LM2596 DC-DC Step Down Modules - Load Testing

[metalphreak]

Had a few of these in my parts box I've never actually used for anything. I figured it would be a good subject to play around with on my new Maynuo DC load.

They range from $1.35 to $3.00 delivered depending on which seller you buy from (ebay, aliexpress, DX etc). There's a number of variations in PCB colour and layout - some have mounting holes which is handy. Very generic specifications given:

Input voltage:4.5-40V
Output voltage:1.5-35V(Adjustable)
Output current: Rated current is 2A, maximum 3A(Additional heat sink is required)
Conversion Efficiency: Up to 92% (output voltage higher, the higher the efficiency)
Switching Frequency: 150KHz

Mine has a National Semi LM2596 according to the label on the IC... Wasn't Nat Semi bought out by TI a long time ago? Sure must be a large stockpile of these parts sitting around in china...

National Semi Datasheet: http://pdf1.alldatasheet.com/datasheet-pdf/view/8744/NSC/LM2596.html

Thermal shutdown and current limit protection
Power Dissipation Internally limited
Temperature Range ?40?C ? TJ ? +125?C
Supply Voltage 4.5V to 40V

I ran my tests with 30V input, 12V output, with a load current up to 2.5A. It didn't last one minute before failing with the temperature probe only showing 65deg celcius!

Fake ICs? Pure crap luck and I got a faulty one? I have two more of these, and a quick test at 2A shows the temperature rising just as fast anyway. My plan is to test them at a lower current to see exactly what they can do without getting too hot. Then I'll have a look at the output waveform (if they don't fail first!  ). The test in the video shows efficiency to be ~85% so it was dissipating 5W to be fair - I would still expect the thermal shutdown and power limiting to work however...

[digsys]

A few months ago, I bought a HEAP of different S/Mode modules, from different suppliers to test and rate. I chose only ones
that claimed 90%+ (up to 95%) efficiency. NONE came close to either efficiency or rated output - NOT ONE. I even bought 5
of each to reduce probability distribution. I haven't dissolved the resin protected ones yet, to see IF they actually used
the chips they claimed were inside (or forgeries). It's still a to-do, I don't keep MEK in my workshop these days (like I used to :-))

[ptricks]

I use quite a few of these switchers (the chip itself not the boards) every month and they are solid performers if you follow the data sheet.

There is a lot of those for sale on ebay and I never buy them because of the way they implement the heat sink is only good for low current loads, the recommended size is 2oz copper and at least 2 square inches.  The inductor used also isn't ideal , at high currents a torroid is best.  They also do not include the filtering of the output to remove switching noise.

[daveshah]

Sure must be a large stockpile of these parts sitting around in china...

Perhaps they were salvaged off boards sent for recycling

[metalphreak]

I use quite a few of these switchers (the chip itself not the boards) every month and they are solid performers if you follow the data sheet.

There is a lot of those for sale on ebay and I never buy them because of the way they implement the heat sink is only good for low current loads, the recommended size is 2oz copper and at least 2 square inches.  The inductor used also isn't ideal , at high currents a torroid is best.  They also do not include the filtering of the output to remove switching noise.

Yep it's certainly not the greatest design. I've used a similar PCB heatsinking on a slightly smaller board and it was capable of dissipating 2.5W continuously before overtemp kicked in (LM1117 reg). Still pushing the LM1117 to the limit of its temperature range but it didn't kill it

Did a quick resistance check around the module. Output V+ is shorted to ground with 1ohm of resistance.

Ah yep, the SS54 5A Schottky diode has failed as a closed circuit. Datasheet says it needs to be rated 1.3x the load current. 1.3 x 2.5A = 3.25A so not sure why it failed.

So this diode is SMA package size. An SS54 is SMC package for 5A versions... I also can't seem to find an SMA package 5A diode at all.

One of the 3 boards is slightly different. It has an SS14 which is a 1A diode! It's the same size as the SS54 too. I guess it's a case of cutting costs absolutely everywhere to get the price as low as possible.

[mikes]

The ones of these I've used had 220 uF input/output caps. If you look at the LM2596 data sheet, that's appropriate for mid range use (maybe 10-20V in / 6-15 V out). Same with the inductor, 33 uf, mid-range, mid power. Unfortunately, the LM2596 doesn't allow a "one size fits all" optimal design.

Having said that, there's an example circuit in the ONsemi datasheet which uses the exact values given above (well, 100 uF input cap, which is even worse), and is labeled "1.2 to 35 V Adjustable 3.0 A Power Supply with Low Output Ripple," so even the datasheet goes against its own advice. It also calls for a 1n5822 (40V/3A), which is worse than the SS54 mentioned above.

I use 'em as more efficient, more flexible replacements for 78xx regulators. Keeping them under 1 A, and not pushing the limits, works for me.

[peter.mitchell]

A few months ago, I bought a HEAP of different S/Mode modules, from different suppliers to test and rate. I chose only ones
that claimed 90%+ (up to 95%) efficiency. NONE came close to either efficiency or rated output - NOT ONE. I even bought 5
of each to reduce probability distribution. I haven't dissolved the resin protected ones yet, to see IF they actually used
the chips they claimed were inside (or forgeries). It's still a to-do, I don't keep MEK in my workshop these days (like I used to :-))

I have used a fair few and as you said, they don't generally meet spec; with that said, most of the time they cost $3-4, which is about the same as the chip itself, so i use them within their limits and for that price they're still pretty good.

[digsys]

I have used a fair few and as you said, they don't generally meet spec; with that said, most of the time they cost $3-4, which is about the same as the chip itself, so i use them within their limits and for that price they're still pretty good. 

Absolutely. ONE reason I'm running the exercise. PLUS, when I saw some of the higher quoted efficiencies, I wanted to know IF
there were new S/Mode chips around that I'd not heard of. Kinda doing to them what they do to us :-)

[peter.mitchell]

I have used a fair few and as you said, they don't generally meet spec; with that said, most of the time they cost $3-4, which is about the same as the chip itself, so i use them within their limits and for that price they're still pretty good. 

Absolutely. ONE reason I'm running the exercise. PLUS, when I saw some of the higher quoted efficiencies, I wanted to know IF
there were new S/Mode chips around that I'd not heard of. Kinda doing to them what they do to us :-)

I have seen some converters from the 'baba and the 'bay that are pretty good though, like 85++% at 50W

[Bored@Work]

For fun check the switching frequency. I had some of these modules where the switching frequency didn't even remotely match the frequency according to the IC's datasheet. The printing of the IC marking also looked strange. My guess was the switching ICs were fakes.

[metalphreak]

Replacement B340A diodes arrived from RS Components today. Soldered one back on and it's back in business 30c fix on a $2 item 

[G7PSK]

I have a similar unit that I run a Panasonic toughbook on, the heat sinking was a bit poor so I soldered a piece of copper to the chip, the unit does have a toroidal inductance but the chip number has been removed, it was sold as being a 4 amp unit.

[Rick Law]

...
Very generic specifications given:

Input voltage:4.5-40V
Output voltage:1.5-35V(Adjustable)
Output current: Rated current is 2A, maximum 3A(Additional heat sink is required)
Conversion Efficiency: Up to 92% (output voltage higher, the higher the efficiency)
Switching Frequency: 150KHz

I recall having looked at that board closely (but decided to get boost+buck).  One very important spec I recalled seeing but not on your list there is "Up to 10W, with heatsink 25W max."  I recall some listed it as high as 15W without heatsink but certainly not above 15W.

If you are driving it at 12V 2A, you are way over that spec.

(I just check on eBay, of the three I looked, none of them listed the wattage.)

Rick

[Cozzmo]

I just ordered some of these (the first batch went AWOL, never arrived so had to order more) to run an Arduino project. Input voltage ~34-36v down to ~7v for the arduino with a few indicator leds and two shields drawing about 300ma peak. Hopefully won't have any issues with them !

[sleemanj]

I believe those converters are "rated" to handle 10 Watts Power Output, not 10 Watts of heat dissipation.  If you add a (large) heatsink you can take that to 20 Watts Output I believe.

You are pumping 30 watts out of that poor wee board, without any heatsink at all, not even a itty bitty one.  I'm surprised it lasted as long as it did

A google translate from a chinese supplier of those same LM2596 buck modules:
"* module using domestic quality LM2596 chip, high efficiency (up to 92%) , heat a small, easily brought up 3A (short-term) , less than 2.5A current work suggests that for a long time use, coupled with the heat at the same time piece (10W output above). "

And my rewording of that into actual english:
"*  'Chinese' LM2596 chip, up to 92% efficient if all the planets align (more like 80-84%), can handle very brief current spike to 3A, more like 2A constant, 10W or more output requires a heatsink"

Of course you have to take specifications of anything an ebay (ali, tao...) says with a decent pinch of salt, not only are they probably exaggerated, but they are going to be exaggerated best case specs. 

I've used (and sold) a few of the LM2577 based SEPIC modules which are similar.  They are crazy cheap and seem to do what they say "on the box", but I haven't actually tested the specs.  They are "rated" (by one of my suppliers) at 12W without a heatsink, and 20W with one.

I have never fitted a heatsink (and they do get toasty) but you can get heatsinks on ebay which fit the entire underside of the board, such as:
http://www.ebay.com.au/itm/Heatsink-w-insulation-tape-Aluminum-LM2596-2577-2587-/260769416022?pt=LH_DefaultDomain_0&hash=item3cb7118356&_uhb=1#ht_2051wt_1399

[dr_p]

yes, but the OP's point is that even overloaded, these should not fail, as the original (a.k.a. not fake) LM2596 has this:

Self protection features include a two stage frequency reducing current limit for the output switch and an over
temperature shutdown for complete protection under fault conditions.

[mikes]

yes, but the OP's point is that even overloaded, these should not fail, as the original (a.k.a. not fake) LM2596 has this:

Self protection features include a two stage frequency reducing current limit for the output switch and an over
temperature shutdown for complete protection under fault conditions.

That's (as stated) self-protection. It doesn't protect against the failure of external components, like the diode mentioned above. I haven't seen any claims that the 2596 itself has failed.

[dr_p]

Oh, my bad.    I misunderstood the original post and thought the IC died.


I also bought some of these for 2$ a pop and for 6$ you get ones with adjustable current limit. AFAIK they're good to have around.

[metalphreak]

Yeah, I initially assumed it was the IC that died. I would have expected the diode to fail open circuit not short circuit. Hard to tell in circuit unless you start removing components. Replacement diode seems to handle it ok (although the temp on the IC does rise too high as I expected). It was probably just a fluke that I happened to test one with a dodgy diode  Knowing now that the diode failed, I would not think that they were fake ICs.

[AmmoJammo]

Old thread, but who cares!

I've measured 3.8amps out of one of these, for a short time, without anything failing

I've also seen one over 100 degrees (on the back of the board) without failing.

Of course, it would be nice if it current limited, or shut down, before it got to this point :p

[DVDRW]

I have this 60v version. I wanted to drive 100W led from 48v offgrid battery. Datasheet says MAX 3A. So wattage limit depend on heatsink and coil size right? Just you have to add larger heatsink and 33uh coil and it works!
Heatsink stayed cold and coil is touchable.
88W @ 2,4A just in case because no current limit.

There are fake 60V ones, so be careful.

[doctormord]

The inductors connection should be as short as possible.

[macboy]

The inductors connection should be as short as possible.

Why? To mimimize lead inductance? 

[Kjelt]

I have this 60v version. 

How did you get that IC off? I tried to remove the capacitors and replace them with Nichicon but that ground plane needed a lot of heat 

[doctormord]

To reduce EMI radiation and efficieny reduction due to wire resistance. (Beside further L*dI/dt losses)