Car PSU proposal -- need design critique

[SirNickity]

That's a great idea.  I'm not really sure the "best" way to detect engine status.  My initial thinking is a zener that will break down above battery voltage and under charging voltage.  This could turn on a FET to switch the regulated 5v on a separate "engine running" pin.

Another option might be to use an accessory input, conveniently stepped down to 5v.  This will trigger anytime the accessory power is on, though -- not necessarily when the engine is running.  That may or may not be what someone would want.

[graynomad]

Load dump is something that should be considered, > 100V I think. Can the TVS etc handle that?

That said I've had a digital speedo connected directly to the battery with little but a 7805 for years.

______
Rob

[spirilis]

I'm hoping the capacitor in the filter will help take some of the hit from a load dump, but even still that regulator I'm using can run continuously up to 36V so it may have more leeway than many devices.


My way to acknowledge engine on/off (actually Accessory on/off) is to wire the Switched-12V line from the car stereo to the gate of a small-signal N-channel MOSFET, which itself drives the gate of a P-channel MOSFET which connects 5V to the "sense line" that the MCUs read.  Pulldown resistor on the Nch gate, pullup resistor on the Pch gate, pulldown resistor on the sense line.  With engine off the sense line should point to GND via that pulldown, then when engine goes on the Nch will yank Pch gate to GND switching it on and sending a lower-impedance 5V source to the sense line.  Could have that wired to an external interrupt to wake things up.

The way I see it, if the accessory is on then you probably expect your gadget to be available.

[graynomad]

but even still that regulator I'm using can run continuously up to 36V so it may have more leeway than many devices.

Not a huge amount more than 36v probably. Anyway it's the other components that are supposed to protect the regulator.

From Wiki re load dump

These spikes may peak at as high as 120 V and may take up to 400 ms to decay.

Just make sure the components can handle this. Load dump is a rare event (say when a battery lead or connector fails) but presumably it happens on occasion.

______
Rob

[spirilis]

From Wiki re load dump

These spikes may peak at as high as 120 V and may take up to 400 ms to decay.

Just make sure the components can handle this. Load dump is a rare event (say when a battery lead or connector fails) but presumably it happens on occasion.

______
Rob

Yeah 400ms is a long time, particularly since the caps are rated 50V .... Oh well, I did make my filter stage external to the main box so it can be swapped out.  Might design something better in the meantime.

This thread I found earlier: http://www.avrfreaks.net/index.php?name=PNphpBB2&file=printview&t=70117&start=0

Someone in there made mention of an SAE spec for how to properly handle these.

Google search revealed some hint in this book: Power Electronics Handbook

Maybe SAE J1211? (would love to find a copy of that...)

[SirNickity]

Load dump is something that should be considered, > 100V I think. Can the TVS etc handle that?

That is the question.  It can handle 1.5kW for a short time (in ms).  My goal is to shunt any short bursts to ground, and blow the fuse on longer anomalies.  In all cases, I want the components to survive.  At the very least, the PSU should sacrifice itself while protecting the load.

Since I have no desire to put my car through the wringer in the name of experimentation, I had considered testing failure modes by feeding the input from a 120v wall outlet for 100ms or so.  (Maybe I should try 500ms to be safe.)  I'm thinking this could be accomplished with a reasonably fast relay and, of course, an Arduino to sequence it.  A 50ft extension cord and a concrete driveway being the test bench, of course. ;-)

The way I see it, if the accessory is on then you probably expect your gadget to be available.

This is the gist of my remaining concern with this feature.  I can't think of any case where the difference between "accessory on" and "engine running" is of any real importance, but end users are funny like that.  There is often an ignition switch as well, where the dash lights come on and the fuel pump engages and all that, but I don't think it will be accessible to most people without tearing into the steering column.  I'm sure it's usually not going to be worth the trouble.

[graynomad]

but I don't think it will be accessible to most people without tearing into the steering column.

You can get access to an accessories signal at the fuse panel.

_____
Rob

[nic579]

Look at this pdf it has a whole pic based system + input and driver circuits for 3-5ohm coils. I am also interested in an advance control but just starting to look into what it would take.  I am going to start with hall pickup and coil drivers with the mechanical advance than move on from there.

[SirNickity]

Long time no post.  Other projects demanding time and whatnot..

So, I'm putting together a test PCB design for this.  I'd like to accommodate a switched "accessory" input, to provide a clean, protected 5v signal line to the downstream load.

My gut feeling is that the vehicle's accessory input should be protected to the same degree as the 12v power input.  I.e., a diode, a TVS, and a current limiting resistor.  Since the accessory input only needs to have enough current drive to turn on a MOSFET or something (remember, this just forms a TTL-level indicator), I should be able to use a higher value R1 (I dunno -- maybe even 1k?) and scale back the TVS and D1 diode to lower current parts, right?  I'll do some simulations, just thinking out loud for now.

Looking back on car stereos and such, it seems the accessory lead is very rarely fused.  But, I see no reason why I should treat it any different, w.r.t. its ability to start an electrical fire if shorted to ground..  Thoughts?

[Docedison]

I was wondering how long it would take for someone to spot the 10 ohm resistor... Most devices of the nature you describe are ignition controlled so I wouldn't worry about 24 V jumps or reverse jumps to the car battery, sense when the ignition is on and let that control the input to your Power supply but no one really caught the real killer and that is Load Drop, a condition where the battery cables are loose or corroded and momentarily "drop" the Load from the alternator, alternators can and will produce a hundred or more volts when open circuited... It's hell on radios, lights, and most of the rest of the 'connected' electronics.

Doc

[SirNickity]

I was wondering how long it would take for someone to spot the 10 ohm resistor...

Well if you saw a problem, you might have said something... 

Most devices of the nature you describe are ignition controlled so I wouldn't worry about 24 V jumps or reverse jumps to the car battery

OK... why is that?  I don't follow you there.

sense when the ignition is on and let that control the input to your Power supply

Negative, sir.  I'm designing for devices that might need power 24x7, but might also want to know when to power-up fully, and when to sleep.

but no one really caught the real killer and that is Load Drop

Nope, surviving load dumps is one of the design goals actually. XD  On that note...

[SirNickity]

Here's what I've come up with for the accessory line.  Have a look and tell me if it makes sense.

Here's the circuit sim code:

Code: [Select]

$ 1 5.0E-6 10.20027730826997 50 5.0 43
172 96 256 96 224 0 6 13.799999999999997 30.0 -30.0 0.0 0.5 Voltage
d 112 256 160 256 1 0.805904783
r 176 256 224 256 0 1000.0
z 240 304 240 256 1 0.8059 15.0
l 256 256 304 256 0 1.0E-6 1.2952880904054838E-4
c 320 256 320 304 0 1.0E-6 13.082409712790922
w 96 256 112 256 0
w 160 256 176 256 0
w 224 256 240 256 0
w 240 256 256 256 0
w 304 256 320 256 0
g 240 320 240 336 0
g 320 304 320 320 0
w 240 304 240 320 0
f 512 256 560 256 0 4.0
w 512 176 512 144 0
r 416 256 480 256 0 1000.0
R 512 144 480 144 0 0 40.0 5.0 0.0 0.0 0.5
w 320 256 352 256 0
w 480 256 512 256 0
g 560 320 560 336 0
w 560 272 560 320 0
r 512 272 512 320 0 100000.0
w 512 320 560 320 0
w 512 272 512 256 0
f 560 192 512 192 1 1.5
O 400 208 352 208 0
r 512 144 560 144 0 100000.0
w 560 144 560 192 0
s 352 256 416 256 0 0 false
w 560 192 560 240 0
r 432 176 432 128 0 10000.0
w 512 208 432 208 0
w 432 208 432 176 0
w 432 208 400 208 0
g 432 128 432 112 0
o 0 64 0 35 40.0 7.8125E-4 0 -1
o 5 64 0 35 20.0 9.765625E-5 1 -1
o 32 64 0 35 5.0 7.8125E-4 2 -1

... and the schematic is attached.

[SirNickity]

Preliminary reference PCB design.  Any obvious design flaws sticking out to anyone?  I still have to give it one final run-through comparing with the netlist and make sure I didn't make any mistakes, then print and populate to make sure the layout works.

BTW, this is designed for Express PCB's Miniboard service.  Two will fit on a 3.8 x 2.5" board.  With the minimum order, that gives me six to run through the paces and fine-tune.  With any luck, maybe one or two of those will get to be production PSUs.

EDIT: Oops -- forgot to connect C1 to R2, which is the connection between the input filter and the regulator.  Won't work very well without that!  I can't see an easy way to resolve that, so it looks like I'll be doing some rearranging.

[SirNickity]

Finally got around to ordering the PCBs for this project.  Here's one.  I'm hoping to populate it and run some tests this weekend.  If it works, I'll start torture-testing it to see how much abuse it can take before something breaks.  >:-)

[jtw11]

Load dump is something that should be considered, > 100V I think. Can the TVS etc handle that?

Automotive transients can often be nearer to 400V than 100V - Intel produced a good application note AP125 on designing systems powered by automotive power lines. Here's a link http://ecee.colorado.edu/~mcclurel/iap125.pdf