Hi all. I'm working on a little circuit that will live in my car...so
I need an automotive voltage regulator. Cost isn't really a huge
concern since I only need a couple. I know many people like the LM2940
though I recall a post about the capacitors it needs being expensive.
What I'm unclear on is whether that was for no load regulation or if
the low ESR caps are always needed.
I'm not so great on power supply design so any help would be awesome!
Thanks!
Josh
--
A common mistake that people make when trying to design something
completely foolproof is to underestimate the ingenuity of complete
fools.
-Douglas Adams
a) 22uF is the minimum amount of capacitance to use
b) no upper bound on total capacitance
c) both low and high esr can cause oscillation
d) a mix of 75% aluminum electrolytic with 25% tantalum gives the
right esr
So I would try 22uF tantalum and 100uF Al electrolytic. Depending on
your application, you might also get by with only using tantalum caps.
In that case, you probably would want to use double the recommended
minimum of 22uf so might try 47uf.
Problem areas would tend to be using *only* Al electrolytics or *only*
low esr ceramics.
> Hi all. I'm working on a little circuit that will live in my car...so
> I need an automotive voltage regulator. Cost isn't really a huge
> concern since I only need a couple. I know many people like the LM2940
> though I recall a post about the capacitors it needs being expensive.
> What I'm unclear on is whether that was for no load regulation or if
> the low ESR caps are always needed.
>
> I'm not so great on power supply design so any help would be awesome!
>
> Thanks!
>
> Josh
On Fri, Apr 11, 2008 at 10:23 PM, Marcel Duchamp
<spam_OUTmarcel.duchampTakeThisOuTsbcglobal.net> wrote:
> From the data sheet,
>
> a) 22uF is the minimum amount of capacitance to use
> b) no upper bound on total capacitance
> c) both low and high esr can cause oscillation
> d) a mix of 75% aluminum electrolytic with 25% tantalum gives the
> right esr
>
> So I would try 22uF tantalum and 100uF Al electrolytic. Depending on
> your application, you might also get by with only using tantalum caps.
> In that case, you probably would want to use double the recommended
> minimum of 22uf so might try 47uf.
>
> Problem areas would tend to be using *only* Al electrolytics or *only*
> low esr ceramics.
Hm...ok, I will see what I have in stock. I'm trying to knock one of
these together quickly to test with so I'm trying not to do too many
parts orders. National has a pretty quick sampling program if I recall
correctly.
Funny thing is that this all might not matter. I'm pulling power off
the car radio CD changer port so it might already be regulated...not
something I really want to chance though. At some point I'll put a
scope on it, start the car and see what happens!
Thanks!
Josh
--
A common mistake that people make when trying to design something
completely foolproof is to underestimate the ingenuity of complete
fools.
-Douglas Adams
> On Fri, Apr 11, 2008 at 10:23 PM, Marcel Duchamp
> <.....marcel.duchampKILLspam@spam@sbcglobal.net> wrote:
>
>> From the data sheet,
>>
>> a) 22uF is the minimum amount of capacitance to use
>> b) no upper bound on total capacitance
>> c) both low and high esr can cause oscillation
>> d) a mix of 75% aluminum electrolytic with 25% tantalum gives the
>> right esr
>>
>> So I would try 22uF tantalum and 100uF Al electrolytic. Depending on
>> your application, you might also get by with only using tantalum caps.
>> In that case, you probably would want to use double the recommended
>> minimum of 22uf so might try 47uf.
>>
>> Problem areas would tend to be using *only* Al electrolytics or *only*
>> low esr ceramics.
>>
>
> Hm...ok, I will see what I have in stock. I'm trying to knock one of
> these together quickly to test with so I'm trying not to do too many
> parts orders. National has a pretty quick sampling program if I recall
> correctly.
>
> Funny thing is that this all might not matter. I'm pulling power off
> the car radio CD changer port so it might already be regulated...not
> something I really want to chance though. At some point I'll put a
> scope on it, start the car and see what happens!
>
>
Why not use a 7805/TO220 in the automotive environment? All you need to
do is install the
recommended caps, but also install a TVS or varistor on the input to
protect from ESD spiking.
The LM7805 is very forgiving, and not as cranky as the LM2940. I use 'em
all over the place.
In the motorcycle environment, I had to add an inductor as well, again
to stop generator and
regulator switching noise. But that is it.
On Sat, 12 Apr 2008 08:50:15 -0700, Bob Axtell wrote:
> Why not use a 7805/TO220 in the automotive environment? All you need to
> do is install the recommended caps, but also install a TVS or varistor on
> the input to protect from ESD spiking. The LM7805 is very forgiving, and
> not as cranky as the LM2940. I use 'em all over the place.
>
> In the motorcycle environment, I had to add an inductor as well, again
> to stop generator and regulator switching noise. But that is it.
In general I would agree with everything you said. It's good practice in
the automotive environment to use ESD and overcurrent protection regardless
of the regulator you choose.
There is one aspect you may have overlooked though and that is dropout
voltage and operation during starting/heavy load/low battery conditions.
As it happens, I've been working on a lot of over-the-road commercial
trucking electronics the past few years and we continue to use LM29xx
series LDO regulators due to the requirement for the equipment to operate
at 7-8V input from the B+ line. A 7805 will drop about 2.5V and that
becomes problematic at low battery voltages.
If the cost constraints allow it, I actually prefer a fairly high frequency
(> 500KHz) switching regulator as it eliminates a lot of on-board power
dissipation at anything less than low operating currents and you can easily
get about the same low dropout voltage as a good linear LDO regulator. When
you get to higher operating currents (say above 100mA), you almost are
forced to a switcher unless you have a heatsink capable of dissipating some
significant power.
On Apr 12, 2008, at 8:50 AM, Bob Axtell wrote:
> In the motorcycle environment, I had to add an inductor as well...
Is there science to picking an inductor value for noise filtering, or
do you just get as large a uH value as will fit within your space,
current, and voltage drop specs?
> On Apr 12, 2008, at 8:50 AM, Bob Axtell wrote:
>> In the motorcycle environment, I had to add an inductor
>> as well...
> Is there science to picking an inductor value for noise
> filtering, or
> do you just get as large a uH value as will fit within
> your space,
> current, and voltage drop specs?
Detroit knew - "There's no substitute for inches", BUT in a
resource limited world the impedance of the filter is
affected by the LC ratio and you may be able to optimise
noise performance by taking impedance into account.
William "Chops" Westfield wrote:
> On Apr 12, 2008, at 8:50 AM, Bob Axtell wrote:
>
>> In the motorcycle environment, I had to add an inductor as well...
>>
>
> Is there science to picking an inductor value for noise filtering, or
> do you just get as large a uH value as will fit within your space,
> current, and voltage drop specs?
>
> BillW
>
>
>
We bought several inductor samples, and connected a dummy output load,
putting it into a box. It turned
out that the LM7805 with a certain inductance value at input worked best
at suppressing spikes at the
output of the LM7805.
Another comment made earlier is to use a switching regulator. Actually
this indeed works the best,
as all the spikes are broken up by the switcher. But it was not used in
this case.
Motorcycles are extremely noisy environments, mostly because the battery
impedance is so high (small
battery).
On Sat, Apr 12, 2008 at 11:50 AM, Bob Axtell <engineerKILLspamcotse.net> wrote:
> Why not use a 7805/TO220 in the automotive environment? All you need to
> do is install the
> recommended caps, but also install a TVS or varistor on the input to
> protect from ESD spiking.
> The LM7805 is very forgiving, and not as cranky as the LM2940. I use 'em
> all over the place.
Hm...good points, same with Matt's post. I think I'm going to use an
LM1117T for my first test and then try to scope the power lines and
see if the radio is actually regulating its output.
Thanks!
Josh
--
A common mistake that people make when trying to design something
completely foolproof is to underestimate the ingenuity of complete
fools.
-Douglas Adams
On Mon, 2008-04-14 at 00:06 -0400, Josh Koffman wrote:
> On Sat, Apr 12, 2008 at 11:50 AM, Bob Axtell <.....engineerKILLspam.....cotse.net> wrote:
> > Why not use a 7805/TO220 in the automotive environment? All you need to
> > do is install the
> > recommended caps, but also install a TVS or varistor on the input to
> > protect from ESD spiking.
> > The LM7805 is very forgiving, and not as cranky as the LM2940. I use 'em
> > all over the place.
>
> Hm...good points, same with Matt's post. I think I'm going to use an
> LM1117T for my first test and then try to scope the power lines and
> see if the radio is actually regulating its output.
FWIW I've used the 1117 for many designs and love it. It just works, and
seems to be able to take quite a beating.
>> > Why not use a 7805/TO220 in the automotive
>> > environment? All you need to
>> > do is install the
>> > recommended caps, but also install a TVS or varistor
>> > on the input to
>> > protect from ESD spiking.
>> > The LM7805 is very forgiving, and not as cranky as the
>> > LM2940. I use 'em
>> > all over the place.
The LM293x family are intended for automotive use. I'm not
sure whether or not ou would consider them "cranky" in the
same manner as the LM2940 but the experience of a friend who
specialises in in-vehicle computer telemetry and metering
systems is that they work very well.
On Mon, Apr 14, 2008 at 12:41 AM, Apptech <EraseMEapptechspam_OUTTakeThisOuTparadise.net.nz> wrote:
> The LM293x family are intended for automotive use. I'm not
> sure whether or not ou would consider them "cranky" in the
> same manner as the LM2940 but the experience of a friend who
> specialises in in-vehicle computer telemetry and metering
> systems is that they work very well.
Er...maybe I made a mistake. I thought the LM2940 was part of the same
family. What part in the LM293x family should I be looking at to get a
5V positive regulator? LM2936 seems to be ok. I'm finding National's
flash/Java whatever it is website to be supremely unhelpful.
Thanks!
Josh
--
A common mistake that people make when trying to design something
completely foolproof is to underestimate the ingenuity of complete
fools.
-Douglas Adams
On Mon, Apr 14, 2008 at 12:14 AM, Herbert Graf <mailinglist4spam_OUTfarcite.net> wrote:
> FWIW I've used the 1117 for many designs and love it. It just works, and
> seems to be able to take quite a beating.
Did you need tantalum caps on both the input and output?
Josh
--
A common mistake that people make when trying to design something
completely foolproof is to underestimate the ingenuity of complete
fools.
-Douglas Adams
>> FWIW I've used the 1117 for many designs and love it. It
>> just works, and
>> seems to be able to take quite a beating.
> Did you need tantalum caps on both the input and output?
*NO* regulator 'needs' tantalum caps on the input :-)
Tantalums turn spikes into smell, smoke, noise, flames and
explosion, followed by a hard short across the terminals -
I've encountered all of these in a single event - in
approximately that order. Even a low energy spike that
exceeds their voltage rating can punch through the oxide
layer that forms the "electrolyte" and then the main power
takes over and enlarges the 'hole'.
If you can GUARANTEE that there will NEVER be voltages
present that EVER exceed the rated voltage then they are
very compact compared to most alternatives.
However, solid Alumin(i)um caps, not to be confused with wet
electrolyte ones, have a similar capacitance density and do
not suffer from the same problems.
Tantalum MAY be OK as an output cap on a regulator. but a
solid Al does as well.
>> The LM293x family are intended for automotive use. I'm
>> not
>> sure whether or not ou would consider them "cranky" in
>> the
>> same manner as the LM2940 but the experience of a friend
>> who
>> specialises in in-vehicle computer telemetry and
>> metering
>> systems is that they work very well.
> Er...maybe I made a mistake. I thought the LM2940 was part
> of the same
> family. What part in the LM293x family should I be looking
> at to get a
> 5V positive regulator? LM2936 seems to be ok. I'm finding
> National's
> flash/Java whatever it is website to be supremely
> unhelpful.
They do seem to be similarly described :-).
There's no certainty of how similar parts in contiguous
numbering ranges are without poring through data sheets, and
then guessing.
The LM2936 is ultra low power, low dropout and nice to use
and ultra expensive. It is limited to 50 mA max output. From
several years ago memory I thing the Iq is about 14 uA
typical, which is why I used to use it in a power-always-on
design.
LM2930 is 1A version.
Iq is far worse than for 2936 - but MUCH cheaper.
The LM294x series says "designed also for vehicular ..." and
the LM293c says "designed originally for vehicular ..." but
that MAY just reflect their original inception dates.
On Mon, 2008-04-14 at 02:13 -0400, Josh Koffman wrote:
> On Mon, Apr 14, 2008 at 12:14 AM, Herbert Graf <@spam@mailinglist4KILLspamfarcite.net> wrote:
> > FWIW I've used the 1117 for many designs and love it. It just works, and
> > seems to be able to take quite a beating.
>
> Did you need tantalum caps on both the input and output?
I don't remember about "need" offhand, I just placed the caps
recommended in the datasheet. I don't believe they were anything
special. TTYL
> Message: 5
> Date: Mon, 14 Apr 2008 16:41:54 +1200
> From: Apptech <KILLspamapptechKILLspamparadise.net.nz>
> Subject: Re: [EE] LM2940 Question
> To: "Microcontroller discussion list - Public." <RemoveMEpiclistTakeThisOuTmit.edu>
> Message-ID: <011901c89de9$dfbcc3d0$e701a8c0@y2k>
> Content-Type: text/plain; format=flowed; charset=iso-8859-1;
> reply-type=original
>
> >> > Why not use a 7805/TO220 in the automotive
> >> > environment? All you need to
> >> > do is install the
> >> > recommended caps, but also install a TVS or varistor
> >> > on the input to
> >> > protect from ESD spiking.
> >> > The LM7805 is very forgiving, and not as cranky as the
> >> > LM2940. I use 'em
> >> > all over the place.
>
> The LM293x family are intended for automotive use. I'm not
> sure whether or not ou would consider them "cranky" in the
> same manner as the LM2940 but the experience of a friend who
> specialises in in-vehicle computer telemetry and metering
> systems is that they work very well.
>
>
>
> Russell
I've used the LM2940S in automotive applications, they are very robust. They
do have a requirement for the ESR of the output cap but it is something like
0.1R to 2R, dependent on current drawn. This is actually quite a wide range
but you should consider the parallel effect of any other large caps on the
board. I generally use tants for the output cap.
On Sun, Apr 13, 2008 at 11:37 PM, Apptech <spamBeGoneapptechspamBeGoneparadise.net.nz> wrote:
> However, solid Alumin(i)um caps, not to be confused with wet
> electrolyte ones, have a similar capacitance density and do
> not suffer from the same problems.
It's surface mount (which is unfortunate for this project) and ideally
I'd like a higher voltage rating "just in case". Still, it's the only
solid aluminum cap I could find in stock at those two (my preferred
vendors).
What do you think?
Josh
--
A common mistake that people make when trying to design something
completely foolproof is to underestimate the ingenuity of complete
fools.
-Douglas Adams
sbcglobal.net> wrote:
