Zero-Draw, Solid State Contactor w/Precharge (Arduino)

Let me first introduce the concep - then below I will give definitions.



No more dead batteries from leaving the controller on
No concern for cell level LVC during discharge
No concern for fires from over charging cells

This will be a new component in my battery monitoring/protecting scheme. No longer will we need to tie into the throttle to terminate discharge - instead this device will be attached inline with the battery (anywhere) and will disconnect the battery from the system in the case of cell level LVC, HVC, or by the user with a mechanical switch

The device has two high current tabs. One should be connected to the battery (positive or negative, it does not matter). The other should be connected to the load (controller, charger, etc).

There are two other connections. One goes to a handle-bar mounted switch that allows the user to **completely** turn off their ebike. This means that no longer will the full battery voltage be sitting on the controller caps 24/7. This is important for many reasons including discharge through leakage current and general safety

The second connection goes over to my (or any) active low detection system.

So why would you want to use this and not just a regular old relay or a high power contactor? Here is a table of reasons.

* Power Usage
---------------------------------------------------------------------------------------------------------
A standard contactor or relay typically requires 12V @ 200mA (call it 3W) to remain in the active state. This is highly problematic for 2 reasons... first it is constantly draining the pack. For a typical 12S2P Lipo pack it would be 440Wh/3W = 6 days.. i.e. if you fully charge your pack and leave the relay on it will deplete your pack after sitting for just 6 days. This is similar to leaving your controller on and even I have done it, ruining entire 18S3P packs.

That is not even the bad part. The bad part is that if you want to create a contactor system good for any voltage... be it 12V or 200V... it can become very difficult to create the 12V 3W low voltage bias supply. It requires a complicated and expensive DC-DC converter that has a limited input range (so you need different versions). Yea - for over 50V folks can use a $1 AC-DC switching regulator but that comes with complications of its own. It requires obnoxious wiring, it adds numerous new failure modes, it draws power, it is bulky, and it just sucks.

A solid state switch built out of mosfets and controlled by a microcontroller draws less than 50uW of constant power. That is 0.00005W. That is in the range of the self discharge of a battery. So low in fact... that I can actually power the system from a pair of coin cells SUCH THAT NO POWER WHAT SO EVER IS DRAWN FROM THE SYSTEM BATTERY.

The device behaves just like a contactor would. There is a "power switch" on the handle bar that can turn the main system power on or off. When the switch is thrown to the ON position the contactor goes through a Pre-Charge (PWM) routine to ramp up the controller power (eliminating connector sparks...) and then goes into a "deep sleep" mode where it draws only microwatts and waits for either the user to signal that the system should be shut down (OFF) or a signal from the Battery Managment System stating that power should be cutt.


* Power Handling
---------------------------------------------------------------------------------------------------------
Contactors are rated for a certain range. The kind of contactor we would use for an ebike (basically the best available in its size range) is good for about 50A continuous and maybe 150A bursts. This is driven by several factors including the gauge of the wire hooked to the contactor - since the terminals are used for heat sinking. To increase power handling you need to either step up into a much larger and more expensive contactor or you need to use two in parallel. Even in quantity - the cheapest contactor available that does not suck is about $40 and that does not include the Precharge circuit (which is absolutely necessary since a bridge resistor is unacceptable for this application) and it does not include the DC-DC used to power it.

Solid state switches can handle much greater loads in much smaller packages (if done right). With 8pcs of TO-220 IRFB4110 we can easily handle 100V 100A continuous with no heat sink (using the wires as heat sinks). Mosfets can be had in 150V, 500V, whatever you like. One design can be populated with the proper mosfet configuration to work for nearly any system. Our baseline with be IRFB4115's - 8 of which will be around $16 in small quantities. This will afford the user a range of 0V to 150V and a constant current of 50A (with heavy cabling). Burst current - 300A easy. We are basically keeping things around or below 70C with minimal power loss.

Power loss through this switch would be around:

IRFB4110 100V 100A bike (10KW)
4mOhms/4pcs in parallel = 1mOhm
Two in series = 2mOhms
10A = 0.2W
50A = 5W ( 100V*100A=10kw, 5W is 0.05% loss, so that is like 99.95% efficient)
100A = 20W (will never happen - 60mph on bicycle for extended periods = crash)

IRFB4115 150V 50A bike (ultra high speed bike - burst to 100A no problem - so call it 15KW until you crash)
10mOhm /4pcs in parallel = 2.5mOhms
Two in series = 5mOhms
10A burns 0.5W
50A burns 12.5W (this is tough to handle thermally, but electrically it is a fraction of a percent of operating power)

I forgot to mention (and it is not in the pictures) that the system switch is in series with a 70C thermal switch
THIS MEANS THAT IT IS TEMPERATURE PROTECTED

View attachment 2


Physically it looks like the above picture. Notice that the tab of the mosfet is also the drain. This serves a triple purpose:
1) Mounting
2) Heat sinking
3) Power conduction

By using copper bars we kill three birds with one stone. The mosfets have a secure mounting location - no "beefed up" PCB boards or bullshit like that. Since copper is just about the most uber heat sink material you can use - it effectively moves heat out to the cabling where it can be dissipated. Yes... the cables are the heat sink. Finally - it makes for the perfect power conductor. No stubby wires to fool around with - just screw on your cable assembly

The mosfets are configred in common source. That means that the switch blocks in both directions. It also means that both the "left" and the "right" mosfets can be directly mounted to the copper bars. In the middle the sources are directly connected together (laying one atop the other) with some help from a very wide and thick PCB trace. Gates are routed (possibly to individual resistors), over to a common 6V zener, then directly to the 5V output of the uController



* Power Source
---------------------------------------------------------------------------------------------------------
This is probably the biggest win for this design. As stated above a contactor always needs a power source - to the tune of a few watts. Yea.. this may be fine if you have a 20KW 4 wheel EV... but if you want a design that can work with anything from a 4S accessory pack to a 200V 5Ah speedster EV then all that DC-DC business is for the birds.

My design takes its power from a pair of coin cells in series. NO POWER IS DRAWN FROM THE BICYCLE MAIN PACK.
No power.
None
Zilch

Coin cells come in a million flavors - common choices are about 50mAh to 200mAh. In that range, with my power draw, and typical usage (2 hours per day) my system will likely out-live the self discharge of the cells... so 5 to 10 years (at best). Worst possible case could be months - but that would be an edge case. I wont bore you with all the low power calculations - just suffice to say it will be like your watch.

How often do you worry about changing the battery in your watch?
That is how little you will worry about the batteries in this contactor

This opens up ALL SORTS of possibilities. Since the contactor takes no power what so ever from the load that it is switching - this means that it is totally independent of load so you can run anything

5V 1000A DC
50V 100A DC
200V 10A DC
Even AC signals!

4S ciggy lither packs
1S cell phone chargers

It does not matter. Although my system touches the sources of the mosfets - it is pretty much self contained.


* Precharging & Contact Welding Issues
---------------------------------------------------------------------------------------------------------

This is huge - possibly the biggest reason to use a mosefet switch
If you put a contactor in a system and have it between a high rate discharge battery (Lithium) and an extremely capacitive load (controller) then it is absolutely CRITICAL that a Pre-Charge system is used (or else the contacts will weld closed!). With Cars - folks usually just hang a 10W resistor right across the contactor. This really blows in my opinion - since it can allow mA of current to be trickled by. This is unacceptable for a small pack like an Ebike - and even more so for a small pack like a 4S5Ah cell 12V mobile Cell Phone Charger.

So... that means that if you want to use a Contactor you need an elaborate pre-charging scheme that can guarantee 100% precharge and do that in a non-obnoxious time. This means power, complexity, wiring, ugh. It means system interdependence. It means over-building for the guy who runs an Infineon with 4mF and under building for the guy that runs a SevCon. Problems Problems that I dont want

With a solid state switch there is no chance of contacts welding closed so all you really need to do is stop the silicone from blowing out on an insane inrush and stop the controller caps from blowing from the same inrush. This could be as simple as a short PWM run (which will cause wear and tear on the controller caps). With a uController in the system - and the ability to control rise times - it is an easy problem to solve.


* Driver circuit Range
---------------------------------------------------------------------------------------------------------

With any contactor system you basically have a limited range that you can work with. One size does not fit all.

With a self contained solid state switch - none of this need be worried about.


* Cost
---------------------------------------------------------------------------------------------------------

My solid state switch costs about 1/4th of what it would cost for a contactor and all the parts for a DC-DC and precharge system. This means lower costs for you and more development cash for me.


NOTES OF POSSIBLE ISSUES:

* It is preferred that the unit be attached to the GROUND legs of any battery so that there is no high voltage being routed up to the user switch. The switch only connects 3V, but since it is basically tied to the sources... if the sources sit at 200V then the switch sits at 206V with respect to battery ground. Care must be taken... this does not mean the switch has to be high voltage... just that it must be insulated well.

Now lets talk about how it works.

One of the coolest f'ing things is that we have an Arduino on the inside



Here is what you are looking at:
* 6V worth of coin cells running into a LDO ultra low quiescent current regulator (2uA & 300mV max)
* A mechanical handle bar switch that allows the user to turn the entire ebike system on and off (also preserving battery life)
* An Atmega 328P Arduino compatible controller (running a low power variant of Arduino)
* A PWM capable TTL output that drives the mosfet bank
* An Interupt capable digital input that comes from the BMS, Detector system, or whatever

The microprocessor sleeps like 99.999999% of the time. This is how you get currents down into the micro amps, or even nano amps. Starting from the off state:

User turns ON switch
Controller performs a Precharge
Controller latches mosfets to the ON state
Controller goes into deep sleep

User turns OFF switch
(there is some hidden complexity here to ensure that the mosfets turn off abruptly)
But basically the system just turns off and the breaker opens

If the system is off there is no need to monitor the BMS

IF the system is ON and a BMS active low signal comes in
It can either be an interrupt (which I dont like) or it can be polled on a 1 second loop. 1 second is nothing in discharge/charge terms - and it can be considered a hysteresis on the system. Either way it is totally programmable so for applications that need instant cut-out we can run interrupts. For other applications we will probably just run a WDT and check once a second

Anyhow - in the case of an LVC even the system blows open the relay and then goes into deep sleep (or can even self terminate with one more mosfet)

Some minutia...
The uController only pulses the open collector input from the BMS. Basically it wakes up, turns on its internal pull up, reads the line, turns off the pull up, and takes action. This dramatically lowers power. A protection diode will probably also be included in the final design.

There will probably be some TVS diodes sprinkled about.

bah... my frigging arms are getting tired. I cant type anymore. Look, Junior approves! (that is Lukes equipment in the background)




Ok - that is it in a nut shell. I am here looking for input from experienced designers. You have a part you want to point me to? You have a concern? Question? Basically I am showing my cards here... anyone can feel free to steal any part of this design - since as of now it is clearly open source. In return all I hope for is that one or two people can help me out. Every little bit helps.

-methods

It just occurred to me that the mosfets could be stood up vertically. The two bars would be vertical and back to back with an insulator - this would make for staggeringly simple connection to a PCB. Source pins go down through the board and fold toward each other - overlapping - and eliminating the need for a big stupid beefed up trace.

Hmmmm... yea...

Anyhow - there was a bunch of other stuff to talk about but I am tired and it is lunch time. There is a whole other aspect to this project that leverages the smarts of the embedded Arduino. There are many possibilities for temperature, current, voltage, time, etc monitoring. We have a companion board that has an RTC, microSD, A/D's, D/A's, Digital IO, Serial links, a screen, buttons, etc.

-methods

I had lunch. Back for more.
So one of the most difficult aspects of transitioning from a hobbiest to a designer/manufacturer is keeping in touch with what people *actually want and need*. Early on it is easy... just make what you want most for yourself and everyone else will like it. Steve and I did that for a few years.... but what happens to folks like us is that we get bogged down in the realities of running a business and we get further and further detached from "the scene". I see it in other active members of the community and I see it in myself.

So... what the hell will we use this thing for? Why would anyone want it? I am looking at it in the following way (and in this order - since first and foremost I have to be excited about something myself... to get myself to work on it)

1) I really, really, really want a solid state DC relay that I can use for my 4S accessory packs (charging phones, jumping cars, powering 12V electronics)

2) I need a replacement for my HVC Breaker (I have sold out that production run) and this is the next logical step. It fits perfectly into the Battery Protection scheme I have been offering/using and I really need something for this summer (For enthusiasts and for OEM's)

3) This will work as a stand alone item that "every" ebike could defiantly benefit from. Even ebikes that have a BMS in the battery or ebikes with a power switch on the box - it is still totally useful to have a stand alone switch to turn the entire system on and off.

* Handles Precharge issues - sparking connectors
* Provides a handle-bar switch for controllers that dont have a power switch (like Lyen)
* Provides a way to turn off accessories like lights - similar to the ignition in a car
* Creates compatibility for controllers that have a high leakage rate (discharge resistor 10K - 50K on the main caps... empty your pack in a week)
* Removes power from the controller for an added level of safety
* Provides a way for other DIY enthusiasts to get a jump on things - they can develop whatever they want and just hook into my active low input for control.

I guess in my mind every bike *needs* a CA, and every bike *could use* one of these.

I also have an agenda here -
I am a big Arduino advocate and this is a great way to get more people involved. You see... this device is completely programmable by the end user via freely available equipment and software. How awesome is that? If even one under-grad student uses this as part of his/her project I would be totally stoked.

anyway... I am going to go build an ebike. Summer is here in California and I have a 24S15Ah pack just sitting on the bench. If anyone can help I really am interested.

* Have a tip non a LDO regulator? (I am looking at microchip)
* Have experience with coin cells?
* Any tips on how to handle the precharge?
* Do you think my idea sucks and wont work? Why? Lets work the numbers on it.

Think you could do it better? How? Show me, and if your way is better I will totally steal your ideas

-methods

I like it, I think lots of people with higher power EV's could use one. It is easy to set up as a hidden kill switche or an alarm isolator too.

I'm trying to get my head around the common source FET arrangement. The Atmega328 seems like a pretty powerful chip for a switch !



But what stops me leaving the switch turned on, and the controller still draining the pack? (EDIT: I guess the Methods LVC boards, duh) Is a 12hr time out feasible, not that you want it shutting down unnecessarily during a ride? Or other method of sensing non-use?


Keep up the great work.

This would make high voltage a lot more useable..
want!

The ATmega328P can be powered from lower voltages, up to 10MHz at 2.7V. The regulator could be ditched out and only 1 coin cell used, but then there was the problem of turning the FETs on. One possibility would be to use a voltage doubler charge pump (2 diodes + 2 caps) controlled by the mega; a little more CPU usage but with the advantage of being able to have some control over the gate voltages, that you need to do the pre-charge - if enough control can be exerted to achieve a bounded known rampup time (real voltage can be peeked back with the ADC or analog comparator interrupt, but I don't know what's going on with FET transfer function and thresholds over the temperature range). Of course, one of the smaller AVRs would also do the job and be cheaper.

I've been playing with the same sort of switches, and using a voltage divider and relatively high gate resistor values to ease the charging of the power caps. I'm very glad to see you thinking about it too, and would love to see the arduino control implemented- even if overkill. It would open up the next step of control and interpretation of status signals at a very very low cost!

EPIC ... Subscribed for the win.

Will buy once this gets released

Very cool. Yes something a lot of people could use. Of course, I'd make it all analog.

On precharging, check out my circuit on page 5: http://endless-sphere.com/forums/viewtopic.php?f=3&t=40142&start=100

The non polar capacitor acts like an integrator and produces a perfectly linear voltage ramp. I=dV/dt x C, so cap charging is constant current. This keeps the FETs in the safe operating area during precharge. If you turn them on too fast, they can blow up as somebody found out. The linear voltage ramp worked out very well in testing. A one second charge time at 50V into 10,000uF works out to something like 500mA charging current.

The coin batteries are a clever solution to having different references during charge and discharge. This was always a problem for pack driven circuits. I have a design for an ultra low quiescent linear regulator that can handle high voltage. I think I could get it down under 10uA. This could replace the coin batteries.

I have about 4 pounds of 2mm 2"x14" nickle plated buss bar. I gave a few pounds to luke. They would make perfect busses for that board. If he has used it then I can send some down.

Excellent thread Methods,

Curiously i am exactly thinking about a wat toreplace a big EV-200 contactor that i hvae for my 2kWh LiFePo4 battery for my Zero DS 2011 range upgrade.

The battery is finished and i am at the point where i need to connect the Signalab 16s BMS to the battery but it's a stock 40A bms. the first idea i had was to use 4 car relay with coil in serie to get an equivalent 48V coul that would be driven by the BMS output and have the contact in parallel. I would het about 80A x 4 = 320A max and have 160mA of current to drive the relay that wold also be activated with a 12V relay that conect to the Zero 12V system so once the zero is ON, the relay let the 48V out of the BMS to go to the relay coil and activate them to conduct up to 320A. this desing is simple but will draw 170mAh every hour... for nothing... and these 1 cu inch car relay just dont enter in the battery case i designed...

Then i thought abot using my EV-200 that would be driven by a 48V to 12V dc-dc where the 48V is from the BMS output... but again the contactor is big and draw 1.9W for nothing... i could also add a 12v relay in serie that would be active only when the zero is ON, but it add some connections between the battery and the motorcycle.. and also the EV-200 is too big.. i would need to install it out of the battery wich is not esthetic.

Then i thought about adding some 4110 mosfet to the signalab bms and use it as is but with more parallel mosfet... i remember that i was able to drive like 20A continuous with minimal heat sink at 58V for only one mosfet...! so for 300A+ i would need 15 of them... and if my controller have regen i woul need to double that !

OK.. wait a minute... 300A is for burst and the average is more like 50-60A on my Zero.. at least... this is what i have recorded with my Cycle analyst... so this mean i could just install like 10 of them for having 100A cont and 400A burst? safely?

Methods, i like your idea.. but please come with something you can sell and i'll take one!

Doc

So funny how you and I seem to be on the same wavelength. I'm a PIC user, but mainly that's because what I have. I have a very similar design to what you have posted here, never tried high current through it yet though.

I'm more of a designer than a builder though, not a fan of assembly line work.

I had an idea for pre charging the caps by slowly bringing up the gate voltage and letting the MOSFET act as the resistor. Soft start. Feel free to borrow it. I don't know if I'd ever get around to really making this in production myself mainly because I value my time at a high cost and I don't know how much one of these would sell for. I'm only moving forward on my throttle interface for the Xie Chang controllers because it's a pet project of mine, don't really see it as making me much money.

Yes, well.... I can't see a point in all this since each controller already has the feature. Simply disconnect thin B+ wire via small key switch or anything and controller won't consume any power. And NO external power elements are needed. No sparks, no precharges. And you can use the same wire to cutoff power during LVC conditions. Plain and simple.
You guys always tend to overcomplicate things.

It was mentioned that the power to the controller was to be turned off completly to isolate the input caps, wont do that way.

I was thinking a while back of using an off the shelf solid state dc relay with a 32V and mA control signal and using a divider network across my 72V or dc-dc and put a precharge resistor in series with a cap which wired into the control circuit would hold the relay off for a few time constants, then hold it in with a few milliamps be a good estop circuit to if you switch the control line too, not strictly an estop as the relay could fail on, but still?

Why would one want to isolate the caps? They practically draw no current. IF they do, probably it is due to 10k resistor in parallel (just ditch it).

I'll throw a long term evolutionary architecture out there, for info and discussion... and it is the direction I am going.

Methy, sooner or later, I think you need to take total control of the controller. As in roll your own micro. That is so you can have source code control of the total "brain" To me the controller is the core. The BMS is a peripheral that communicates with the brain. The charger is a peripheral that communicates with the brain. The brain is always on... though it "sleeps a lot.

You have a safety disconnect... be it the "big red switch" or a lanyard that pulls out of a connector. But otherwise when the battery is connected to the controller and stays connected. You rolled your own LTC BMS so you have the capability to do firmware. I think you will eventually go this route.

I believe the battery "cell compartment" should be separate from the BMS. Modularity rules. You have already demonstrated, I believe, the ability to communicate with the BMS. Now to master the charger. This month I played around with an off line 5V 1 amp cell phone charger. I modified it to provide 7.3 volts and current limiting to about 0.6 amps to charge 2 A123 cells I used to power my photography strobes. Switch mode, off line power supplies can be mastered and integrated into a "holistic Controller/BMS/Charger" integrated suite. The trouble is the UL certification, once we start modding them.

I think the only switching FETs that should be in the power path are the H bridge. Charger switches and outside the controller contactors are stopgaps to the "ultimate solution."

That said, I think this evolutionary step is great! and I don't want to derail your thread on a brilliant switch with blathering about a future... Integrating the coin cells to drive the gate is pure, unadulterated brilliance, and I salute you! This is a GREAT addition for existing hardware.

Since you are always thinking... even when you are asleep, I wanted to throw something out there to fill any dead time in your "noggin" that you used to have. BTW the munchkin looks great, nice and healthy and bright eyes!

This is absolutely a missing link that most casual hobbyists have been missing for a long time.

What's cool, is that since the FETs are never linear and rarely switched, you don't need nearly as many as you would in a single leg of the bridge of the controller.

You could even program overcurrent protection right into the module itself, and autoprotect for overtemp etc, even without a feedback loop from a CA/BMS.

Since you are using a micro you could also pulse the gates to bring the charge in the onboard controller caps up slowly. This is pretty trivial to implement as I'm sure you realize.

I like this idea. I need a high current version though, 300A continuous, 80V. If I was to match the contactor, which has a specified contact resistance of 0.2mohm, it would take 40 IRFB4110.

zombiess said:

Since you are using a micro you could also pulse the gates to bring the charge in the onboard controller caps up slowly. This is pretty trivial to implement as I'm sure you realize.

Click to expand...

Isn't his current implementation, which seems to be a PWM ramp, just that?

bearing said:

I like this idea. I need a high current version though, 300A continuous, 80V. If I was to match the contactor, which has a specified contact resistance of 0.2mohm, it would take 40 IRFB4110.

zombiess said:

Since you are using a micro you could also pulse the gates to bring the charge in the onboard controller caps up slowly. This is pretty trivial to implement as I'm sure you realize.

Click to expand...

Isn't his current implementation, which seems to be a PWM ramp, just that?

Click to expand...

Yeah this is an interesting idea, Lineartech makes a chip just for this purposes (pwm the gate drive to limit inrush). It might be tricky to not blow up under a bunch of different controller scenarios though.

bearing said:

I like this idea. I need a high current version though, 300A continuous, 80V. If I was to match the contactor, which has a specified contact resistance of 0.2mohm, it would take 40 IRFB4110.

zombiess said:

Since you are using a micro you could also pulse the gates to bring the charge in the onboard controller caps up slowly. This is pretty trivial to implement as I'm sure you realize.

Click to expand...

Isn't his current implementation, which seems to be a PWM ramp, just that?

Click to expand...

I missed that he wrote that until you just pointed it out, had to go back and re-read it.

circuit said:

Yes, well.... I can't see a point in all this since each controller already has the feature. You mean one of the features?

Maybe that is because you did not read my original post. This solid state contactor is to replace my HVC Breaker, i.e. Charge breaker, i.e. a device that terminates the charge current when any cell in the pack goes over 4.25V. Tell me again exactly how you intend to accomplish this?

The fact that this device can also block discharge current is a benifit, not the core goal.

Simply disconnect thin B+ wire via small key switch or anything and controller won't consume any power. And NO external power elements are needed. No sparks, no precharges.

Oversimplification. So you wont have any sparks when you have to service your battery and re-connect to the controller? You wont need to precharge eh? hmmm... yea...

And you can use the same wire to cutoff power during LVC conditions. Plain and simple.

Yea - plain and simple until you actually implement it yourself instead of over-simplifying a problem and jumping to conclusions that do not meet the stated requirements

You guys always tend to overcomplicate things.

Click to expand...

Or is it that guys like you tend to bluster in and shoot from the hip? All of us have of course considered leveraging the ability to turn off the regulator in the controller - but that is just a tiny part of the problem. It is very ebike specific and does not address any of the other applications that could benefit from a high power solid state DC switch.

I dont want to discourage others from contributing by jumping on you - but your input was pretty much belligerent and not too helpful.

-methods

circuit said:

Why would one want to isolate the caps? They practically draw no current. IF they do, probably it is due to 10k resistor in parallel (just ditch it).

Click to expand...

Because having power on the controller rail means that there is power on the mosfets means that there is power only 1mm of silicone away from the motor. The only thing between a 20KW machine exploding into violent action and your fingers is a few traces and a few bits of code. Maybe you have not seen a powerful EV go out of control - but once you have - you will realize that true safety can not be obtained until the power is removed from the controller.

Bikes ? Yea - ok - most peoples bikes are toys

Motorcycles? EV's? These are a totally different class of power and danger.

-methods

Gregory said:

Is a 12hr time out feasible, not that you want it shutting down unnecessarily during a ride? Or other method of sensing non-use?

Click to expand...

Good idea

-methods

I appreciate your input. I did consider running at 3.3V and boosting to drive the fets but it adds a lot of complexity and power. A charge pump only works if you keep pumping it... which means cycles, which means power right?

I will think of it some more.

I also considered some of the monitoring options - but steered away from that for electrolysis issues. The fewer high voltage lines that come in the better.

-methods

Njay said:

The ATmega328P can be powered from lower voltages, up to 10MHz at 2.7V. The regulator could be ditched out and only 1 coin cell used, but then there was the problem of turning the FETs on. One possibility would be to use a voltage doubler charge pump (2 diodes + 2 caps) controlled by the mega; a little more CPU usage but with the advantage of being able to have some control over the gate voltages, that you need to do the pre-charge - if enough control can be exerted to achieve a bounded known rampup time (real voltage can be peeked back with the ADC or analog comparator interrupt, but I don't know what's going on with FET transfer function and thresholds over the temperature range). Of course, one of the smaller AVRs would also do the job and be cheaper.

Click to expand...