Newbie mistake: Using a programmable power supply to test LEDs

[Sienna]

Hello everyone.

I have just been blowing up a lot of red LEDs when I try to test them (such as after making small RGB panels).  This is probably common sense, but having released a whole bunch of magic smoke, I figured I would relate the experience just in case any other beginner has the same 'bright' idea I had.

I just recently got a Rigol DP832 - my first real lab bench power supply.   Well, the DP832 is really cool in that you can just program in 15mA or 20mA, and viola, you have an instant LED tester, right?  No messing with LM317 circuits or resistors!  (Who doesn't make use of a $400 power supply to make $.02 LEDs light up!   )

Well, maybe, but only if you are careful.   

After making a whole bunch of RGB LED panels, I wanted to quick sweep them with the power supply to test all the LED connections.  (If everything lights up, then I know the surface mount soldering is OK.)
So, I do what any good beginner does, I set the voltage at 15V (four LEDs in series per color), set the current at 20mA, and figure I am great - the power supply will regulate to 20mA, and I will be safe as I try all the different color channels.  Turn the output on, and commence to go down the line with the probe.  Blue works, Green works, but those Red channels, man the LEDs must have been a bad batch!  They work once, maybe twice, but they all seem to be dying!  What crap! 

After lots of  , it turns out that those Red LEDs just can't stand being over-voltaged to double their Vf for the very brief time it takes the current regulator to kick in.  (The Red LED chain only took about ~8.2V for it to reach 20mA current, compared to the Blue @12.8V and Green @13.6V.  The delta between my chosen voltage of 15V and Vf was much smaller on Green and Blue, which is likely what saved them from my stupidity.)

One the one board that I didn't blow up, I started with the output of the power supply off, hooked up the Red chains, and turned the output on.  Viola!  No troublesome red channel!  When allowed to ramp from 0V (vice having to clamp from 15V), the Red LEDs never exceeded their ratings. 

Moral of the story:  Blowing up LEDs is a great way to get more familiar with your brand new power supply, the peak detect function on your DMM, and getting clued into the world of 'duh'!

TL/DR:  Even if your power supply has a CC mode, never test an LED by connecting an active output with a high voltage to said LED.  Instead, start with the output disabled, wire the LED, and turn on the output to allow the CC mode to kick in before you over voltage the LED.

[Andy Watson]

If I remember correctly, one of Dave's teardowns shows a thumping big capacitor immediately behind the output terminals on those power supplies. That cap will momentarily supply much more current than you've set on the current limit.

[IanB]

Yes, absolutely. With all items of lab equipment there is a correct way to use them, and this is not often taught in the user guide. I am sorry for the loss of all your LEDs.

I confirm your conclusions though:

• Never connect an energized power supply output to a device under test. Always wire up the device with the power supply off.
  
  
• Never turn on the power supply through its mains switch with the device under test connected.
  
  
• If the power supply has an output on/off switch, use that to energize and de-energize the output once you are comfortable the settings are correct.
  
  
• If you are testing a device for the first time, start with the output voltage at zero and gradually ramp it up to the desired value  while monitoring the device current. Turn off immediately if anything doesn't look right.
  
  

One important thing to keep in mind is that all power supplies have a capacitor across the output terminals to help with regulation. If you connect an energized power supply to some device under test this capacitor will discharge through the device outside the current regulation loop.

[qno]

So much for not using a resistor..

It would have saved you desoldering the red LED's

[c4757p]

After lots of  , it turns out that those Red LEDs just can't stand being over-voltaged to double their Vf for the very brief time it takes the current regulator to kick in.

The I(V) curve of a diode is exponential. If 8.2V gives you 20mA, double that will give thousands of amps, likely limited by the circuit's series resistance, and possibly creating a PED (projectile-emitting diode).

[tszaboo]

After lots of  , it turns out that those Red LEDs just can't stand being over-voltaged to double their Vf for the very brief time it takes the current regulator to kick in.

The I(V) curve of a diode is exponential. If 8.2V gives you 20mA, double that will give thousands of amps, likely limited by the circuit's series resistance, and possibly creating a PED (projectile-emitting diode).

By the way, previously I was able to blow up LEDs and shoot the lens to the other side of the room  . Now it doesn't seem to work, I've tried 3 different LEDs. Does anyone know a specific type of LED which can be fired? I'm guessing it needs to be high brightness. (and yes, I have a degree to do this  )

[AndyC_772]

You can use CC mode to test your LEDs. All you need to do is connect a shorting link in parallel with the device you're testing, then switch on the PSU, then remove the link. Having the short will force it into CC mode, and it'll stay in that mode when you remove the link. Your device gets the programmed current and no more.

[c4757p]

Many well-designed all analog power supplies start up quite gracefully, so the mains switch can often be used as a load switch. My Power Designs 2005A has sufficiently fine control of current to set LED test current, then shut off the supply, connect the LED, then turn on the supply. Proper design of the analog section of modern digital power supplies seems rather lacking, though (e.g. the mammoth capacitors right behind the banana jacks on this PSU).

By the way, previously I was able to blow up LEDs and shoot the lens to the other side of the room  . Now it doesn't seem to work, I've tried 3 different LEDs. Does anyone know a specific type of LED which can be fired? I'm guessing it needs to be high brightness. (and yes, I have a degree to do this  )

Is... is this a joke?

[alm]

Sounds like it's indeed the output cap discharging through the LED. If it would take the current limit long enough to kick in to destroy a LED I would consider it a design flaw, but obviously the power supply will be unable to limit the current from the discharging cap. One reason to prefer power supplies with fairly moderate output caps. Shorting the power supply before connecting the LED should indeed prevent this problem.

• Never connect an energized power supply output to a device under test. Always wire up the device with the power supply off.
• Never turn on the power supply through its mains switch with the device under test connected.

So you can't connect anything with the power supply turned on, but you can't turn on the power supply with anything connected to it either? I'm pretty sure this boils down to never use your power supply. I don't see a problem with turning a well designed power supply on with a load connected, and I think the LED might still be alive if the output was turned off before connecting the LED (since the cap might be discharged). Of course it's good to be aware of design flaws like turn-on spikes, but I prefer to use decent power supplies without these problems.

[IanB]

• Never connect an energized power supply output to a device under test. Always wire up the device with the power supply off.
• Never turn on the power supply through its mains switch with the device under test connected.

So you can't connect anything with the power supply turned on, but you can't turn on the power supply with anything connected to it either? I'm pretty sure this boils down to never use your power supply. I don't see a problem with turning a well designed power supply on with a load connected, and I think the LED might still be alive if the output was turned off before connecting the LED (since the cap might be discharged). Of course it's good to be aware of design flaws like turn-on spikes, but I prefer to use decent power supplies without these problems.

 

I'm giving advice to newcomers so they don't blow up their stuff by accident. If you are an expert with good knowledge of your power supply and the device being tested you can do whatever you wish 

What I mean of course, is not to suddenly apply the full operating voltage to an unknown or untested device. Set the voltage to zero, set the current limit to the lowest sensible value, connect the device under test, then slowly wind up the voltage. If you hit the current limit before you reach the target voltage you may have a fault or a short and you can stop rapidly before damaging something.

Also, power supplies with a dedicated output on/off switch ought to have a controlled ramp up to the operating point when this switch is used. The same may or may not be the case with the mains switch, depending on the quality of the power supply. So whenever you have a new power supply or other item of test equipment in your hands you should test it thoroughly and become familiar with its operating characteristics before using it to do real work.

[Matje]

...

• If the power supply has an output on/off switch, use that to energize and de-energize the output once you are comfortable the settings are correct.
  

Beware! Only do that if you have checked that the power supply doesn't do something nasty - some chinese devices will do really bad things.

See the attached image for what a RG-303Pro (sold under different brands) will do on its main outputs on switching them off, this is for 3 volts output. This happens even with really high load. Mine has a sticker now: "Before switching off turn current limit down", it's the only way to avoid frying something. Needless to say I try to always remember to disconnect stuff before switching outputs off, just to be sure.

[tszaboo]

Is... is this a joke?

No I am asking it seriously. Blowing up stuff for fun is fun.

[EvilGeniusSkis]

Is... is this a joke?

No I am asking it seriously. Blowing up stuff for fun is fun.

I once popped a 6mm TH LED by connecting it to ~3V (2 AAs). don't know if i connected it backwards though.

[Jon Chandler]

I confirm your conclusions though:

....

• If you are testing a device for the first time, start with the output voltage at zero and gradually ramp it up to the desired value  while monitoring the device current. Turn off immediately if anything doesn't look right.
  
  

....

.

This seems like a good idea but I learned the hard way it's not.  I was testing a 3 watt LED with a constant current switching supply module.  I wasn't sure exactly how this was going to work, so I followed the start slow plan...and promptly blew up the power supply module!

The module was an eBay special, so perhaps it lacked some proper safeguards.  Think about what happened.  The module is designed to output 300 mA at whatever voltage (over some limited range) is needed to maintain that current.  If the supply voltage is lower, the switcher is going to draw a lot of current in an attempt to generate the needed output.  Enough current to let out the smoke in my case.

After a few cuss words and a little thought, I connected a second module, applied the recommended voltage and it worked perfectly.

[rexxar]

Is... is this a joke?

No I am asking it seriously. Blowing up stuff for fun is fun.

I once popped a 6mm TH LED by connecting it to ~3V (2 AAs). don't know if i connected it backwards though.

I bought some remarkably cheap LEDs to give to the kids in my electronics class. We were all very surprised when they blew up on 6V. The kids seemed to think it was rather exciting!