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

[methods]

Ok - poked Totem Pole into Digikey and clicked on the first thing I saw

http://www.nxp.com/documents/data_sheet/PMD9003D.pdf

After a very brief look - it almost looks like I could just use that to drive the mosfet directly with a 9V source.... I have to re-learn all the details about BJT's as I have long forgotten that stuff. Any time I was ever forced to use one power consumption was never an issue... so I would just size base resistors by randomly grabbing a resistor out of the box and putting my finger on the BJT to see how hot it got

I did 5 years of ultra low power work but I always just focused on the uController aspect - never had to worry about the details outside of my little module... and that is pretty much childs play with an MSP430 or other super low power uController. Done a few years now playing with mosfets.... but never when power was a concern. If I needed a gate voltage I would just plop down an isolated 5V to 15V DC-DC 1W uModule Total problem solver... use that to power an isolated gate driver and the deal is done.

Now trying to mix ultra low power with very high power ( we are talking about 30uW controlling 7.5KW here...) I have tipped the scales

Just saying that gets me excited tho... a 30uW system that is *super smart* having wireless (battery) control over 7.5KW. Amazing.... and all I do is complain :x

-methods

 

[grindz145]

Damn 4.2 uA is awesome. I'm working on a project with a MSP430. I'm tempted to drop it and throw an Attiny in there and run Arduino code. I wish there was an open-sourced equivalent for the MSPs.

 

[methods]

Damn 4.2 uA is awesome. I'm working on a project with a MSP430. I'm tempted to drop it and throw an Attiny in there and run Arduino code. I wish there was an open-sourced equivalent for the MSPs.

That 4.2u is actually with the much more capable 328P. Running at 3V datasheet calls out 3.9uA - so I know I am doing something right. That is still with the WDT running - turn that guy off and we are talking about sub 1uA.

I programmed the MSP430 for a few years using Eclipse. You know what? I hated it. Sure it was extremely powerful... but back then they were the only ones that could do ultra low power, hardware multiplication, dynamic memory allocation, etc. Now days - pfffttt... even an Arduino can do hardware multiplication (Mega and up)

The ATtiny is pretty limited - both in program space and functionality... but for this deal I am really just picking it because it will be easier to inspect during assembly :wink: That 328P can run for near the same price, space, and power... but that pin density really limits the allowable slop in production. We are not even using reflow profiles at this point - just cooking in the oven and watching for reflow - eyeballing differences in part sizes.

Yea - Let me tell you....

I started programming 8086 in ASM... then C using a turbo compiler.... then PIC ASM.... then C++, Java, then up to LabView... I have done a significant amount of "hard work" with every kind of programming you can do... so it is not that I "need" to use an Arduino boot loader - it is more like I *appreciate* how f'ing amazing it is.

A boot loader that will allow you to write one piece of simple code that will work on like 20 different uControllers!
The ability to just snatch code snippets off the net and BAM - just run them - without even touching the code in any way
The joy of using Eagle and Arduino and sharing experiences with other hackers/wackers
The prospects of how quickly you can teach others (Steve at JozzTek picked up LabView and Arduino in only a few months)

Those who put their nose up at Arduino are most likely the kind of guys who get paid a salary by someone else :wink: As soon as a guy works for himself and blows 40 hours or 80 hours (like I just have) working hard to solve tiny stupid problems... all of a sudden the efficiency of Arduino just shines.

Not fast enough? Just crank that CPU up to 20MHZ
Not powerful enough? Yea - I doubt that - Try a Mega
Cant do low power? PHFFT - check out RocketScream - those guys have an awesome 8MHZ mini that does like 1uA sleep

Sorry for the rant - you got me all excited

Arduino is just the tip of the iceberg for collaborative open source work - I am really excited to see where we are at in 20 years. I will be a 60 year old programmer blabbing away then just like I am now - all excited - ready to roll.

-methods

 

[methods]

Man - buggers in the details
This one was good for around 100uA and logic level control - but then they went and hung a 100K on the output - which is another 90uA :x

http://www.onsemi.com/pub_link/Collateral/MC34152-D.PDF

-methods

 

[grindz145]

A boot loader that will allow you to write one piece of simple code that will work on like 20 different uControllers!
The ability to just snatch code snippets off the net and BAM - just run them - without even touching the code in any way
The joy of using Eagle and Arduino and sharing experiences with other hackers/wackers

That's exactly it. When I think about how many times people are re-writing uart code for micros everyday (for instance) it just seems ridiculous. Especially when you're doing it for yourself. It always takes forever to set up all the basic junk, and it's not rocket science, but it is a ton of details, pain-in-the-ass register searching and time counting.

 

[Njay]

Man - buggers in the details
This one was good for around 100uA and logic level control - but then they went and hung a 100K on the output - which is another 90uA :x

http://www.onsemi.com/pub_link/Collateral/MC34152-D.PDF

-methods

Sometimes you can turn things on/off with a micro pin directly or through a transistor.

 

[bearing]

I think you can use an inductor which saturates before reaching charge current. You only need it while switching on, and if it's big enough, the FET will be fully on before it saturates. When switching off, you don't want it, so it would be a good thing to have it saturated.

I'm not sure if you can buy an off the shelf inductor which saturates at a fraction of the charge current, yet has a low enough resistance to take the full current, though.

Nice find on that gate driver!
How often will it switch?

 

[zombiess]

Man - buggers in the details
This one was good for around 100uA and logic level control - but then they went and hung a 100K on the output - which is another 90uA :x

http://www.onsemi.com/pub_link/Collateral/MC34152-D.PDF

-methods

Sometimes you can turn things on/off with a micro pin directly or through a transistor.

I've done that before with a NPN transistor setup as an emitter follower with no resistor on the base, just ran it directly to my micro controllers output pin to switch the load (a relay). It's worked fine for me on a few projects, but I don't know what it's power consumption would look like since I've never tried to engineer anything as low as Methods is.

 

[grindz145]

I was just thinking a little bit more about using the Arduino code. I seem to remember something about the license mentioning something about non-commercial use. Are there any issues there?

 

[megacycle]

When this is finished we can all pay our charitable contributions to the cause

 

[grindz145]

When this is finished we can all pay our charitable contributions to the cause

The open source guys aren't looking for cash generally, if it's GPLd it's totally cool, but otherwise someone might just get up in arms.

The more I think about it, the chance is pretty low, especially when your project is totally open source the way Method's shit always is.

 

[fechter]

Life would be a lot easier with a 9v battery and 3v regulator. If the power draw is low enough, the processor could run off a cell tap with coin cells. A good low quiescent linear regulator would allow operation off a big 9v transistor battery. Those work great on my smoke detectors. Battery holders are dirt cheap too. Change once a year.

One other thing I'm not clear on; what is the output from the HVC/LVC device. Don't those need power to work or something?
A standard optocoupler output needs a pull resistor to work, more power consumption...

My favorite TPS2811 gate driver, the one that only takes 2uA standby, could possibly be driven on the input by a variation of that charge pumping scheme. The TPS2811 series needs a little over V/2 on the input, so 3v won't be enough. The input impedance is very very high, so it won't need much current to trigger.

 

[methods]

I was just thinking a little bit more about using the Arduino code. I seem to remember something about the license mentioning something about non-commercial use. Are there any issues there?

Hex code is hex code.

-methods

 

[methods]

They are darlington pair - so 5V, Open Collector, Ground to maximize sensitivity - or just short 5V and OC and use two pins like a regular opto.

Ah fechter... you poor, poor man... you dont get to enjoy the pleasures of microcontrollers

* 5V goes to a uController output
* Output goes to a uController input (or I just tie them together to save a wire)
* Ground goes to ground

* Turn on output
* Read input
* Turn off

Happens so fast that only a tiny blip of power is drawn

Another way is to use the internal pull-up resistor (turn it on and off) and just use two wires - GND and (Signal/5V) - so only two wires.

You do bring up a good point tho - if I want to go into deep hibernation and wake up on interrupt instead of polling then I will probably need to wire up a constant pull-up resistor that has a tiny leakage. I will tie 5V and Output together (and lose some sensitivity at ultra cold temps - temps below where lipo even works).

Otherwise - I could just wake up for a few uSeconds every second or two and that would be fine.

-methods

One other thing I'm not clear on; what is the output from the HVC/LVC device. Don't those need power to work or something?
A standard optocoupler output needs a pull resistor to work, more power consumption...

 

[methods]

I dont pay attention to any of that crap. As far as I am concerned no man can own an idea, a method, or a likeness. All of that is total bullshit made up by lawyers and jerks :wink: If anyone wants to steal anything I have done - and do it better - I say hell yea - do it - and I will do the same.

I can say this - there are at least 300 online retailers selling different Arduino variants and all of them are running Arduino - or some custom version of the boot loader.

-methods

When this is finished we can all pay our charitable contributions to the cause

The open source guys aren't looking for cash generally, if it's GPLd it's totally cool, but otherwise someone might just get up in arms.

The more I think about it, the chance is pretty low, especially when your project is totally open source the way Method's shit always is.

 

[fechter]

Ah fechter... you poor, poor man... you dont get to enjoy the pleasures of microcontrollers

* 5V goes to a uController output
* Output goes to a uController input (or I just tie them together to save a wire)
* Ground goes to ground

* Turn on output
* Read input
* Turn off

Happens so fast that only a tiny blip of power is drawn

OK, that sounds like it's covered. There are similar approaches that can be implemented using analog parts.

 

[methods]

OK, that sounds like it's covered. There are similar approaches that can be implemented using analog parts.

If ever there were words of a man in denial....

-methods

 

[grindz145]

There are similar approaches that can be implemented using analog parts.

But you can't necessarily fix them as quickly.

 

[etriker]

OK, that sounds like it's covered. There are similar approaches that can be implemented using analog parts.

If ever there were words of a man in denial....

-methods

They could still make analog tvs but they last too long. I have one in the bedroom from 1997.

The new tvs don't last nearly as long,

If it is one that hooks up online then the auto updates from Vizio can get messed up and kill your mainboard and Vizio may or may not sell you a new mainboard.

They might give you a discount on a new, improved Vizio if you yell at them enough for killing your tv with messed up online automatic updates ?

 

[methods]

Those are valid points you are making - but they are not logical arguments against the use of micro controllers or in favor of analog engineering. The true force between the negative results you point out is greed and engineering for profit.

They design those products to fail because they are in the business of selling new products, not creating the best product that they can.

So the real question is.... how do we bring back the spirit of the real analog engineer? Guys like Fechter who do it only because they want to help everyone else and do the best work they can? That is the question I want to ponder... how do we bring back the engineering spirit and leave the designed obsolescence behind.

-methods

They could still make analog tvs but they last too long. I have one in the bedroom from 1997.

The new tvs don't last nearly as long,

If it is one that hooks up online then the auto updates from Vizio can get messed up and kill your mainboard and Vizio may or may not sell you a new mainboard.

They might give you a discount on a new, improved Vizio if you yell at them enough for killing your tv with messed up online automatic updates ?

 

[grindz145]

how do we bring back the spirit of the real analog engineer? Guys like Fechter who do it only because they want to help everyone else and do the best work they can? That is the question I want to ponder... how do we bring back the engineering spirit and leave the designed obsolescence behind.

I think about this alot too. I think part of it, is that we have to make longevity cool again. In a culture where the 1 year old phone is getting old, keeping a bike (or bms) for 20 years is hard to even fathom, but to make alot of the investments in sustainable design take hold, I think bringing back longevity is important. That way you can really appreciate the design, because it's stood the test of time.

 

[methods]

The only way I can think of to do that (in this constantly evolving environment) is to start building things to be more modular. That way - Fechter can design an awesome mosfet switch that can interface to whatever is the latest and greatest for years and years.

The all in one solution (like the cell phone) is doomed to repeat the cycle.

Modularity is awesome.

-methods

 

[liveforphysics]

I like the 9V smoke detector battery idea too.

You can get non-rechargeable lithium ones that have a pretty amazingly high energy density.

http://www.amazon.com/Ultralife-Long-Life-Lithium-Battery/dp/B000167NYO

$7, 1.2Ah at 9V, 10 year shelf-life, and snaps into or out of a $0.25 standard clip when/if replacement is ever needed. Having 1.2Ah wouldn't let you run wild with your energy consumption, but I think it would make it a lot easier to make a product that could do a self-contained decade of use and then snap in a standard size format replacement available all over the world for under $10.

It's downside is volume and weight. If you want something super tiny, this thumb-sized brick is no finger-nail sized coin cell, but it's also not like we're making a medical implant, we're packaging it onto a bicycle so using a cubic inch of space doesn't seem entirely like a deal-breaker to me at least.

 

[circuit]

draws almost nothing, wayyy less than any contactor or bunch of expensive mosfets (including losses in heat). And really nothing when switched off.

 

[fechter]

draws almost nothing, wayyy less than any contactor or bunch of expensive mosfets (including losses in heat). And really nothing when switched off.

Yes, true and I like how the UVC is normally closed so any wiring faults will fail safe.
This arrangement still does not address the big spark you get when connecting the battery. If you never remove the battery, it's not such a big deal, but for those who remove the battery frequently, a precharge circuit would really help. It also keeps full pack voltage on the controller caps all the time. There is some thought this could lead to premature cap failure, but I'm not sure if that is true. For sure cap leakage could become an issue at elevated temperatures. It would also be better to disconnect the controller during charging to prevent damage from transients coming through the AC mains.