A cautionary tale with cheap ebay Voltage Regulators.

http://en.wikipedia.org/wiki/Short_circuit

"A short circuit is an abnormal connection between two nodes of an electric circuit intended to be at different voltages. This results in an excessive electric current/overcurrent"

  • wiki

touching the clips results in max amp discharge causing the wires to give off heat. is this not a "short" ? I'm serious.....

You probably want to focus on the part where it talks about very low / no resistance.

You cannot have a voltage drop over a wire with no resistance.

huh? (video... look at the video)

A cautionary tale with cheap ebay Voltage Regulators

The Title... or more correectly how to abuse a component without really trying. The regulator is rated at 2A output current... for short periods of time because there is no effective heat sink. Read ALL the data sheet INCLUDING the thermal considerations. The voltage drift is most likely due to components on the board getting hot. The OP once mentioned that the module was too hot to touch and it is my opinion that IF as I believe this is the case, then the module needs to have a heat sink soldered to the bottom of the switcher. A 12 mm X 25 mm strip of copper about .6 mm thick with a hole 1/4 from one end a 90 deg at the half point, (middle) soldered to the switcher and mounted to a good heat radiator. IMO the component is being abused and it is very simple to use it properly. I appear here to be a hard nose but that is my training, I had to learn the hard way to be Certain that anything I did was 100% Perfect. I am non trying to be the expert, more to share my thoughts only, If I appear otherwise it is unintentional and I apologize. {Edit, RKJ}

Bob

dhenry: You probably want to focus on the part where it talks about very low / no resistance.

You cannot have a voltage drop over a wire with no resistance.

A wire with no resistance? Can you spare me a little of yours as my supply of super-conductors has run low?

Lefty

Well, it is a switching regulator, so... if you apply a large load to the output, then remove that load, it is likely to overshoot. With an undersized output filtering stage, it's entirely possible that its regulation suffers when faces with current transients. Now, if the voltage increases steadily with a consistent heavy load on the output (say, a 5 ohm resistor) then it's probably a component behaving out of tolerance. Maybe a saturated inductor? Or comparator oscillation? I don't know.

It would be interesting to put a scope on the feedback voltage divider to see what the comparator input looks like under overload conditions.

Whether this is "abnormal abuse" or something that should be handled properly by the IC's protection circuitry is debatable. I would like to know that if something in the load circuit failed, the PSU would do everything it its power to behave according to the three-tiered model: Protect the user, protect the load, and protect itself -- in that order. As a realist, I don't necessarily expect that in a regulator bought for spare change from a no-name overseas merchant via eBay.

It would also be interesting to wire up the same circuit with known-real chip and parts and see if it behaves similarly...

Hmm. Noticing that actually SHORTING the output of a typical buck switching regulator is sort-of like operating the inductor like a boost regulator…

Also noticing that the input and output caps in the photos on eBay are smaller than the “typical circuit” in the datasheet. (Even if “220” really means 220uF, and not 22uF, which is another possibility.)

Indeed it would and I for one would be very happy to do so. The very few 3 or 4 of the 'Chinese' boards have all worked fine. I am conservative in nature and have no desire to deal with excess heat so I try to remember that 10 to 20% of the load current will be lost as heat and plan accordingly. The concept of counterfeit IC's crops up from time to time and for some, I can believe that the money for IC's that are "Pure Unobtanium" is too good to pass up and China is a Long way away. There is as well the send me the defective ones and I'll test them and let you know what my 'engineers say". This has enough holes in it for it to work occasionally. However to begin to think about faking chips of that nature is unbelievable particularly in the quantities that are sold every day. It is possible that only one house in Shenzhen has all the phonies.. but I've seen, bought, used and have several 'variants' with and without meters, current adjustment, in both buck and boost mode Too.. I have designed and built switching regulators with both National Semi and Linear Tech's devices and by my experience that heatsink is going to get very warm with a 6 or 7 volt I/O differential and a 1A load. It will take a few minutes and so might fool someone initially because of the thermal mass of the device, but the bottom line is as always if it is too hot to touch, comfortably, IT IS TOO HOT... Put a heat sink on it. The rest of the errors reported could just as easily be measurement errors. I am Not Trying to imply that the measurements are anything but what they claim or to be as I have no intention of starting any kind of discourse over a minor difference of opinion. Possibly the wiring to the regulators might be long enough to require extra bypassing. There are many possible causes and again due to the high volume of sales coupled with the low numbers of complaints reported here in the Forum I am for one somewhat inclined to be skeptical of "Cheap Chinese Knock Off's or counterfeit IC's as the reasons for performance issues, especially of that nature.

Bob

No I don't think it means 22 uF, that marking is used for ceramic caps... to the best of my knowledge and experience.. Or at least in the past 50 years.. give or take a day. That they are marked 220 uF means little though, those are Electrolytics on that PCB and the SMT components place them at being less than ..? 20 years old.. Surplus at any rate. Seriously the parts if more than a year old are suspect, highly suspect... I put my scope on both input and output and use a 10 uF tant and check for significant ripple reduction on both input and output noise with any of those switchers.. It's called Due Dilligence. If I see any the caps are old and dried out, I replace them. For the cost of the board... Well below what I could purchase them in less than 25 quantities blank and bare of anything but the etching and screening.. Thank You, I will try to understand my 'resource' and make the best of it. Much more interesting for me to find out Exactly why and what to do to fix/avoid the issue and report that... Now... that's really useful.

Bob

SirNickity: Well, it is a switching regulator, so... if you apply a large load to the output, then remove that load, it is likely to overshoot. With an undersized output filtering stage, it's entirely possible that its regulation suffers when faces with current transients. Now, if the voltage increases steadily with a consistent heavy load on the output (say, a 5 ohm resistor) then it's probably a component behaving out of tolerance. Maybe a saturated inductor? Or comparator oscillation? I don't know.

It would be interesting to put a scope on the feedback voltage divider to see what the comparator input looks like under overload conditions.

Whether this is "abnormal abuse" or something that should be handled properly by the IC's protection circuitry is debatable. I would like to know that if something in the load circuit failed, the PSU would do everything it its power to behave according to the three-tiered model: Protect the user, protect the load, and protect itself -- in that order. As a realist, I don't necessarily expect that in a regulator bought for spare change from a no-name overseas merchant via eBay.

Ah now that explains it... cheers

I bought a lot of 10 of the ebay LM2596 boards for about $15 shipped.

http://www.ebay.com/itm/10pcs-LM2596S-power-module-DC-DC-Buck-Converter-Step-Down-Module-/400358408922

After some testing, I suspect that the LM2596S chip on these boards is counterfeit. The LM2596 spec sheet says that the switching frequency is 150 KHz, but these measure more like 50 KHz. The output ripple is supposed to be under 30 mV, but I see 200 mV with a 5V/300ma load. Also, if I set it for 5V at no load, I measure 4.75V at 1.6A load, more droop than expected.

Last year I bought a very similar-looking device from ebay for $6, only slight differences in the layout, matching this item:

http://www.ebay.com/itm/DC-DC-Buck-Converter-Module-LM2596-Breadboard-Power-Module-1-23V-35V/321158804705

That device shows a 150 KHz switching frequency as expected and much lower output ripple.

Chagrin: You're driving a LED based on voltage, then you're making it worse by using a regulator circuit that is specified at +/- 2.5% regulation. Grumpy Mike would kick your butt for doing it that way. There's nothing about a LED V/I curve that suggests anything but eventual failure.

If Grumpy Mike is mad, you know that what you are attempting to do is actually going to work. If he seems to like it, scrap your design. There's no hope left for it.

He seems to think that if a butterfly flaps its wings in taiwan, it will damage your arduino (that's not much of an exageration). Connecting an LED without a resistor will not harm the Arduino, only the LED. If it is done in short pulses (under 2s), no harm can be done.

so let me get you a phillips rebel and see how that theory holds

If you supply the correct /regulated/ voltage along with no ripple...

The LED will be fine.... show me a dimly lit led running at 2v (1ma draw), no resistor required! Resistors are purely in place to protect the led from voltage spikes (current limiting in this case)

LED forward voltage drops 2mV per degree Celsius.

and your high powered ones get hot pretty quick, so what do you do

you could be normal and use a current limiting device, or you could make some feedback loop to constantly regulate the voltage as it changes

I work with high powered LED's for a living, its not uncommon to watch them zip on up from 23 to 60-70c, there's almost a volt difference using the general guideline posted above

besides I was responding to the post above, last I checked the arduino outputs 5 volts

i'm using a 10 watt high powered LED with a TIP31 (68 ohm resistor) and PWM (analogWrite) with no problems at all, i draw around 800ma @ 14.5v using analogWrite(pin,200) (11.6watts) but with good heatsinking which my LED does have ... it's happy to draw around 1.2amps (12v LED)

"it works fine" is not an acceptable answer

whats the voltage drop over time, whats the die temp over time, whats the(true) current draw over time, what is all that in an enclosure, whats your ambient temperature?

and I mean serious time not 5 min and welp its good

I don't care what you do with your 1$ ebay find, but you are only seeing a small part of the picture and preaching it like its the bible

it may work fine for now, it may work fine for a while the point is you dont know and you only want to argue that fact, while people who do have a clue sit here shaking their head

Do you understand what the role of the power transistor is?....

Unlike a mosfet, the TIP31 is current based, using PWM, you're limiting the current to the LED (EG, same as a resistor)

ughhh, man some people need prozak on here....