Need an Nchannel Power Fet w/ VGS down to 3V

[methods]

The market is flooded with options.
I would steer toward IR if possible.
I could use a pointer if someone has a fresh database.

HIGH LEVEL: I need to control a power fet with a voltage range of 3V to 4.2V so as to run a uController directly off of a 1S cell and not use a regulator. Without this - I must run a 5V regulator - and much power is lost to the regulator and uC running unnecessary high. I want no regulator in my circuit of any kind.

EDIT: Solution
LTC7000




Requirements:

N-Channel Power Fet
100V
VGS, fully on, down to 3V
Low on On resistance... I can parallel them... and they can get hot... say... below 10mOhms
Price - dont care. $1 or $10 same to me.
Package - dont care... tho I would prefer to be able to dump heat into a traditional heat sink as I want them to have a very high overhead
I have no experience with these aluminum impregnated PCB boards... but... maybe it is time.


Application: Bi-Directional power switch

100V max, down to any voltage even 0V
100A burst, 70A continuous melting connector max, 30-50A general operating from freezing to 120F ambiant
No power draw what so ever from switched lines
10AWG on 45A Andersons (I have tested these to 70A continuous... that is the failure point with enough 10awg heat sink)

1S Lithium powered micro controller running Android
Super low power design - awake only fractions of the time - running in Mhz
No regulator - running directly off of the 3V to 4.2V swing
Powered by coin cells or a long life pouch or a solar ....
Onboard super low power reference voltage @ 3V for temp measurement (blipped RTD's)

Smarts on board - serves as a general switching point for powering down (or up) any system

Set of parallel power mosfets for input and output
Direction agnostic full power in both directions
Slow turn on and turn off... by design and as a result of design... perhaps manage low IR pack spark
Pulsed VGS on a cap, will require tuning, as with a boot strap for a high side switch... uC sleeps and blips to keep the mosfets on

1 year target shelf life
1 month absolute fail fallback
In the event of a dead onboard battery - must fail OPEN

Battery can be charged externally or be a primary cell

Input active low (closure of a 10M pull up filtered)
Multiple inputs for LVC condition, HVC condition, User Input condition, with different behavior and hysteresis, different lockout conditions, etc. depending on input

I started this project back in the day.
Its time to close it out
Nothing on the market that does this... certainly nothing that does it well

I will share the source code and build files - I dont want to get into this business... I do want a Dog Gamn solution that I can use
I will build it, test it, tune it, and use it.
Anything else - you are on your own.

You help me, I help you.
Lets make it happen.

-methods

 

[methods]

No current measurement
Fuse built onto the board
100A
This is how we will handle short circuits

Yes... we could measure current... but I am just not that into it. It will start to suck power trying to be quick
A fuse is quick

For slow response... a thermal mass and an RTD onboard
It will open itself in the case of continuous abuse

So - Fast Fuse for abrupt abuse
Slow overheat protection for inch-inch abuse

Can be powered by any DC-DC, from 3V up to 6V - or any scavenging power source

-methods

 

[methods]

Cant run off of 2S due to uController limitations - 6V max
Would need to run a Reg for 2S
Reg would need to be reliable and low dropout down to 6V
Reg would need to output 5V... or at least 3.3V... to be compatible with a wide range of gate level fets

But... we could say to hell with power, run 2pcs 18650, and replace once a month
Treat it just like the flashlight on your bike
Run as large or as small of a source as you like
Possibly DC-DC it off the main pack

In that case the design becomes trivial and any number of fast acting behaviors could be included

It is my opinion that other than short circuit protection.... nothing needs to happen fast in a Battery Management system
3 seconds... that is how often I intend to act on the circuit.

This effort keeps coming to the front for me.
I wake up thinking about it...

-methods

 

[methods]

Yes I already built it once... but that was a one-way switch
Inferior

-methods

 

[methods]

This will meet my requirements
This is what I want

So tired of punting on this project.

Yes - super ugly drawing



I will leave "hooks" in the PCB to implement the low power version, the 1S version, the coin cell version
This will get the project moving tho

Extra pads for shunting off the gnarly switching noise - resistors, caps, TVS diodes, whatever needs to be sprinted on it.
I want to play with turning the switches on and off slow... never done that... always just slammed them on.
Only time I ever did PWM of high current was on BMS V2-4... where I was switching 125V 20A @ 3khz... it was fuglyugly - no inductor

This one may need a DIY inductor... been meaning to do this for a LONG TIME. Bought the magnet wire already.

Some big caps - fast and ceramic - handle all that noise
Missing a TVS on the 5V section - gonna need to be clamped hard - all sorts of parasitics are going to bugger things up

-methods

 

[methods]

Yea - so I came full circle

I need to drop my requirement for low power and low VGS - it is not a requirement at this time
This opens up 20 or 30 mosfet options which are superior
This drops price to the super low point

I can build this cheap
It will work

Potted in an aluminum heat sink - I will play with something other than epoxy this time.

I have a friend with a machine shop close by now... I can finally put some hours on a CNC machine.
Aluminum will be from scrap - every one will be a snow flake... with personality

Learning:
How to break out of a design freeze-up by adjusting requirements to get the ball rolling
How to make a DIY inductor to deal with inrush
How to pot with materials other than epoxy
How to machine a relative true area inside of a random ME
How to manage turn on and turn off using PWM with low power on the mind - never ran a cap on the gate of a fet before

HEY - BTW - a very rad dude gave me a 3D printer... so... NOW I CAN MAKE ADVANCED POTTING MOLDS

K - off to the Bay Area for a KickStarter BBQ

-methods

 

[methods]

Last Post...

Uh... yea... I finally cleared out my old EE warehouse
I have enough parts in hand to build this - like... now.

Its simple enough to bread board and start testing.
I will go back to the "old way" and just haywire something up.

Kids these days... have about given me a heart attack... trying to adjust to the concept of Laying Out a PCB FIRST instead of second.
Sigh.. I was taught to build everything haywire... then spin a PCB once proof of concept was done.

I have seen... with the drop in cost and lead time on PCB's... that it makes a ton of sense to skip the haywire phase

BUT I HAVE LEARNED THIS:

MANY DESIGNS DO NOT MAKE IT OUT OF THE MIND, OFF OF PAPER, AND INTO REALITY... WHEN COMMITMENT TO A PCB DESIGN IS IN THE CRITICAL PATH.
It just makes things... eh... too formal.

I am rolling it back to the old school and just building it.
Super Excited.

I will take pictures of the Home Brew method and post in a thread with links back to the original concept (the purple Arduino powered coin cell switch)
Fun fun fun

Remember - a lot of us are learning "how to do it"
It serves no purpose to show off at a high level if guys cant emulate you.
I am going to build this in a way that anyone could... even a high school student

Anyone is free to nab the design and use it for a project or whatever.
Anyone who wants the code... while have to pass my programming test :wink:
That AINT the ability to go to GitHub... that's an actual test of programming competence.

-methods

 

[Hugechainring]

I skimmed what you want to do, with admittedly only partial comprehension. Recently I was working on figuring out how to to force an Anker Power Core 10000mah pack with QC 3.0 technology to output 6.0VDC for a particular application.

What about using a Qualcomm Quick Charge 3.0 HVDCP Controller like ON Semi's NCP4371? Voltage input range is 2.2-28V, and output voltage can be commanded in I believe .2v steps starting at 3.6V. This isn't my area of expertise, but I think these arrangements don't use a regulator that you are trying to avoid.

On my side, what I want is a USB cable with a built in charge controller that I can plug into an Anker power pack, thus outputting a predetermined voltage other than the standard 5V.

 

[methods]

I skimmed what you want to do, with admittedly only partial comprehension. Recently I was working on figuring out how to to force an Anker Power Core 10000mah pack with QC 3.0 technology to output 6.0VDC for a particular application.

I assure you that your partial comprehension was the fault of the writer... as I admittedly am guilty of posting "mid thought".

What about using a Qualcomm Quick Charge 3.0 HVDCP Controller like ON Semi's NCP4371? Voltage input range is 2.2-28V, and output voltage can be commanded in I believe .2v steps starting at 3.6V. This isn't my area of expertise, but I think these arrangements don't use a regulator that you are trying to avoid.

I only skimmed the data sheets... and thanks for the link... I am shooting (myself in the foot) to be able to bring quiescent current in at under a uA in the final revision. I will read deeper into the sheets but I suspect with all that communication functionality there will be a price to pay in uA. My last BMS was running a few uA and the charge controller *IIRC* utilized a latching relay which drew 0uA (or effectively leakage currents on the PCB). I have been smashing my head against the wall trying to justify control loops slow enough to bring a "smart" (or programmable) controller in at under 1uA. The relaxed requirements in this thread are a "methods, get off your ass and start compiling some code with the ammeter hooked up" effort.

On my side, what I want is a USB cable with a built in charge controller that I can plug into an Anker power pack, thus outputting a predetermined voltage other than the standard 5V.

Thats no problem what so ever if I understand your requirement.
Assuming power levels under 15W (as the highest actual USB output I have seen around is 3A * 5V)
There are a TON of Buck/Boost converts available with variable output.

I have not personally fooled with whatever commands I would need to send up the USB pipe to unlock the higher current levels.. but my first attempt would be to just cut off a USB cable and draw down the 5V supply until I saw errors (per the old USB phone chargers). I would then make a "multi-adapter" for the output that matches whatever plugs I needed to use, and tune it case by case with either a DMM or an LED ladder.

Here are some units I have experience with.
The Boost only or Buck only have more power handling IIRC
The Buck/Boost models are a tradeoff... and of course we could get more power out of any of them.

All Electronics is like the Flea Market... Sometimes you run off with a KILLER DEAL and sometimes you get jacked and hate yourself.
DC-DC Boost for $8

I suspect that selecting a DC-DC with a FIXED input voltage of 5V will get you better efficiency, so better power handling, as any adjustment on the input equates to compromise in performance.

Here is a $3 Gamble....

But when I want something that works... I look at Meanwell via the TRC site.
With Meanwell you get only +/-10%... but... it does not blow up or burn a hole in your pocket.

To completely rewrite your requirements... eh... we can make you a portable power supply that "makes your wildest dreams come true" for a lot cheaper.
Reference materials here from Justin RE flight requirements for maximum battery size.
I have personally flown all sorts of sketch batteries... Just tell TSA that you "are the man" and that they "need to settle down" lol.

-methods

 

[methods]

So that USB charger deal... to me looks like:
(or at least how I would design it is)

Xpcs 18650 Cells in Series Parallel - optimized for the output voltage - which appears to be USB... so 2S XP
A USB Charge controller to charge the unit up when your laptop is wall powered
Perhaps a quick charge input for when you are in a hurry
USB bank of outputs... where mine would be capable of melting USB wires

I would just hook a 50W 5V supply directly out the USB door and see what happens.
I would hope that any modern device has the ability to manage its own charge rate.
I know I certainly would not trust a host device to manage my charge current... as the charging stations appear to do.

That kind of stuff must always be handled on board (thinking high power charging here) to maximize global compatibility.
Chargers should be dumb IMHO.
I may be proven wrong.

-methods

 

[methods]

So I was approached for a couple of small jobs.
The type of job that... could take 300 engineering hours... or 30
Since I am looking at taking one or two of the jobs on a flat fee - here is the path I would take for solving this riddle.


Drop Mosfet Switching, pick up gas filled contactor
https://www.waytekwire.com/item/770...HnZTVmy8eAId2TrgJsrt0yVvcMerVrsfgixoCf2fw_wcB

Here is the Data Sheet:
http://www.gigavac.com/sites/default/files/p115.pdf

Voltage: 0V to 1200V (yes... thats a real 1.2kV broken under load)

Coil power is only: 45mA @ 48V... so a couple watts

The P115 is the part number
F is for 48V coil
B is for 12V coil

12V coil is the most generally applicable.
48V coil will work from 30 - 64V... so 10S minimum and 15S maximum (calculated by lowest LVC vs highest HVC)

Hmmm... 10S to 15S covers most peoples needs
2 watts is three orders of magnitude less current draw than load... so if someone goes full throttle for a minute... thats the same as running the coil for 16 hours
Can anyone say... WE FINALLY FOUND A USE FOR THOSE CRAPPY 5W SOLAR PANELS?!?!

So... in the event that we go a direction like this, BMS would look like this:

Variant A
Variant B
Variant X

Variant A is the hardware design. Lowest part hardware triggers driving an open collector bus.
Expandable to any size pack
Arduino on the buss - drives a gate level fet to drive coils
Would need either a 12V source or a DC to DC... either way Arduino will need power
Arduino needs 2V to 5V.... or can inefficiently take 12V
No balancing
No data coming out
I really, really, really do not want to pursue this - but ultra low power.

Variant B is the LTC design.
Breaks up into 4S to 12S chunks
Balancing ... low current through the low pass filter resistors
Full data stack available via USB for PC or BlueTooth for Android
Data buss instead of oldschool buss
POWER HOG... So would need to be implemented with a deadman timer and bootstrap switch.

3 seconds... that is the number I am rolling.
A BMS needs only to respond once every 3 seconds

Prove me wrong - assuming there is a fast acting fuse to deal with short circuit and thermal probes to catch the rest.

-methods

 

[methods]

Setting of Requirements:

* Pack shall conform to the regulations I posted in another thread.
Variant A is eliminated as it can not be guaranteed that a channel has gone bad. Variant B comes with error detection to guarantee safety.

* Pack (which we are assuming is about 50V) shall be able to be charged from any charger - even a 100V charger - so long as the continuous current is less than the continuous charge current allowed of that cell combination or that which causes excessive heating (programmable by application in the "smart switch"). So - an "almost" retard proof charge system.

* It shall be impossible to over-discharge the pack - ether by direct loading or by self discharge (check). Any discharge in excess of C rating shall be detected by cell sag. Any short circuit or extreme discharge shall be detected by fast acting fuse (set to wire gauge).

* User shall be able to visually confirm that pack is in good working order. Either by LED light or by Android interface.

* Pack shall be IPXX rated - at user request

* Pack shall rest in a "sleeping state" which draws NO POWER - or leakage currents only. Say - 1uA

* Pack shall be brought online via a BootStrap switch and a Dead Man Timer shall be set. Heuristics for this will be tough without a Shunt or Current Sensor. A pack should know how much current it is dumping for data log reasons... but we often boot this job out to the CA or other. It requires an opamp or two to get it right with a shunt... so... a hall sensor... but those are expensive, big, and power hungry.

Side bar - Heuristics.
How to determine what state the battery is in without access to the discharge current.

States: Idle, Charging, Discharging

Suggestion 1) Watch the current dummy
Suggestion 2) Integrate cell voltage to extract direction. Cells not changing? Idle. Cells going up only? Charging. Cells going up and down? Discharging
Suggestion 2) User Switch to control dead man time... 1hr? 5hrs? bypass?

(Simple to create a user interface on a PC or Android to set such things in the Flash memory of an Arduino)

Ok... I feel it..
We are converging on a new design

I want to do this ONE MORE TIME and it better be right.

Requirements:

* Pack is to be built once and never opened again. It should not need a balance charger or any bullshit. It must be able to self balance.... LVC and HVC at the power level

* No fiddly bits - look to production. A single PCB shall be used to avoid nightmare assembly

* Balance taps shall be protected to their AWG via self resetting fuses at the terminal taps on a PCB

* Design shall scale from 4S 1Ah to 40S 400Ah (with population options on balancing as pack goes larger)

* Design shall withstand any noise thrown at it such that it works under the hood of a Tesla at full throttle.

* Design shall be quick and dirty - fast to market - using reasonable components - as I would build it.

* Design shall steal every good idea I have ever seen... and may I gouge out my eyes if I ever build anything inferior again out of fear of retribution from "people who own ideas".

-methods

 

[methods]

Ah... I found the mosfets that have a Vgs of less than 3V.





That part there will buy us 5 things.

1) We can now run our uController at 2V or 3V to save significant power
2) We can now PWM the fets HARD and FAST if we want to... and we will be buffered from the snap-back.
3) We have both fets running off one channel (at no time should they be out of sync) and high side to maintain ground integrity
4) We get current sensing ... with a fast response... so if I install an 80A capable shunt in there... we should be dialed.
5) We can now use any mosfet on the market - not limted to VGS that are logic level (and less noise immune)

Although we choose to keep Battery Ground isolated from Chassis Ground... Eh... If I can keep ground, ground, ground.. and switch high... great.
Ground switching leads to trouble when people get clever with their "wiring"
Most people understand and respect the high leg.

-methods

 

[methods]

And to get started... the Vanilla flavor.

Fewer parts = more likely to get implemented.


LTC7001

-methods