Go Down

Topic: Arduino Power Supplies (Read 3266 times) previous topic - next topic

Docedison

are covered in an article in Dec 2012 Circuit Cellar Magazine. Good article with some small and clever idea's

Bob
--> WA7EMS <--
"The solution of every problem is another problem." -Johann Wolfgang von Goethe
I do answer technical questions PM'd to me with whatever is in my clipboard


majenko

usps://docs.house/den/magazines/stack4/box3/magazine392.mag

:)

How do you provide a link to a magazine ...?

oric_dan

Online magazine articles at circuitcellar.com are apparently restricted to those
having a subscription. I think.

The thrust of the article is many Arduino boards use the miniscule SOT-223 v.regs
and these will overheat rather nastily if dissipating more than about 0.5 Watt. Of
course, this has been commented on in numerous threads on this forum.

The author's choice of workaround is to use a TO-220 7808 in the Vin line to reduce
dissipation in the SOT-223 device.

The other criticism is that the "tuning fork" style contacts used in Arduino female
header pins do not make very good contact for current driving situations, especially
when inserting circular pins.


retrolefty


Online magazine articles at circuitcellar.com are apparently restricted to those
having a subscription. I think.

The thrust of the article is many Arduino boards use the miniscule SOT-223 v.regs
and these will overheat rather nastily if dissipating more than about 0.5 Watt. Of
course, this has been commented on in numerous threads on this forum.

The author's choice of workaround is to use a TO-220 7808 in the Vin line to reduce
dissipation in the SOT-223 device.

The other criticism is that the "tuning fork" style contacts used in Arduino female
header pins do not make very good contact for current driving situations, especially
when inserting circular pins.




Well any modern linear regulators have internal auto shutdown protection if they overheat or too much current is drawn, so I think more is made of regulator heat then really needs to. Besides an Arduino board is great at controlling stuff but is pretty poor at powering external stuff. I would only use the boards 5v pin to power pull-ups, switch contacts, etc. Anything needing any real current should not be powered externally from the arduino's +5vdc bus. And once you commit to external regulators, using linear regulators in this day of very inexpensive Asian switch mode voltage regulators makes little sense to me.
Those arduino shield connectors work great if used as they were designed, which is male .1" square header pins, not random round wires. Again the arduino is best used as a temporary prototyping board and once a circuit is proven and debugged, and you wish to keep the project, it should then move on to a standalone chip design or small carrier board where you are free to make permanent soldered connections for better long term reliability.

Lefty

oric_dan

Quote
Well any modern linear regulators have internal auto shutdown protection if they overheat or too much current is drawn, so I think more is made of regulator heat then really needs to.


The author mentions that too, saying thermal shutdown is around 175 degC = 350 degF,
and he'll be darned if he wants his Arduino board running anywheres near that hot. Ditto
for me.

Practically speaking, he computes that, for Vin = 12V, you can draw only a miniscule
70 mA off the SOT-223 v.reg before it overheats beyond what he's comfortable with.
Ditto for me.

Quote
And once you commit to external regulators, using linear regulators in this day of very inexpensive Asian switch mode voltage regulators makes little sense to me.


CN - not that all over again, sheesh.

Quote
Those arduino shield connectors work great if used as they were designed, which is male .1" square header pins, not random round wires.


Of course, these are what everyone buys,

http://www.makershed.com/product_p/mkseeed3.htm


westfw

Quote
Practically speaking, he computes that, for Vin = 12V, you can draw only a miniscule
70 mA off the SOT-223 v.reg before it overheats beyond what he's comfortable with.

Computes, or measures?  There is "some" heatsinking, and the reverse-protection diode, and ...

Still, that kind of thinking is why the "Freeduino" ended up using a full-sized 7805, with room to add at least some heatsink if you wanted to.

oric_dan

He decided that 165 degF max is all he wants "his" v.regs to go to. This is in line with
his idea that any hotter than that and he cannot hold his thumb on the part for more
than a few seconds. [this will displease people who argue for the max value from
the datasheet, 350 degF or whatever, but that's a philosophical question more than
anything else. Nothing destroys electronics like excess heat, after all, and some people
just don't like boards you can cook an egg on].

In any case, given the heating coefficients, 150 degF works out to 0.5 W dissipation.
So, I = 0.5W/(12V - 5V) = 70 mA.

And even if you're happy with your v.reg being able to boil water, you're still talking
less than 200 mA, which is a long way from the 1 Amp used in the PR notices.

The other thing about the 7808 vs a dc-dc converter from CN, he was only talking about
a quick-fix for under-designed boards.


Constantin

#8
Dec 07, 2012, 08:47 pm Last Edit: Dec 07, 2012, 08:51 pm by Constantin Reason: 1
I happened to like the article not because of its criticism of the Arduino board - we've been through those before - but because he illustrated well how various design choices impact the linear power supply. For example, one solution is to simply run a external power supply with a lower voltage into the Arduino board, i.e. 7-9VDC. The SOT223-3 package also likely regulates less than 500mA of output power under real world circumstances without getting super-hot, no matter what respective data sheets may claim.

There are already solutions for more rugged versions of the Arduino on the market - the Ruggeduino comes to mind. If you take a look at the size of the Ruggeduino regulator (DPAK?) and the land area around it, it's pretty apparent that its designers placed emphasis on keeping the thing cool. Some versions of the Freeduino also featured large LM7805's in TO-220 packages that are easy to interface with large heat sinks. I used a similar power supply to convert 24VAC to 5VDC for the RBBB and its relay board that runs a backwashing assembly in my sprinkler system. The 7805 still gets warm but thanks to the heat sink, not uncomfortably.

To me, the main takeaway was to 1) use well-regulated external power supplies. 2) minimize any power conversion on-board, if possible 3) ensure that the desired power output can be met without burning a hole in your PCB by reviewing the various parameters and figuring out ahead of time if the on-board power supply can handle applicable currents. It re-affirmed to me that using an external, well-regulated power supply for my home-grown boards is a valid design choice as it separates the heat from the MCU, its temperature-dependent ADC output, etc.

The current Arduino move towards 3.3V-based MCUs can certainly help in this regard. For one, having multiple 5V (fused!) power sources feed a given on-board regulator can be as simple as having an array of (Schottky?) diodes that feed into Vin. Secondly, post diode, the remaining voltage that has to be 'burned off' is low enough as to minimize the work that a linear regulator has to do, minimizing its heat generation and maximizing its useful life and power capacity at the same time.

oric_dan

Quote
It re-affirmed to me that using an external, well-regulated power supply for my home-grown boards is a valid design choice


I always thought that was kind of overkill, as it's easy and cheap to choose a plain old unregulated
wall-wart that is satisfactory for most jobs. 6V @ 1A, 9V @ 300mA, &etc. The output voltage of
such wall-warts drop in direct proportion to load current increasing.

OTOH, on my home-grown boards, I use a TO-220 5V @ 1A LDO v.reg [with room for a real
heatsink], feeding a DPAK 3.3V @ 800mA LDO v.reg. Will handle about anything I do, powered
by wall-wart or batteries.

Constantin

Overkill? Small switch mode external power supplies are available for about $5 these days and are usually less expensive than unregulated wall-warts using standard transformers. Seems nuts to continue using the latter when the former usually offers more output at less cost and heat than the latter.

I totally hear you on your regulator choices, however. That said, the latest shields I developed feature a single 3.3V voltage regulator in a SOT223-3 form factor. However, I felt I could get away with it because the incoming voltage is 5VDC and I left plenty of land around the regulators to help dissipate the heat + I used vias to help spread the heat to the other side. Plus, my power demands are likely less than 100mA. Overkill, I guess.  :)

oric_dan

Quote
Small switch mode external power supplies are available for about $5 these days


We just had a long and very unhappy thread about people using such supplies, which were not
approved by Approval Agencies. So, I hope yours have all kinds of nice Approval stickers on them.
As I say - save $2, burn down the house.

3.3V SOT-223 is probably ok feeding off of 5V, and not running heavily loaded. On other threads,
people have usually used something like 0.5 watts as being the max dissipation you want in such
a package.

Constantin

Not sure what approvals mean anymore in an era where such approvals are easy to fake. Just take apple 'cube' power supplies as an example. Many fakes are good enough to fool even experienced buyers unless you know what to look for and test the unit. For that reason, the reseller is actually more important than ever.

But I take your point re: the need to buy stuff that can cut the mustard. It's not about saving a few bucks though. Old school linear power supplies with hunky transformers are increasingly difficult to find.

Docedison

@Dan... That thread was more concerned with the abuse of the regulators as I remember. The brunt of the thread was more about how poorly they performed when operated at max ratings. I've built and used in production switchers that had no "Approval Stamps or Stickers". The only caveat is to remember the ratings when using the device and having enough familiarity with the product to really basically remember that if it gets hot it needs a heat sink. For that matter Most of the 'reputable' CN houses will and do warn on the Ebay page that when using the device at max rating that more heat-sinking is required. I see no restriction to using a switcher to supply an Arduino for digital or analog processing at any Vref level or input level consistent with the current and present ground noise. a millivolt resolution should be readily do-able with a switcher and a little filtering. If I was real lazy I'd use a capacitance multiplier (2N3904 1 1K resistor and a 100uF cap, although a 47uF would work as well. This is applicable for any low power (under 200 mA) load from a switcher. Higher load currents force a Pi section filter possibly a 2 section (2 inductors and 3 capacitors) Pi filter for and additional 6 Db/octave roll off. Unless the analog sampling is at a multiple of the switcher simple oversampling and either some RC filtering or averaging / decimation will remove any possible switching artifacts. The biggest issue is that switchers art very poorly understood by the majority of the Arduino community I've encountered so far... However the Arduino Community is about learning, as is the Arduino Product. The real issues come about when the Arduino is used for more than digital control signals or to drive a led or two, the PCB wasn't intended to drive large loads directly...
I apologize for not providing any more information about the Circuit Cellar Article. I really thought that everyone read that magazine... It and Byte (in its early days) were my bibles for all things microprocessor and a big part of my early education.

Bob
--> WA7EMS <--
"The solution of every problem is another problem." -Johann Wolfgang von Goethe
I do answer technical questions PM'd to me with whatever is in my clipboard

oric_dan

Quote
Not sure what approvals mean anymore in an era where such approvals are easy to fake. Just take apple 'cube' power supplies as an example. Many fakes are good enough to fool even experienced buyers unless you know what to look for and test the unit. For that reason, the reseller is actually more important than ever.


So, from this, I assume the $5 power supplies you're using have no Approval stickers on them
at all?

Quote
Most of the 'reputable' CN houses will and do warn on the Ebay page that when using the device at max rating that more heat-sinking is required.


What does reputable from CN even mean? If it's not stamped as approved, it's certainly not been
through the approval process - at least successfully. And as Constantin says, even if it is stamped
approved, who knows. In general, I'd be suspicious of anything coming from CN and costing very little.
I wouldn't mind using a CN dc-dc converter run off a battery, but plugging into the power mains and
then going out shopping is another story.



Go Up