Pages: [1] 2 3   Go Down
Author Topic: Car PSU proposal -- need design critique  (Read 3667 times)
0 Members and 1 Guest are viewing this topic.
Anchorage, AK
Offline Offline
Edison Member
*
Karma: 42
Posts: 1176
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

I'm seeing quite a few car project ideas come up.  I have my own in the works and have been studying PSU design principles that take into account the unique and challenging environment that 12v electronics have to deal with.  So, I have attached a switching power supply design based on posts and app notes I've read.  I've taken the best ideas I found and put them together in hopes that it'll be nearly bullet-proof.

I'm not an accomplished engineer, so I'm hoping someone out there with some experience can throw in their two cents.  I'm hoping to put together a prototype, put one in my car, and (continuously) run some kind of simple yet persistent sketch, where I can tell if it crashes or reboots.  Like a software clock or something.  I plan to leave it running until it stops, so there's a baseline for how reliable it can be expected to be.  Maybe I can put one on my Ninja, too ... that's a *really* hostile power environment.  ;-)  Any car should be better than that.

Alright, enough jibber-jabber.  Experts, tear it to shreds...

Here's how it works:

Battery and fuse -- obvious.  The diode (D1) protects against reverse connection, and negative spikes.  The 10-ohm resistor (R1) is a current-limiting device in case of high-voltage positive spikes or a fault in the PSU.  D1 will also ensure the caps inside don't discharge externally during starter cranks or spikes.

The transient voltage suppressor (D2) will clamp the input voltage to 20v.  This is designed to catch alternator load dumps.  The previous resistor will prevent this from turning into a short if the car is jumped at 24v, but I don't know if the fuse, resistor, or TVS will survive.  I'd like to test that.

L1 and C1 are an input filter to reduce some of the noise from ignition / injector pulses and whatnot.

The IC is a TI MC33063 adjustable switching regulator.  I used the reference step-down design from the datasheet.

L3 and C4 are an optional output filter to reduce ripple, as specified in the datasheet.  The cleaner the better, and it's about $1 worth of parts.

All inductors and caps need to be capable of sustained output (or ripple) current -- I'm shooting for about 1A, though the sheet claims 5v @ 500mA.


* carpsu5v.png (12.8 KB, 675x290 - viewed 160 times.)
« Last Edit: May 16, 2012, 12:58:55 am by SirNickity » Logged

Global Moderator
Dallas
Offline Offline
Shannon Member
*****
Karma: 199
Posts: 12768
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset


How will the circuit behave if the battery is reversed?

And you may want to include a fuse / short circuit protection on the output side.
Logged

Anchorage, AK
Offline Offline
Edison Member
*
Karma: 42
Posts: 1176
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

Ideally the diode (D1) would prevent anything from happening.  Do you see any reason to the contrary?  I do intend to build one as a sacrificial torture test victim, so we'll see.

Output shorts and current limiting should be handled by the regulator -- it claims to handle such faults in a graceful manner.  However, I would not oppose a 500mA fuse (ala USB power entry circuits) on the connection between this and the load.  Good suggestion.

Anything else?
Logged

Global Moderator
Dallas
Offline Offline
Shannon Member
*****
Karma: 199
Posts: 12768
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset

Do you see any reason to the contrary?

I am completely unqualified to answer that question.  I just know that mechanics occasionally make mistakes and was encouraging you (and anyone else following along) to spend a few moments considering the unexpected.

Quote
Output shorts and current limiting should be handled by the regulator -- it claims to handle such faults in a graceful manner.  However, I would not oppose a 500mA fuse (ala USB power entry circuits) on the connection between this and the load.  Good suggestion.

Thanks.  The suggestion certainly does not come from personal experience.  I would never foolishly leave an Arduino plugged into a USB port while I moved loose wires around.   smiley-grin

Quote
Anything else?

If you get impatient for more replies, bump the topic and I'll move it over to General Electronics.  I'll give it a forum tour to maximum the number of eyeballs.
Logged

Anchorage, AK
Offline Offline
Edison Member
*
Karma: 42
Posts: 1176
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

While I'm waiting on the parts fairy to deliver, I did some simulations on the input stage (everything in front of the switching IC) of this circuit.  Here are some things I found:

- It looks like R1 (the current limiter) should be 47 ohms instead of 10.  With a 1W resistor, a short-term 24v jump-start should be perfectly safe.  The TVS (essentially a 20v zener diode) will conduct and clamp the output to 20v, but with a real-world average of 26v input, that's still only around 600mW through R1.  It touches 1W just under 28v, which would be asking too much in my opinion, so a running 24v vehicle could damage the resistor unless the fuse blows first.  (This is assuming a similar curve on nominal vs. charging voltage in 24v vehicles as compared to 12v vehicles, but I have no experience with that.)  Using a 68 ohm resistor would move the 1W threshold to around 29v.  The trade-off here is a slightly longer charge cycle for the filter cap (C1), but with a 13.8v input, the output still reaches 9v within about 30ms at 47 ohms, and 45ms at 68 ohms.

- I seem to have over-sized L1 (the input choke) quite a bit.  Even at 10uH, there's very little fluctuation after the L1/C1 filter.  This should keep the switching regulator happy enough.  With a 24v, 40Hz AC input, I get about 300mV ripple (mind, this is before the regulator).  At 400Hz, it's around 50mV.  At 4k, the simulator doesn't show anything meaningful.  It's well out of my reach to do proper noise analysis, though.

- Reverse polarity protection seems to be working fine, with nothing but leakage current through the diode.

I'm using http://www.falstad.com/circuit for my tests.  You can import my test circuit and play around with it.  The voltage input is variable (set to -30 to +30), and can be wiggled around in real-time to test the filter.

Here's the code:

Code:
$ 1 5.0E-6 10.20027730826997 50 5.0 43
172 160 208 160 176 0 6 27.6 30.0 -30.0 0.0 0.5 Voltage
d 176 208 224 208 1 0.805904783
r 240 208 288 208 0 47.0
z 304 256 304 208 1 0.805904783 20.0
l 320 208 368 208 0 9.999999999999999E-6 0.009189132775834264
c 384 208 384 256 0 4.7E-4 20.216092108038907
w 160 208 176 208 0
w 224 208 240 208 0
w 288 208 304 208 0
w 304 208 320 208 0
w 368 208 384 208 0
g 304 272 304 288 0
O 448 208 512 208 0
r 448 208 448 256 0 2200.0
w 384 208 448 208 0
g 448 256 448 272 0
g 384 256 384 272 0
w 304 256 304 272 0
o 0 64 0 35 40.0 0.4 0 -1
o 9 64 0 35 40.0 0.1 1 -1
o 17 64 0 35 7.62939453125E-5 0.2 2 -1
o 5 64 0 35 40.0 9.765625E-5 3 -1
o 12 64 0 34 40.0 9.765625E-5 4 -1

@CB:  If you'd like to move this around, feel free.  I want to get this right so folks can grab a parts list, build it, and concentrate on their project instead of having to figure out how to reinvent this particular wheel.
Logged

UK
Offline Offline
Shannon Member
****
Karma: 222
Posts: 12549
-
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

Consider putting a bridge rectifier on the input side rather than a diode?

While I've never designed a switched mode power supply, I thought it was possible to design them so that they tolerated an input voltage up to the breakdown voltage of the switch. In that case, if you target 24V or 30V or whatever as your peak input load, there'd be no normal situation where your crowbar overvoltage protection was required.

If this is intended to be a generally robust supply it would be worth providing enough output protection for either output line to be shorted to ground, 12V or the other output line. If using a fused protection, you would earn lots of brownie points by using resettable (or even self-resetting) breakers rather than a disposable fuse.
Logged

I only provide help via the forum - please do not contact me for private consultancy.

Offline Offline
Full Member
***
Karma: 0
Posts: 204
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

Just wondering, we plug sensitive electronics like GPS, ipods, etc into our lighter sockets all the time and they don't fry left and right. Do they have this type of circuitry built it or are our boards just sensitive to spikes and such?
Logged

Offline Offline
Newbie
*
Karma: 0
Posts: 24
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

three  good  ideas to include
1 25 A 400 piv full wave bridge on the input makes this able to be used in positive or negative ground vehicles or ones with  reversed  positive on the  power outlet
2 on the output stage  use the following 7812 + 12v reg  ti makes these with a 125v input  rating  use that version its  rated at 1.5a so use something like a poly fuse ptc http://www.mouser.com/Circuit-Protection/PTC-Resettable-Fuses/_/N-5g3e?Keyword=ptc&FS=True  in the output  also  make sure to place a 4700uf cap between output and ground and a 480uf cap between input and ground  they should both  be rated at 100 v
3 do the same circuit  but use the 5v version of the same reg as listed above   this gives you  stable 12 and 5 volt dc psu  ill look in my  design library  for a  power management ic that will provide a soft on function  that allows the  system to  come up  easy to avoid damage as well as over under voltage  protection  to prevent turn on if either is present
this  should be  a circuit that can be applied to  direct vehicle power or to the  front end of a dc switching supply   be advised that the full wave bridge is to be placed at the input of the switching supply along with two 68,000uf 50v caps for the  primary  stabilization of the input voltage the input  should have a 25A 32 volt  atc fuse before the  bridge and a 50A 32 volt atc  fuse after the  primary  stabilization  circuit this fuse is there to  protect the  secondary systems
also a 50a 1000 piv diode should be placed  in the  positive leg  from the  bridge to prevent power from returning back  through the  input as  there wil be enough stored energy in the primary  stabilization  system to  possibly weld with if it shorted out
this  suggestion is from experience  as i have  built  Mobil  dc psu for  Mobil server systems with as much as 4000A at 12 and 5 volts on the outputs and 9,000 A on the +3.3 volt line those however  had a 480 volt 3 phase primary  but were able to be  run on 480 volts dc in emergencies and they did do that in two  situations
Logged

Anchorage, AK
Offline Offline
Edison Member
*
Karma: 42
Posts: 1176
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

GREAT feedback, guys.  I really appreciate it.  Keep 'em coming.

Quote
Consider putting a bridge rectifier on the input side rather than a diode?
Quote
25 A 400 piv full wave bridge on the input makes this able to be used in positive or negative ground vehicles or ones with  reversed  positive on the  power outlet

Hmm... I didn't think about positive grounds.  As devil's advocate though, I do wonder if there's any danger of being too accommodating?  I.e., if the PSU works under potentially accidental conditions, it may mask an improper hookup until the load starts interfacing with external circuits.  Maybe that's the user's problem.  Another possible solution is to add a bi-color LED.  Green = polarity OK; Red = polarity reversed.  Anyone with a positive ground is likely to know this beforehand, and can either flip the LED, accept red as their indicator, or leave it out of their build entirely.  I'll have to see what options exist for ultra low current bi-color indicators (or just use two in reverse parallel.)  Further thoughts?

Quote
While I've never designed a switched mode power supply, I thought it was possible to design them so that they tolerated an input voltage up to the breakdown voltage of the switch. In that case, if you target 24V or 30V or whatever as your peak input load, there'd be no normal situation where your crowbar overvoltage protection was required.

Fair point..  Ideally, I would like to be well below the maximum input range of the switching chip to allow room for things like inductor-induced swings (is that even an issue here?) and temperature derating.  I'm sure I can find a suitable (cheap, easy-to-use, minimal external part count) replacement IC with a higher rating though -- I think this one is safe to 40v anyway, so there's already wiggle room.  It's certainly possible just to use a higher clamping voltage on the TVS and reduce the margin on the input at the users' discretion, so maybe clamping at 28v would be a better choice.  Personally, I don't think I would ever consider 24v jumping as "normal", but that's my definition, not everyone else's.  Incidentally, I wonder how the average user's existing car electronics would handle this situation.  Opinions welcome.

Quote
If this is intended to be a generally robust supply it would be worth providing enough output protection for either output line to be shorted to ground, 12V or the other output line.

I'll see what I can do.  :-)

Quote
If using a fused protection, you would earn lots of brownie points by using resettable (or even self-resetting) breakers rather than a disposable fuse.

I think I might plan to have and test some various options, including the inbuilt output protection from the IC alone, and intentionally abuse it to see the reaction in each case.  Then, offer some part numbers (fuse, breakers..) along with test results, and leave that up to the user based on their needs.  In a perfect world, the output fuse will be nothing but a fail-safe, where the IC can and does handle 99% of faults, and the fuse only blows if something is really very wrong.

Quote
on the output stage  use the following 7812 + 12v reg  ti makes these with a 125v input  rating  use that version its  rated at 1.5a

Can you explain why you would recommend a linear regulator?  Is it just for simplicity?  Using a power-management controller would help if there's no need for the project to be running 24/7, but otherwise, wouldn't it be prudent to use the most efficient conversion possible?

Quote
ill look in my  design library  for a  power management ic that will provide a soft on function  that allows the  system to  come up  easy to avoid damage as well as over under voltage  protection  to prevent turn on if either is present

I thought about doing some kind of over/under protection using a discrete transistor...  If you have a cheap and easy IC to suggest, that would be interesting.

Quote
also  make sure to place a 4700uf cap between output and ground and a 480uf cap between input and ground  they should both  be rated at 100 v
Quote
the full wave bridge is to be placed at the input of the switching supply along with two 68,000uf 50v caps

I was thinking 100v, too.  4700uF seems like a lot, though.  68,000uF seems like gross overkill.  Keep in mind -- the design goal (at least for this iteration) is to deliver less than an amp at 5v.  Probably much less, typically.  That in mind, is there a good reason to go this high, or were you thinking of something larger scale?  It would significantly complicate the requirements for the input stage, I think, to build this large without justification.  That could be my ignorance, though, so please elaborate if I'm mistaken.

Quote
Just wondering, we plug sensitive electronics like GPS, ipods, etc into our lighter sockets all the time and they don't fry left and right. Do they have this type of circuitry built it or are our boards just sensitive to spikes and such?

I really don't know.  I expect many of the cheap ones out there are built to generate revenue, and the well-being of the load is the onus of the end device's engineers and the user.  Things that stay in the car over the long haul (your stereo and the engine computer) are probably fortified with vitamins.  Just an uneducated guess.

At any rate, my goal is a PSU as robust as possible that can be built by anyone with experience soldering, while keeping the cost, size, and power draw down to a minimum.
Logged

United Kingdom
Offline Offline
Tesla Member
***
Karma: 224
Posts: 6593
Hofstadter's Law: It always takes longer than you expect, even when you take into account Hofstadter's Law.
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset

I think your schematic is not quite right. According to the datasheet at http://www.ti.com/lit/ds/symlink/mc33063a.pdf, the input supply should go to pin 6, not pins 1/7/8.

R1 is too high. At 1A output, the input current will be around 0.7A at 10v input, and a 10 ohm resistor will drop 7v. This doesn't leave enough margin for the regulator (it is a bipolar regular with a darlington output, saturation voltage is specified at 1.3v max). I suggest reducing it to 2.2 or 3.3 ohms. I'm not sure you should worry about a 24v jump start, you will have other things to worry about (like the 12v vehicle battery exploding). However, that regulator is specified for up to 40v input, so if you use input capacitors rated at 50v then it will survive a 24v jump start anyway, if you use a higher voltage TVS diode.

Otherwise it looks OK to me. I wouldn't use a bridge rectifier on the input, you already have a protection diode and if you are going to use mosfets to switch anything running from the 12v supply, you will need a common ground anyway.
Logged

Formal verification of safety-critical software, software development, and electronic design and prototyping. See http://www.eschertech.com. Please do not ask for unpaid help via PM, use the forum.

Anchorage, AK
Offline Offline
Edison Member
*
Karma: 42
Posts: 1176
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

I think your schematic is not quite right. According to the datasheet, the input supply should go to pin 6, not pins 1/7/8.

You're right... the supply should connect on the pin 6 side of R2, not the pin 7/8 side.  Thanks for catching that!  :-)

R1 is too high. At 1A output, the input current will be around 0.7A at 10v input, and a 10 ohm resistor will drop 7v.

I don't understand -- how's it dropping voltage?  Unless the TVS is clamping there's no voltage divider, and it should only act as a current limiter.  What did I miss?

I'm not sure you should worry about a 24v jump start, you will have other things to worry about (like the 12v vehicle battery exploding).

I've heard of people doing this, but TBH, I have no idea if it's safe at all.  Batteries are more complicated than my simplistic understanding of them, so I tried to accommodate this technique if or when it's used.  I certainly wouldn't do it, as I don't have the prettiest face in the world, but I'd still like to keep it.  Maybe it's safe enough for short periods.  I'll leave that to the professionals, and try to build a PSU that can keep up.

I wouldn't use a bridge rectifier on the input, you already have a protection diode and if you are going to use mosfets to switch anything running from the 12v supply, you will need a common ground anyway.

Yeah, I'm leaning that way.  I do like the idea in theory, so I may provide it as an optional alternative with caveats.  BUT, having the PSU cheerfully cope with reverse connection, and then having the load self-destruct when it connects to external circuits is not exactly my idea of friendly, nor easy-to-use, so I think I'll design for one-way operation by default.
Logged

United Kingdom
Offline Offline
Tesla Member
***
Karma: 224
Posts: 6593
Hofstadter's Law: It always takes longer than you expect, even when you take into account Hofstadter's Law.
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset

I don't understand -- how's it dropping voltage?  Unless the TVS is clamping there's no voltage divider, and it should only act as a current limiter.  What did I miss?

The entire current drawn from the 12v input flows through that resistor, that's what causes the voltage drop.
Logged

Formal verification of safety-critical software, software development, and electronic design and prototyping. See http://www.eschertech.com. Please do not ask for unpaid help via PM, use the forum.

Anchorage, AK
Offline Offline
Edison Member
*
Karma: 42
Posts: 1176
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

Uh, wait.. never mind.  I'm too young to claim senior moment, so I'll stick with "noob".
Logged

Maryland, USA
Offline Offline
Full Member
***
Karma: 0
Posts: 162
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset

Heh I'm doing something quite similar, although using more pre-built components... and I don't have a TVS on mine, just an LC filter up front.  The switching regulator I'm using is a premade muRata Power Solutions OKI-78SR-5/1.5 (http://www.mouser.com/ProductDetail/Murata-Power-Solutions/OKI-78SR-5-15-W36-C/?qs=sGAEpiMZZMtwaiKVUtQsNemMZL4TplJBqOl7845nWHA%3d) 1.5A unit, fits the same pinout as a linear regulator which is nice.

Good to see all the various options spelled out on the table though!  My setup doesn't exactly handle 12V going into the output stage at the PSU, but one specific leg of the gadget network has a filter with zeners, limiting resistors and fuses on the power and data (Dallas 1-wire DS18B20 which will be located in the engine bay, need to harden it for potential shorting from water/salt if the primary weather sealing gets breached)
« Last Edit: May 22, 2012, 09:48:38 pm by spirilis » Logged

Maryland, USA
Offline Offline
Full Member
***
Karma: 0
Posts: 162
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset

Also one thing I would add, you might want to minimize the use of LEDs that will be on during engine shutoff (unless this is not intended to be used while the engine is off) ... and any microcontrollers should handily know when the engine is shut off & go into reasonable sleep modes...
Logged

Pages: [1] 2 3   Go Up
Jump to: