Forums

Power supply in automotive environment

Started by Jake McGuire January 20, 2004
Sadly, I'm running up against the limits of my meager mixed-signal
design knowledge.  I'm trying to put together an embedded device
intended for use in an automotive (well, motorcycle) environment, and
am trying to figure out how to power the damn thing.

A few of the components I need to use (video overlay chip, MEMS
accelerometers, op-amps) require +5V.  And since I'm going to to be
doing analog sampling and then A/D conversion, I suspect the power has
to be fairly clean.  None of my signals are going to be above 100Hz,
so I might even be wrong on that point.

Anyway, as near as I can tell, my options include powering from the
vehicle +12V system via a linear regulator, using a lithium-ion
battery along with one of the nifty USB-powered charger chips and a
switching regulator to step up to +5V, or just whipping up a 6-cell
NiMH or NiCad pack and regulating it down to +5V, then using an
external charger to recharge the thing.

The switching option seems like the best combination of operational
and design simplicity, but I'm worried about the effect of the 50mV
P-P 1MHz power supply ripple on my already poorly-understood analog
circuitry.  It seems like I should be able to LC notch filter that
out, but is this going to mess up the feedback on the power supply
chip?

The vehicular +12V supply seems the next simplest (operationally), but
I've heard all sorts of nasty things about the quality of that power,
and how it's full of noise and spikes, occasionally reverses polarity,
etc etc.  Is handling this just a matter of putting in appropriately
rated caps and a reverse-polarity protection diode, or is there more
to it?

Is implementing this going to be so much of a pain in the ass that I
should just solder together a bunch of AA NiMH cells and accept the
need to swap out battery packs every now and again?

(note: I'm not planning on actually selling this to strangers, so I'm
not concerned about UL testing or functioning 100% of the time, just
"mostly working".)

-jake
Jake McGuire wrote:

> Sadly, I'm running up against the limits of my meager mixed-signal > design knowledge. I'm trying to put together an embedded device > intended for use in an automotive (well, motorcycle) environment, and > am trying to figure out how to power the damn thing. > > A few of the components I need to use (video overlay chip, MEMS > accelerometers, op-amps) require +5V. And since I'm going to to be > doing analog sampling and then A/D conversion, I suspect the power has > to be fairly clean. None of my signals are going to be above 100Hz, > so I might even be wrong on that point. > > Anyway, as near as I can tell, my options include powering from the > vehicle +12V system via a linear regulator, using a lithium-ion > battery along with one of the nifty USB-powered charger chips and a > switching regulator to step up to +5V, or just whipping up a 6-cell > NiMH or NiCad pack and regulating it down to +5V, then using an > external charger to recharge the thing. > > The switching option seems like the best combination of operational > and design simplicity, but I'm worried about the effect of the 50mV > P-P 1MHz power supply ripple on my already poorly-understood analog > circuitry. It seems like I should be able to LC notch filter that > out, but is this going to mess up the feedback on the power supply > chip? > > The vehicular +12V supply seems the next simplest (operationally), but > I've heard all sorts of nasty things about the quality of that power, > and how it's full of noise and spikes, occasionally reverses polarity, > etc etc. Is handling this just a matter of putting in appropriately > rated caps and a reverse-polarity protection diode, or is there more > to it? > > Is implementing this going to be so much of a pain in the ass that I > should just solder together a bunch of AA NiMH cells and accept the > need to swap out battery packs every now and again? > > (note: I'm not planning on actually selling this to strangers, so I'm > not concerned about UL testing or functioning 100% of the time, just > "mostly working".)
The one sentence, mostly working answer is Nat Semi Simple switcher for +5v and a 5 volt A/D with an internal 4.096v reference. Pay close attention to data sheet recommendations for reference bypassing and/or filtering. Put a diode and a fuse in series with the battery, followed by a 20-some volt zener across the regulator input and you should be reasonably safe. I would not guarantee it for mass production though.

Jake McGuire wrote:

> Sadly, I'm running up against the limits of my meager mixed-signal > design knowledge. I'm trying to put together an embedded device > intended for use in an automotive (well, motorcycle) environment, and > am trying to figure out how to power the damn thing. > > A few of the components I need to use (video overlay chip, MEMS > accelerometers, op-amps) require +5V. And since I'm going to to be > doing analog sampling and then A/D conversion, I suspect the power has > to be fairly clean. None of my signals are going to be above 100Hz, > so I might even be wrong on that point. > > Anyway, as near as I can tell, my options include powering from the > vehicle +12V system via a linear regulator, using a lithium-ion > battery along with one of the nifty USB-powered charger chips and a > switching regulator to step up to +5V, or just whipping up a 6-cell > NiMH or NiCad pack and regulating it down to +5V, then using an > external charger to recharge the thing. > > The switching option seems like the best combination of operational > and design simplicity, but I'm worried about the effect of the 50mV > P-P 1MHz power supply ripple on my already poorly-understood analog > circuitry. It seems like I should be able to LC notch filter that > out, but is this going to mess up the feedback on the power supply > chip? > > The vehicular +12V supply seems the next simplest (operationally), but > I've heard all sorts of nasty things about the quality of that power, > and how it's full of noise and spikes, occasionally reverses polarity, > etc etc. Is handling this just a matter of putting in appropriately > rated caps and a reverse-polarity protection diode, or is there more > to it? > > Is implementing this going to be so much of a pain in the ass that I > should just solder together a bunch of AA NiMH cells and accept the > need to swap out battery packs every now and again? > > (note: I'm not planning on actually selling this to strangers, so I'm > not concerned about UL testing or functioning 100% of the time, just > "mostly working".) > > -jake
First verify that the supply is 12V, most motorcyles use a 6V supply. If this is the case and your current depands are small a zenner diode and a few capacitors would be my choice.
Grant Killey wrote:

>First verify that the supply is 12V, most motorcyles use a 6V supply.
That was true at one time, but not for about the last 40 or more years.
> If >this is the case and your current depands are small a zenner diode and a >few capacitors would be my choice.
-- ======================================================================== Michael Kesti | "And like, one and one don't make | two, one and one make one." mkesti@gv.net | - The Who, Bargain
It depends on your power requirements.

If it's a few hundred mA or less, I'd use a series diode, a reservoir
capacitor and a linear regulator (78M05 etc). You could include a series L
to reduce "noise".

The limits of this circuit will likely be dissipation and cooling of the
regulator ((12v - 0.7v -5v)* Current)

For more than this, use a monolithic switcher (Linear Tech make some very
nice ones and their website has afree simulator for designing with them).
You can get wound components for these from many suppliers (e.g. Coilcraft)



"Jake McGuire" <jamcguir@yahoo.com> wrote in message
news:dbb82eb7.0401201254.2ffe6b9a@posting.google.com...
> Sadly, I'm running up against the limits of my meager mixed-signal > design knowledge. I'm trying to put together an embedded device > intended for use in an automotive (well, motorcycle) environment, and > am trying to figure out how to power the damn thing. > > A few of the components I need to use (video overlay chip, MEMS > accelerometers, op-amps) require +5V. And since I'm going to to be > doing analog sampling and then A/D conversion, I suspect the power has > to be fairly clean. None of my signals are going to be above 100Hz, > so I might even be wrong on that point. > > Anyway, as near as I can tell, my options include powering from the > vehicle +12V system via a linear regulator, using a lithium-ion > battery along with one of the nifty USB-powered charger chips and a > switching regulator to step up to +5V, or just whipping up a 6-cell > NiMH or NiCad pack and regulating it down to +5V, then using an > external charger to recharge the thing. > > The switching option seems like the best combination of operational > and design simplicity, but I'm worried about the effect of the 50mV > P-P 1MHz power supply ripple on my already poorly-understood analog > circuitry. It seems like I should be able to LC notch filter that > out, but is this going to mess up the feedback on the power supply > chip? > > The vehicular +12V supply seems the next simplest (operationally), but > I've heard all sorts of nasty things about the quality of that power, > and how it's full of noise and spikes, occasionally reverses polarity, > etc etc. Is handling this just a matter of putting in appropriately > rated caps and a reverse-polarity protection diode, or is there more > to it? > > Is implementing this going to be so much of a pain in the ass that I > should just solder together a bunch of AA NiMH cells and accept the > need to swap out battery packs every now and again? > > (note: I'm not planning on actually selling this to strangers, so I'm > not concerned about UL testing or functioning 100% of the time, just > "mostly working".) > > -jake
"Jake McGuire" <jamcguir@yahoo.com> wrote in message
news:dbb82eb7.0401201254.2ffe6b9a@posting.google.com...
> Sadly, I'm running up against the limits of my meager mixed-signal > design knowledge. I'm trying to put together an embedded device > intended for use in an automotive (well, motorcycle) environment, and > am trying to figure out how to power the damn thing. > > A few of the components I need to use (video overlay chip, MEMS > accelerometers, op-amps) require +5V. And since I'm going to to be > doing analog sampling and then A/D conversion, I suspect the power has > to be fairly clean. None of my signals are going to be above 100Hz, > so I might even be wrong on that point. > > Anyway, as near as I can tell, my options include powering from the > vehicle +12V system via a linear regulator, using a lithium-ion > battery along with one of the nifty USB-powered charger chips and a > switching regulator to step up to +5V, or just whipping up a 6-cell > NiMH or NiCad pack and regulating it down to +5V, then using an > external charger to recharge the thing. > > The switching option seems like the best combination of operational > and design simplicity, but I'm worried about the effect of the 50mV > P-P 1MHz power supply ripple on my already poorly-understood analog > circuitry. It seems like I should be able to LC notch filter that > out, but is this going to mess up the feedback on the power supply > chip?
You can put an LC notch filter or just a lowpass filter after the regulator, but AFTER the feedback. Do not include the filter inside the feedback loop, or it will likely oscillate. You could also put a linear regulator after the switcher, and if you use a low dropout part, the losses aren't very high. Just remember to set the output voltage of the switcher high enough to give you the headroom required for the linear reg.
> > The vehicular +12V supply seems the next simplest (operationally), but > I've heard all sorts of nasty things about the quality of that power, > and how it's full of noise and spikes, occasionally reverses polarity, > etc etc. Is handling this just a matter of putting in appropriately > rated caps and a reverse-polarity protection diode, or is there more > to it?
On a 12V system, usually you have to be able to handle at least 24V on the input side. Put in a reverse polarity protection diode, and a fuse (or polyswitch), then follow that with a 24V power zener to ground to limit incoming voltage spikes. We used to use a MR2535L part from Motorola which was meant for this specific purpose. A datasheet can be found here: http://www.onsemi.com/pub/Collateral/MR2535L-D.PDF
> > Is implementing this going to be so much of a pain in the ass that I > should just solder together a bunch of AA NiMH cells and accept the > need to swap out battery packs every now and again? > > (note: I'm not planning on actually selling this to strangers, so I'm > not concerned about UL testing or functioning 100% of the time, just > "mostly working".) > > -jake
I would think that you should be able to get standard regulators to do what you want, without having to resort to a battery supply. Mike Anton
In article <100tgb5p7aom585@corp.supernews.com>, 
manton@nocompusmart.abspam.ca says...

> On a 12V system, usually you have to be able to handle at least 24V > on the input side. Put in a reverse polarity protection diode, and a > fuse (or polyswitch), then follow that with a 24V power zener to > ground to limit incoming voltage spikes. We used to use a MR2535L > part from Motorola which was meant for this specific purpose. > A datasheet can be found here: > http://www.onsemi.com/pub/Collateral/MR2535L-D.PDF
I am actually facing the design of a power supply for automotive use as well. I came up with the following and would appreciate comments... Take 12V DC into a bridge rectifier (~ terminals). Output of rectifier "+" through poly fuse to power supply circuit. Place a TVS (such as the above MR2535L) across the "+" & "-" terminals of the rectifier. My thinking is if there is an overvoltage the MR2535 alone will eat the spikes. However, if for some reason there are negative spikes on the 12V, the TVS alone won't do anything (and other solutions use a second TVS to shunt negative spikes into the positive terminal). With a rectifier all spikes are positive going & combined with the TVS, they should be clamped...right? Thanks, Jay.
Jay wrote:

> In article <100tgb5p7aom585@corp.supernews.com>, > manton@nocompusmart.abspam.ca says... > > >>On a 12V system, usually you have to be able to handle at least 24V >>on the input side. Put in a reverse polarity protection diode, and a >>fuse (or polyswitch), then follow that with a 24V power zener to >>ground to limit incoming voltage spikes. We used to use a MR2535L >>part from Motorola which was meant for this specific purpose. >>A datasheet can be found here: >>http://www.onsemi.com/pub/Collateral/MR2535L-D.PDF > > > I am actually facing the design of a power supply for automotive use as > well. > > I came up with the following and would appreciate comments... > > Take 12V DC into a bridge rectifier (~ terminals). Output of rectifier > "+" through poly fuse to power supply circuit. Place a TVS (such as the > above MR2535L) across the "+" & "-" terminals of the rectifier. > > My thinking is if there is an overvoltage the MR2535 alone will eat the > spikes. However, if for some reason there are negative spikes on the > 12V, the TVS alone won't do anything (and other solutions use a second > TVS to shunt negative spikes into the positive terminal). > > With a rectifier all spikes are positive going & combined with the TVS, > they should be clamped...right?
Personally, I'd prefer a single series diode for reverse polarity protection. I'd not want the ground of my device to be a diode drop above battery ground. Of course, if you do it my way you have to deal with those negative spikes.
Jim Stewart wrote:

> Personally, I'd prefer a single series diode for > reverse polarity protection. I'd not want the > ground of my device to be a diode drop above > battery ground. Of course, if you do it my way > you have to deal with those negative spikes.
You can make a low-drop bridge recitifier out of MOSFETs... use a MOSFET with gate and drain (or source? I can't remember) wired together as a diode. Um, I'd need to look in my transistor book to remember which way round it would then work, but I'm sure a *real* electronics engineer will just know :-) ABS
On Thu, 22 Jan 2004 19:52:14 +0000, Alaric B Snell
<alaric@alaric-snell.com> wrote:

>Jim Stewart wrote: > >> Personally, I'd prefer a single series diode for >> reverse polarity protection. I'd not want the >> ground of my device to be a diode drop above >> battery ground. Of course, if you do it my way >> you have to deal with those negative spikes. > >You can make a low-drop bridge recitifier out of MOSFETs... use a MOSFET >with gate and drain (or source? I can't remember) wired together as a >diode. Um, I'd need to look in my transistor book to remember which way >round it would then work, but I'm sure a *real* electronics engineer >will just know :-)
There is no magic way to make a MOSFET into a low-drop diode. The bridge rectifiers made out of MOSFETs use additional active circuitry to drive the gates appropriately. It is not as simple as just shorting two pins of the MOSFET. -Robert Scott Ypsilanti, Michigan (Reply through this forum, not by direct e-mail to me, as automatic reply address is fake.)