Automotive power distribution controller

[Smouldering Iron]

Greetings all. This is my first post on your fine forum. To be honest it is a little intimidating considering there are such switched on people (pun intended!) here, but I think I will be OK seeing you seem pretty nice to noobs like me.

A couple of mates and myself are into all things automotive, interested in electronics and micro controllers, and wanted a project that would be a learning experience. Please note that “interest” has no bearing on “knowledge” or “competence” 

What we’ve decided to have a crack at is a solid state power distributor and controller for automotive applications. “What the heck is that?”, I hear. Well it basically replaces the fuse box with micro controller controlled electronics. It takes inputs from switches and sensors and distributes power to the appropriate circuits based on those inputs and programmed logic. For example, press the horn button, it activates the horn circuit. Flick on the indicator (turn signal), and it flashes your indicator (at a rate you can specify in the software). A more complicated example, when the coolant temperature is over x degrees (user defined once again) turn on the radiator fan. That is the basic operation, once you start pulling in CAN data from the engine ECU etc it opens up a lot of possibilities, but at that point we’re starting to get ahead of ourselves.

There are several example of the same thing currently being sold commercially, and I understand that automotive manufacturers are starting to use the same sort of technology too. Some of the motorsport and military products are:

http://www.racepaksmartwire.com/Smartwire_Home.aspx
http://www.motec.com.au/aboutpowerdistribution/powerdistributionoverview/
http://www.piedrafita.com/sistemas/en_UK/soluciones/electronica.php

We started with some arbitrary goals. Total current handling of 120A, 12 inputs, 30 outputs. If we can get a prototype to do that then we’ll look at adding stuff like a bus for additional inputs, sensor input via CAN etc etc.

We’re using Arduino as the controller. We don’t have the experience to know whether this is the best option, but we’ll see how we go.Initially we’re looking at the output circuit. We need to design a circuit with the appropriate components that will handle the current and heat load. Once we have something that works and have done some testing with it we’ll look at how to monitor the current and feed that information back to the controller.

Looking forward to your comments & critique.

cheers,

Damo.

[UnaClocker]

Seems like there was a kickstarter for a device like this a year or two back. It didn't meet it's goals, too niche of a project.
Anyways, sounds like a Body Control Module to me. Good goal. I'd use an ARM M0, to get the nice low power draw you need in an automotive application (ie, when the car is off, it's not going to drain the battery). You should roll an alarm system into the unit while you're at it. Alarm logic is really really simple.

[Corporate666]

Is this a commercial device?

Arduino is definitely not a good option, IMO, for many reasons.

[Smouldering Iron]

Is this a commercial device?

Nope

Arduino is definitely not a good option, IMO, for many reasons.

I was pretty sure this would be the case and that there are technically better options available. In Arduino's defence it's cheap and accessible to a lot of people. Taking that in consideration is there something else you think we should be looking at as an option?

[Skimask]

Automotive manufacturers aren't starting to use these sorts of things.
They've been using things like these for a few decades now.
Arduino?  Ya...No....
Where's the redundancy?
Where's the safety?
Name one person that's going to trust an Arduino to turn on a cooling fan at the right temperature...vs. having an overheated engine go up in smoke due to a bit flip.
Nope....

[Smouldering Iron]

Seems like there was a kickstarter for a device like this a year or two back. It didn't meet it's goals, too niche of a project.
Anyways, sounds like a Body Control Module to me. Good goal. I'd use an ARM M0, to get the nice low power draw you need in an automotive application (ie, when the car is off, it's not going to drain the battery).

The ARM has some beneficial features by the looks of it, thanks for the suggestion

You should roll an alarm system into the unit while you're at it. Alarm logic is really really simple.

That's actually one application I hadn't considered, and like you say it wouldn't be too hard to do at all.

[max_torque]

Well, like all good projects, it needs splitting into several managable chunks that you can start separately before you "pull it all together" !!


1) High current circuitry, protection, and safety

This concerns switching, measuring, distributing the vehicle battery voltage.  What power switches are you going to use (HINT: google "Infineon PROFET".....), what current can you source, how are you going to measure that current. What about fusing, over voltage protection (google "automotive load dump"  What is the systems thermal limitation, with all circuits loaded, is it realistic to load all circuits at once (hint, check alternator output!), how are you going to heat sink the device, what are the ambeint temps the system will operate in.  Will you "derate" in higher ambients. How will you carry the high current internally? thick copper pcb, busbars, descrete wiring etc  And finally, the hardest one, what IO connectors are you going to use?  (<<< that is a LOT harder than you might think in low volume, unless you want to spend $$$ on each connector

2) Enclosure, housing, mounting, environmental protection etc:  What will the device be housed in, will it be waterproof, what about vibration, condensation, etc?

3) Control and communication:  Starting with an simple ardiuno is actually a good idea.  prove the concept first, work out your control strategies.  Consider the safety case, but don't get stuck on it for now (assuming you aren't actually going to sell these things yet!).  Use Google to brush up on your CAN knowledge, so you understand how a typical automotive CAN bus is operated.  How many "hardware" inputs will you need, what about "outputs" etc.  Get a spec sheet together defining your project, so you can keep track of those decisions and issues around them (like delegating uC pins to functions, will a 48pin micro be enough etc?)

4) High level architecture:  How many power outputs do i need?  What current do i need to be able to measure, or act on, both  peak and continuous.  (Hint, best option is to split your power distribution into two systems, imo, a master, with say 6 to 8 high current outputs, that then feeds power "buses" to slaves that then actually controls the end consumers, as this architecture minimises vehicle wiring)



Answer some of these^^ questions and get back to us!  (Good project btw, but actually pretty complex and covers a lot of knowledge from more than just basic electronics etc ;-)

[max_torque]

Name one person that's going to trust an Arduino to turn on a cooling fan at the right temperature...vs. having an overheated engine go up in smoke due to a bit flip.

This^^^ is not a "safety case" as such.  Sure, it ends up with a failed engine, but it would not be considered critical (especially as the EMS system would go into "limp" if the fans fail to cool the engine. ie there is already a "catch all" second line of defence.


However, the all power going off, whilst you are driving at 70 mph on the highway IS a safety case, which is difficult to mitigate against.............


(also, looking at the manufacturers reliability data, there is little obvious difference between an 8b AVR and a 32b ARM in terms of hardware reliability. And, the simpler 8b device is likely to have simpler code, and hence less like to have a critical bug in it!  IE critical system safety is less about the hardware and more about how you implement and use it...........

[max_torque]

And for maybe a bit of inspiration, here's a CAN controlled slave output module i did earlier:





8 switching output channels with a "protected" 4amp continuous (8amp pk for 10 sec) rating (over voltage, undervoltage, over temp, no load detection etc, current feedback for each channel)

1 synchronously rectified PWM output, 10A continuous, 20A peak (10sec)


This^^ is what i called a "corner" module.  It sits on the end of one of the high level power busses (4 wire bus, CAN H & L, KL15, switched KL30) and is used to run lighting and other end consumers in the vehicle

[nukie]

Hi you are up against big players like Fujitsu, Hitachi, Bosch, Siemens are you up for it? Make sure your bank account is loaded and have a team of lawyers ready because big guns are coming your way when you start making money from your product.

You can look up patents and spec sheets from these manufactures this will help you with your directions. Car ECU requires a lot of testing, make sure you minimise the risk.

[Monkeh]

Hi you are up against big players like Fujitsu, Hitachi, Bosch, Siemens are you up for it? Make sure your bank account is loaded and have a team of lawyers ready because big guns are coming your way when you start making money from your product.

Is this a commercial device?

Nope

[max_torque]

Hi you are up against big players like Fujitsu, Hitachi, Bosch, Siemens are you up for it? Make sure your bank account is loaded and have a team of lawyers ready because big guns are coming your way when you start making money from your product.

No, the big guns have much better things to do, like sell millions of their products to the automotive OEMs!

No one  gets to be a Tier1 automotive supplier without years of investment and process/quality assurance.


That's not what the OP is on about.  They want to make a small "hobby" power output controller, plenty of people already do this, but the one thing you will notice is that all of the existing devices are very expensive considering what they actually contain.  That should give you a big clue as to the niche/low volume nature of the market for these devices!  For example, Motec (a well established name in aftermarket engine management) sell there power distribution models for well over $1500, and yet they contain around $50 worth of electronics.  Now either they are making a LOT of money (they aren't ;-) or the volumes are so low, that in order to even get close to re-couping their investment and development costs, they must ramp up the price dramatically.

One thing that is quiet clear is that with the exception of the better, much more expensive PDMs, people have massively skimped on the IO connectors, simply because an automotive rated 30A per pin IP67 sealled connector system IS hugely expensive!  The better units use the "autosport" series of MIL spec connectors, with gold plated pins, and if you want a shock, go and check out how much each of these is to buy, and if you want a REAL shock, check out the price of the tooling (crimp tool etc.........)

[TheEnd]

You'll be fine.

The market for these will be race and rally cars. I've worked on plenty in the past, and stripping out the old loom and thinning it down is a big part of it.
Everything is done for light weight and simplicity, and quite a popular style is to move the main fusebox to the glovebox area, and replace the fuses with resetable breakers where you can just dab a button and hope you get a circuit back on line.
eg, say on a rally car you manage to smash a light, and the fuse blows take out everything.
Hit the reset, and hope the remaining lights come back on.

Some of the other ideas in some of the links are good, an add on fan module with temp switches is a good idea, frequently OEM fans can be removed and manual ones set up, so a fan controller built in is nice.
Central locking and electric windows are almost always binned unless it's a complex system with frameless windows and hard to replace with plastic.
Drivers for heated screens is smart too, something that is often done is an aftermarket front heated screen is fitted, and the interior heater fans are junked, and a timed controller to run an output for one of those PWM in a rapid defrost and then a lower heat level would work.

It's certainly not rocket science, it is something that is being marketed so there is a niche, and if you target it right, ie for stripped down cars, I can see it being popular.

Currently what people do is strip the interior out, follow as many wires as they can, and start snipping out individual wires one by one. Sunroofs, rear radio antennas, airbag equipment, it all gets traced and binned, but a replacement body and light module would make things far easier.

[rx8pilot]

I have spent the past year designing a family of power distribution and monitoring products. Not for automotive use, but very similar - 12v nominal voltage + high current + multiple battery inputs. It is all MCU controlled and monitored with 3 independent MCUs (AVR). There are multiple data busses for monitoring, user control, and graphical display. All solid state circuit protection - no fuses or breakers.

The goals seem to be similar. For my application, the biggest concern was being fault tolerant. Over/under voltage, reverse polarity, capacitive loads,  noise immunity, high-temp operation, vibration are all a reality. Failure or even small glitches in the system can be very expensive (in my case anyway).

I will say that I was surprised at the real complexity of the challenge. Obviously not impossible, but there are tons of details that surprised me along the way. I had only dealt with small delicate electronics previously. The brutal nature of managing high-current loads that are hard to predict caught me off guard. Having a FET that turns big loads on/off is easy enough. The hard part is monitoring voltage and current during the switch and determining if it is shorted or just a big capacitor. What I ended up with was very delicate mV range signals in close proximity to very high current components that are also close to a microcontroller. Not really rocket science maybe, but I thought it be be a total piece of cake. I was wrong. To get a very high-reliability circuit that could deal with a long list of bad things took some time.

I started by working out the circuit protection for each channel. This part of the circuit is all hardwired and does not depend on the MCU for circuit protection directly. The MCU monitors it's status and uses that info to decide what to do next. I have two versions of the circuit - one type of output is a fixed current while the other is MCU programmable. Both types are standalone and will function even if the MCU is dead. One of the challenges is to determine how much inrush current you can allow before the circuit shuts down. If you act too fast, any transient will shut the circuit off. If you act too slow, things will smoke (like your output FETs). In an automotive environment, you will be connecting outputs to all sorts of things that you do not control, making it hard to decide how balance the parameters. In my case, I spent some tedious time calculating the point at which the FETs in my circuit would be compromised and dialed it back from there. My gear is not supporting life in any way - but my customers sure feel like it does since there is so much money involved in my application when something goes bad.

How far along have you gotten on this project? You had asked for critique but provided nothing to critique - except for a concept/idea.

[ajb]

As others have alluded but not directly stated, don't rely on the MCU for low level protection.  It takes an incredible amount of care and testing to make a truly robust, let alone truly safe microcontroller based control system.  Even if you're only controlling non-essential accessories to start with, it makes sense to do your low level protection in hardware rather than software.  Remember that a semiconductors switch like a FET can be destroyed in a fraction of the time it would take a fuse to blow or a breaker to trip.  It's probably worth the extra few bucks to use protected switch devices that have overcurrent/overtemperature protection built in.

The nice thing about the Arduino hardware is that they're not just Arduinos, they're also AVR/SAM dev boards.  You can start with the Arduino environment to do some proof of concept and basic hardware testing, and then when you're ready switch to Atmel Studio and work with the 'bare silicon' as it were.  I don't think that's a bad way to go.  Speaking of microcontrollers, read up on watchdogs.

[max_torque]

TOP TIP: Don't spend ages "rolling your own" output protection and monitoring, just use some that's been designed EXACTLY for this job, and is already in use in several million cars worldwide today. Companies like Infineon have spend thousands of man hrs of R&D and countless sums of money making just this kind of thing:

http://www.infineon.com/dgdl/Introduction+to+PROFET%E2%84%A2.pdf?folderId=db3a30431400ef68011421b54e2e0564&fileId=db3a304332ae7b090132b527d9173083

For example, take apart a fuel pump pwm controller, supplied by a global Teir1 supply to companies like VW, BMW, Mercedes etc, and made in the millions, and what do we find?




Oh, look, a fully protected, high side, half bridge driver, just sat their, doing all the "hard stuff" for you in one, easy to use package!

[max_torque]

TOP TIP #2:   If you are really serious about getting into automotive electronics, buy and teardown as much vehicle electronics as you can!  Ebay is a cheap source of s/h modules of all descriptions, often selling for tiny amounts of money (especially for the ones that never go wrong and so have no s/h market value!).

I just went and found my "Box of vehicle electronics" and this is what i got when i tipped it out onto the floor:


Plenty of learning lessons in all that ;-)

If you do this, you can effectively leverage all that learning, experience, and expensive R&D effort put in by the top automotive systems engineers.  If you try to understand how these devices work and are designed/made, you quickly start to see the same things.  Similar components (load dump TVSs, smart switches, Freescale microcontrollers, high input voltage, reverse polarity protected voltage regulators, CAN chokes, CAN transcievers etc etc etc), similar layouts, similar bus bar-ing, and similar safety processors.
   You could spend the rest of your life going through datahsheets trying to chose you own components, without any real idea if they are suitable in the real world for your application.  By tearing down and inspecting and questioning already existing devices you get a massive "head start"!

So, as people round here are want to say "Don't turn it on, take it APART!"  ;-)

[Precipice]

Then you find the ideal Infineon device, which isn't available to mere mortals... Same for On. Aaargh!
So you use a Tier-2 part from ST, which turns out to be crap and full of lies. Double-Aaargh

But yes, use automotive parts for automotive products, or life becomes intolerable fast.
(Well, you can hack stuff together, but transistors are a lot cheaper and smaller when they're integrated).

[max_torque]

Then you find the ideal Infineon device, which isn't available to mere mortals... Same for On. Aaargh!
So you use a Tier-2 part from ST, which turns out to be crap and full of lies. Double-Aaargh

To some extent, for the more specifically automotive Infineon devices, then yes, limited low volume availability can be an issue, but most of their ProFET range is available because of it's industrial usage. They are NOT cheap in low volume, but that's a different issue ;-)

[batteksystem]

In one of my project several years ago, we use IR3313, quite expensive but very good performance for constant loading.