minutes or hours.
I wasn't sure if feeding power into the drain for the FET that should be off would be a problem. That's why I threw in some diodes.

DC_Kid
I have access to a hot air station so soldering shouldn't be a problem.
I don't know anything about gate charges, but I just want a simple on-off type operation and hopefully it doesn't switch very often.
One FET should always be on and one should always off. like you said just a fail switch
adding pull downs is probably a good idea.

Thanks guys.

 

the gate charge is a capacitor. when you touch say a 9v battery to a empty 220uF cap, what happens? current rushes into the cap w/ current being limited by inherent factors. the IC needs to be able to handle that gate charge, etc. the fet you listed looks like a low gate charge, but that alone is half the puzzle, you still need to make sure the IC can handle that, etc. you may even want to put a 10 or 100ohm between IC and gate, and then maybe a 5k or something from gate to gnd for the pull down, etc.

 

DC_Kid
So if I follow correctly the 10/100 ohm resistor would be an inrush current limiter so I don't try to pull too much from the controller pin when turning on the gate?

Sorry for the mspaint schematic but I'm at work

 

An N-MOSFET will not work properly for your requirements. To switch plus 12V you need to use P-MOSFETs. You connect the source to the +12V and the drain is the output. To prevent crosstalk from the selected device to the unselected device you can use a diode as shown or, if you want minimum voltage drop, two P-MOSFETs connected back-to-back (source to source).

You need 12V to drive the gates so you need an NPN buffer transistor (2N2222 or similar) with 10k from the collector to the +12V supply (collector also going to the gates) and 50k from the base to the ATtiny outputs.

 

Assuming you correct the MOSFETs and driver so you can control a 12V source from a 5V drive...

OK, so one side is on and active. Then it fails bringing the output low. Some time later the ATiny notices this, then some time later the ATiny changes it's output drive, and some time later the MOSFET turns on so some time later the 12V output is restored.

If you have a primary supply (Line Power) that is higher in voltage then the secondary supply (intermittent power) all you need is two diodes to keep them separate and always driving the load, and it doesn't have a changeover glitch.

If the voltages are not just so, extra diodes can be used.

 

DC_Kid
So if I follow correctly the 10/100 ohm resistor would be an inrush current limiter so I don't try to pull too much from the controller pin when turning on the gate?

Sorry for the mspaint schematic but I'm at work

thats the idea, but crutschow is right, nfet you wedge load between Vcc and drain with source to low side, or gnd in this case.

use pfet and then gate driver uses low to turn fet on, thus its a pull up resistor to 5v to keep the fet off, but IC will need to handle all or the gate charge, etc.

just find p version of your fet. in post #6, not sure why you have +12v on both sides??

use p like this (you can use a small dual nfet IC for the switch if your IC can only be used as a high(on) signal, etc.

 

crutschow
Thanks for the heads up.

Here is another FET with roughly similar properties but P channel.
http://www.digikey.com/product-detail/en/CSD25401Q3/296-24260-1-ND/2038330

You mention needing to use a transistor to run the gate off the supply voltage, but on the data sheet for the fist FET it says it was optimized for 5v gate drive, and on this one I just posted it shows an Rds of 8mOhm at 4.5 Vgs. I was hoping that would mean not needing any extra bits. Not that adding a transistor is a problem I just thought it would keep it simple.


ErnieM
I guess I should explain a bit more. The idea is to use this for things like solar or wind power or sources that aren't entirely reliable. But I would like it to switch over to line power when then panels or turbine or whatever fall below a set point. Then kick back over to them when they come up again. I wouldn't put anything critical on this but maybe some lights or pumps or things that wouldn't be too fussy about losing power for a moment if for some reason it took the controller a while to catch on.
I was just hoping for more of a "one-or-the other" type set up instead of load sharing between them.

 

.................
You mention needing to use a transistor to run the gate off the supply voltage, but on the data sheet for the fist FET it says it was optimized for 5v gate drive, and on this one I just posted it shows an Rds of 8mOhm at 4.5 Vgs. I was hoping that would mean not needing any extra bits. Not that adding a transistor is a problem I just thought it would keep it simple.
........................

That minus 5V gate voltage is Vgs (gate to source voltage), so with +12V on the source, the gate has to go to 12V (Vgs = 0V for P-MOSFET OFF) to +7V or less (Vgs= -5V or greater) to turn the P-MOSFET ON. Since the ATtiny output can't go to +12V, an easy way to do that is to add an NPN buffer.

 

i would add some filtering around the 5v reg and use that for the tiny and fet pull-up, etc.

the basics of the circuit is there, but as you will find out even basic circuits needs some extras here and there, like some basic pre and post filtering on the 5v reg so the IC has some stable/clean voltage to run on (i built a circuit for a module that was run on automotive power, and the 14.4v in automotive is not a battery, etc). as another example, maybe a self healing inline fuse to the 5v reg, because as-is, if the reg decides to short then you'll be replacing more than just the reg, etc.

so, get the basics of IC, fet, and gate driver down on paper, and then folks here can help in suggesting some improvements, etc.

and as crutschow said, buffer with a N device, and if you wanted to isolate the power fet from the logic level side, use a opto IC as the buffer, etc.

 

An N-MOSFET will not work properly for your requirements. To switch plus 12V you need to use P-MOSFETs.

Forehead slap! I was so focused on the orientation and diodes that I didn't even notice they were sitting on the top side of the load. Forest for the trees.

+1 for the simpler arrangement proposed by ErnieM. Simpler is better. You don't need a computer to OR together supplies.

 

Thanks everyone, I'm glad I asked before just trying to build the circuit.


crutschow
Thanks for all your help.


DC_Kid
I will add some filtering and safety in there I was just trying to keep the diagram simple while talking about how to drive the FETs, thanks


wayneh
don't worry about the forest for the trees this newbie was in the desert of ignorance thinking he had a good idea.
Also I'm hoping to make it a bit smarter than just Oring. Status output, current monitoring, data logging, don't entirely know yet (cost and my own ignorance are a factor) so that's why I wanted a controller. And I might even have to move away from the tiny series and use a mega if I'm going to add bells and whistles.


I'm attaching a schematic of how I think crutschow was explaining things. Hopefully he or someone else can look it over and let me know if I'm heading in the right direction for driving a p channel this time.


Still plan on using the TI – CSD25401Q3 P channel FET
http://www.digikey.com/product-detail/en/CSD25401Q3/296-24260-1-ND/2038330


and Crutschow mentioned the P2N2222 transistor
http://www.digikey.com/product-detail/en/P2N2222AG/P2N2222AGOS-ND/920244

 

........................
I'm attaching a schematic of how I think crutschow was explaining things. Hopefully he or someone else can look it over and let me know if I'm heading in the right direction for driving a p channel this time.

Still plan on using the TI – CSD25401Q3 P channel FET
http://www.digikey.com/product-detail/en/CSD25401Q3/296-24260-1-ND/2038330
.....................

That's the circuit I suggested.

That MOSFET is a soon-to-be obsolete unleaded surface-mount device which is designed for flow soldering on a PCB. I expect you want a standard leaded device instead.

 

... I'm hoping to make it a bit smarter than just Oring. Status output, current monitoring, data logging, don't entirely know yet...

Fair enough. Depending on your needs, you might like to look at data acquisition using something such as a LabJack or DATAQ. These are devices that attach to your computer via USB and allow all sorts of communication to peripheral sensors and controllers (outputs as well as inputs). What you've described so far could be easily accomplished with such a thing.

 

Hmm I noticed that last night too. I might try this one instead then.
http://www.digikey.com/product-detail/en/SI7145DP-T1-GE3/SI7145DP-T1-GE3CT-ND/2442037

As for soldering I hadn't given any thought to problems using lead solder on lead-free pin plating. A quick googling says its a great way to cause cracks in your joints. But the data sheet for this FET says the pins are just exposed copper with no plating.

I have a hot air station so surface mounting isn't an issue, but you are right I am using leaded Kester ep256 paste.

 

Hmm I noticed that last night too. I might try this one instead then.
http://www.digikey.com/product-detail/en/SI7145DP-T1-GE3/SI7145DP-T1-GE3CT-ND/2442037
.............................
I have a hot air station so surface mounting isn't an issue, but you are right I am using leaded Kester ep256 paste.

So you are going directly to a PCB without doing a breadboard?

 

So you are going directly to a PCB without doing a breadboard?

Ill probably put the FET on a breakout board and breadboard the rest. That should keep things pretty simple.
I guess I could dead bug it but a tiny break out would only be $1-2 for a set of 3.

 

if you want to conserve power, then use a fet for the gate driver.