Connect with us

LEDs LEDs LEDs......Help

Discussion in 'LEDs and Optoelectronics' started by varkpos, May 14, 2010.

Scroll to continue with content
  1. varkpos

    varkpos

    13
    0
    May 14, 2010
    Evening all,

    This might seem simple, but I have to control a parallel array of leds via a Arduino. My challenge is with the circuit before an input form the micro controller.
    All the leds are 0603 and soldered onto a knitted wire mesh, this in turn is then casted into resin blocks. (Wifes Uni Art project, dont ask) There is 22 of these in total and 9 has been made. Each consist of 7 leds connected in Parallel without serial resistors.

    My fault, I did not add any resistors to the leds, and I know this is wrong! Space is/was an since I discovered this I was planning on joining the 0603led with a 0603 Resistor, then redo all.

    However supplying the 7 led array with a 3.3v power source from a regulated source, and the fact that the leds is casted into resin seems to provide a stable scenario where it follows the operating curves of the leds. Tested this and left it on for 3 days. No thermal runaway. Ie burnt leds.

    Right now with that out of the way, what I want to achive is, to turn on and off, ie flicker this 22 units at various intervals with the arduino. Thus my thoughts is to add a transistor or mosfet to switch the ground on each casted object, by an input voltage from the arduino.

    My question is:

    Could someone please assist me in selecting the correct transistor or mosfet for this implementation. Think the ones I have , the 3v supply is too low.

    Then to determine the best operating voltage, 3v seemed to work fine and my thoughts were to keep the voltage low, for heating, and thermal runnaway. Might even use multiple 3v supplies, to ensure I am on the safe side, and connect the resin encasted objects with 7 leds each, 3 or 4 together to a wall wart.

    Then what mA my supply should be. Eash time I think I have a winning circuit someting changes and all is back to the drawing board.

    My thoughts was that a mosfet would draw less current, as I don't want to over supply the resin casted led arrays.

    My components is
    LED's http://www.farnell.com/datasheets/80903.pdf
    Mosfet http://uk.farnell.com/jsp/search/productdetail.jsp?SKU=1653658

    Any help or assistance would be much appreciated, as I need to get this project on the way. If your advice is that this is not going to work then I will have to start over, adding all the LEds ect.
    This is currently driving me crazy, and need to have the finished for an exibition in less than 3 weeks.
    Thank you in advance.
     
  2. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    25,412
    2,780
    Jan 21, 2010
    Is there a series at all when you power those LEDs from 3.3V? If not, then you are very lucky, but if it worked for 3 days then I'll go with the empirical evidence -- just make sure the regulator is stable!

    What current is drawn by these strings of LEDs?

    It sounds like you need a mosfet as the additional voltage drop across a transistor may impact the LED brightness significantly.

    I'd go for a logic level mosfet. Many of those start to turn on with a couple of volts on the gate and may work. If you have access to a higher voltage rail (say 5 to 10V) then using the Arduino to switch a transistor which in turn provides gate drive via the higher voltage rail may be a good option.

    The latter option would be required for those mosfets as Vgs(th) is around 4V

    A BSS138 would be a better option as its Vgs(th) is between 0.8 and 2V. Even with this, the Vds may be higher than you want.

    On the other hand, there are also some very low Vce(sat) transistors.
     
    Last edited: May 14, 2010
  3. varkpos

    varkpos

    13
    0
    May 14, 2010
    Steve, i am not sure what you meant with your first question but I will elaborate.
    I added a 100Ohm resistor, in series to one end of the string of 7 leds connected in parallel, this was powered by the 3.3v output form the arduino.

    I am not sure, as I have not measured this yet. I will try tonight and let you know.

    My thoughts were to use the one's I have already made in the setup with a 3v or 3.3v source, then build the rest with a resistor in series to each led, the way this should be done.

    Then I can run them of a normal 5v rail for the ones with the resistor in series.

    Would the parallel leds without the resistors, work the same on a 5v rail, if I raise the 1 series resistor in relevance to the voltage change from 3 to 5v, or should I stay with 3v on the ones I have already made.
    My main concern is what current I would need to power the 3v objects casted without a resistor? Might have to split them up, ie power form different sources.
    Any advice.
     
  4. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    25,412
    2,780
    Jan 21, 2010
    Sorry, I missed out the word "resistor" from that question.

    If the LEDs worked from a 3.3 volt source for quite a while at an intensity that was acceptable, then I see no reason why you can't run them from a higher voltage with a suitable series resistor to limit the current to the same value.

    If you construct the new units to have suitable series resistors to allow each LED to operate from 3.3V, then it would also be possible to add another series resistor (a single one in series with the module) to allow it to operate from 5V. Indeed, you could build this into the epoxy module so that you have a ground connection, a 3.3V and a 5V connection and get the best of both worlds (that is more difficult if you're epoxying the wires in place, in which case, employ the resistor at the other end of the cable).
     
  5. 55pilot

    55pilot

    434
    3
    Feb 23, 2010
    While it is hard to argue with a successful test, you have to know the difference between getting lucky and successfully testing. You also have to realize that your test is valid only when it is done under a realistic condition. I am afraid your test is not valid.

    Your description is scattered. Putting things together, I am assuming that for your test you had 7 LEDs in parallel, connected to a 100 ohm resistor, running from 3.3V.

    Assuming a LED forward voltage of 2.7V, you were dropping 0.6V through the resistor, running 6mA through it or about 1mA through each LED. At such low currents, if the LEDs are well matched, things would work, as long as the LEDs stay at the same temperature. If one of them heats up, its forward voltage will drop, it will hog most of the current and it will get noticeably brighter than the others. At those currents it is unlikely that there will be a runaway, but I can not be sure of that. Did you intentionally heat one of the LEDs to see what happens?

    If you were running at higher currents, say 5mA per LED, you will definitely get a runaway and a failed LED as soon as one of the LEDs got warmer than the others. Cut the resistor back to 10 ohm and re-run the test. Then heat up one LED and see it get brighter and brighter and then fail.

    If you keep running the setup for several weeks and months, the LEDs will age differently and one of the LEDs will hog more current, leading to a brightness mismatch at low currents and a runaway at higher currents.

    If the LEDs in a group happened to be from different bins or even different batches, they will have different forward voltages, leading to hogging, current mismatch and eventual runaway even at low currents, especially if one LED is a lower bin # than the others.

    Bottom line is that despite your "successful" test, your setup has no hope of working reliably in the long run. You may avoid your wife's wrath in the short run by using what she has created, but in the long run you will end up paying for it when the LEDs do not work reliably, which is inevitable.

    I would suggest that you fess up, explain the problem, throw away what is done (or modify if you can), take her out for dinner to make up for your mistake because it is always the guy's fault (wonder how lesbian relationships handle it), and do it correctly by either having the LEDs in series (preferred) or in parallel with individual series resistors. The reason putting them in series is preferred is because the current through all of them will be equal and they will appear to have the same brightens (assuming they are from the same intensity bin) which I think is important for your requirement.

    I know this may be a bit more technical than appropriate, but in the past Steve has volunteered to translate it into human speak :)

    ---55p

    Edited to add: I just noticed that your test was done with "3.3v output form the arduino" That may actually be less than 3.3V which means your LEDs may be running at much less than 1mA which makes your test even less valid.
     
    Last edited: May 14, 2010
  6. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    25,412
    2,780
    Jan 21, 2010
    Always a pleasure 55p :p

    Actually, at that low current, as long as the maximum forward current of the LEDs is not exceeded if one LED takes ALL the current, no damage will result. You'll just have one very bright LED and 6 dim ones.

    I presumed your test was done at the current you would use, and by encapsulating the LEDs in epoxy, it is likely that only self-heating will be an issue, the mass of the epoxy should ensure any reasonable external heating affects all LEDs.

    As 55p suggests, low currents tend to cause less problems, and this is due to the smaller amount of self-heating.

    If you got all of your LEDs from a single strip of tape, then it's also likely that they will be fairly closely matched (certainly more than LEDs picked at random.

    Or don't (if the current you're running these at is sufficient).

    Now, one has to ask how long will this piece remain active. Is it an "installation" that will be shown once, broken down, and possibly not shown again (or if shown again there will be opportunity to repair)?

    Are these LEDs on continuously, or is the duty cycle low? Is it switched off at night?

    All of these factors will mitigate the potential problems.

    My impression is that the incorrectly wired units are a small number of prototype blocks, and that there are more to be made. Depending how these are being made, and whether they're all identical, it may be possible to make extra and render the issue moot.

    If that is not possible and the current ones really need to be used, I'd try to use them at the lowest duty cycle possible for the application.

    It would also be nice to get confirmation that your tests were realistic, i.e. at the full brightness that will be used.

    I am not quite as pessimistic as 55p, but we all know that I'd also prefer to see individual resistors on each LED.
     
  7. 55pilot

    55pilot

    434
    3
    Feb 23, 2010
    What you call pessimistic, I call pragmatic and realistic.

    If one wants to design reliable things that will work every single time, one needs to deal in the hard cold reality of worst case parameters and plan for the undesirable but possible scenarios.

    Optimism causes too many problems and invariably leads to unreliable designs. I avoid it like a bad venereal disease :D

    ---55p
     
    Last edited: May 14, 2010
  8. varkpos

    varkpos

    13
    0
    May 14, 2010
    Gents,

    Firstly thank you for all your advice, it really helps more that you can think, and I am learning a lot.

    I followed the advice and started to solder all the new arrays as it should be. This took quite a bit of time, and I was well pleased with myself for purchasing the thinnest 0603 LEDs on the planet. Days later, he he, I had all the 0603 LEDs soldered directly end to end to the 0603 resistors and this serial marriage connected to the various spots on the array.

    Tested it all and it works fine. I can mention that the intensity of the led is much more that with the 3v setup with no resistors.

    To answer some of the questions. The 22 arrays consisting of 7 leds each will be controlled by a arduino (this I still need to configure and figure out the code completely) triggering a mosfet on and off at high speeds to form a pulsating effect on the arrays, each of these will be random over the arrays, lighting them up form dim to full about once a min. Thus the intensity on each array will be very low. I am also planning to add a timer switch to the piece that will deactivate all at night, and start up again at the appropriate times.

    (Still have the 8 old arrays and will have to see if my wife can redo the mesh to rebuild them too. If not I might try to use them at 3v and keep the duty cycle on them even less, than the rest. Almost nothing.)

    No I started to do some measurements to post, as per some of the questions, but my reading must be wrong, or my dig meter is on the way out. Brought some test samples into work and will try a meter here to compare.

    Main source voltage : 5.01v (ATX power source for testing, will be a wall wart on implementation) Still need to determine Amps required.
    0603 LEDs http://www.farnell.com/datasheets/80903.pdf
    LED Fv 2.9
    9-10mA
    0603 Resistor 220Ohm
    LEDs in array 7.

    Measuring the current with the Multi meter in Series to the circuit it reads 135mA. Could this be possible? This must be wrong.

    Connected up one LED with a 220 R and measured the current on that.
    Apparently the voltage measured across the single led is 3.8v.
    Current measured by the 1 LED and 220 R is 50mA 1?

    This should be way above the limits of this LED.

    I connected up the mosfet. http://www.vishay.com/docs/91015/91015.pdf
    And it works perfect to switch the led on and off. First tried with a single led, and then with the full array of 7 leds( with R) using the same 5v form the PS to trigger the mosfet with a 1kR to ground.

    D – LED – 5v
    S -- GND
    G – 1KR to ground
    I_ Swithch to 5v

    My main question is what current would the mosfet use to switch this of and off? I know the arduino can muster about 40mA per pinout.

    Looks like all would be well, only need to sort out the required current.

    Thanks
     
  9. varkpos

    varkpos

    13
    0
    May 14, 2010
    Confirmed the Current reading with a different meter.

    Single LED and 220 Resistor on a 5v Supply :54.7mA
    7 Led + 7 R220 Array on a 5v Supply : 163mA

    This is much more that I anticipated. Well, The figures must be correct?

    I would then need a hefty power supply or 2 - 3 seperate supply's

    Any thoughts
     
  10. Resqueline

    Resqueline

    2,848
    2
    Jul 31, 2009
    If I understand this correctly you have 1.2V drop over 220 ohm resistors. That calculates to 5.45mA. Are you sure you didn't get 22 ohm resistors? What're they marked?
     
  11. varkpos

    varkpos

    13
    0
    May 14, 2010
    This would be a great time to swear if I was a sailor. I never measured the resistors. They are indeed marked 27x. Looks like they sent me 22 ohms instead of the 220ohms I ordered.

    This just gets better and better.Looks like I will have to start again....
     
  12. 55pilot

    55pilot

    434
    3
    Feb 23, 2010
    Did you order 220 ohm or part number 220? Resistor part numbers are two digits followed by a 3rd digit indicating the number of zeroes.

    The part number 220 means the value 22 followed by no zeroes or 22 ohms.

    The part number 221 means the value 22 followed by a single zero or 220 ohm.

    The part number 223 means the value 22 followed by three zeroes or 22000 ohm or 22K ohm.

    I have a hell of a lot more experience than you and despite that I never built more than one of anything. I always build one and test before making more.

    I am still not very clear what exactly the device is. I understand that you have an 0603 LED and a 0603 resistor directly soldered together and encapsulated in some resin. How much load does the solder joint have? Will the resin support the joint firmly or will it allow bending loads to get to it? If the joint is not supported and kept rigid, it will fail. Same for the joint between the wire and LED or resistor.

    ---55p
     
  13. 55pilot

    55pilot

    434
    3
    Feb 23, 2010
    If the voltage is actually 5V, then we have a 2.2V drop over the resistor. If the resistor is 220 ohm, the current should be 10mA.

    If the current reading is correct, the diode forward voltage will likely increase to 3.3V (extrapolating from figure 2), the voltage drop across the resistor will decrease to 1.7V and the resistor will have to be 31 ohm.

    If the extremely long thin wires are used, the wire resistance plus a 22 ohm resistor can come close to 31 ohm. Add to that a power supply that is below 5V and the theory that the resistors are 22 ohms becomes plausable.

    ---55p
     
  14. varkpos

    varkpos

    13
    0
    May 14, 2010
    55p,
    I ended up seeing what I did wrong and you are correct.
    The part number was 220, and on the resistor 22X. I never measured the resistor.
    When I did the first test, I never measured the current. The leds was shining bright, but I thought that was do to being close to the 10mA range. All is still working but looks like I have gone from bad, 3v leds with no R. to worse. 5v Leds with 22Ohm Resistors. Great.

    Well the good thing is this gives me the chance to redesign my scope and change my power source. I had to use 2. One for the leds and one for the arduino. Might just as well up the input voltage to 12v use a 1K2 Ohm resistor. Will definitely get the correct ones this time.

    This nightmare is for an art piece. The 7 led array and resistors is in a 16x8x2cm clear resin block. Luckily I haven't casted them yet. That would have been the end then, as it takes a lot of time to make the initial structure to host the leds.
    The joints are solid as they are in the middle of loads of resin, this also keeps the temp way down on the leds.

    As it were there weren't going to be time to redo the 8 I have casted without the resistors in the led array. So I will just have to chance it on a 3v supply and keep them at a very min duty cycle, and if they die, then those 8 will just be without led or less.

    Can any one recommend a good transistor that can lower 12v to 3v that I can power those 8 with. maby 3 or 4 should cover the current that they would draw.

    Must admit this is a bit of a blow, but I will just have to order the new R will have them in the morning, and then its All night trying to make up for this.

    Would also have to get different mosfets, as the 3v may be to low for the ones I have. Sigh
     
  15. varkpos

    varkpos

    13
    0
    May 14, 2010
    The wire mesh, which I connect the LEDs to is 0.2mm coated wire, and is a REALLY long.
    Thus the figures could be correct.

    I did measure a 3.8-3.9v over the one led I tested once I saw the current readings. All ads up. The surprising thing is the beating that these leds took. Ive been soldering those that I did, and tested each array, while I am soldering the next one. Then I also left one LED burning at that rating for, the whole of yesterday an last night. Thus I thought all was correct.
     
  16. Resqueline

    Resqueline

    2,848
    2
    Jul 31, 2009
    I won't claim to have the full view of what you're doing here but if you just put a diode (or two) in series with the 5V supply that should lower the current to a safe level.
    I don't understand where 55p got the 2.2V resistor drop from (what it should have been?). I'd say you'll need around 90 ohms (82 or 100 depending on final Vf) to get around 20mA.
    55p adressed the mechanical strength due to resistor & LED being soldered together. It's a very valid point. The end plating on the components will crack off if there's any stress due to bending forces. Will the resin be supported or could it be exposed to bending forces?

    I just got a cheap ($3) LED flashlight (SKU 2044) off u-know-where. It has 12 5mm LED's and runs on 3 AAA batteries. The thing has NO resistors for the LED's, draws 900mA (4W). That is 75mA per LED (average), or 3 times overcurrent.. It'll be interesting to see how many battery sets the LED's will go through before starting to fail.
     
    Last edited: May 18, 2010
  17. 55pilot

    55pilot

    434
    3
    Feb 23, 2010
    5V - 2.8V for the average LED Vf (per the linked datasheet) leaves 2.2V for the resistor. The 2.8V number is a middle of the road guess for Bin #2. It could be higher or lower if Bin #1 or Bin #3 were supplied.
    I have posted several times before the some of the LED flashlights out there are absolutely horrible designs. It is so bad that last year one of the major LED manufacturers told me that they were considering restricting sales to only qualified companies because of the horrible reputations LEDs were acquiring due to these bad designs.

    ---55p
     
  18. varkpos

    varkpos

    13
    0
    May 14, 2010
    This wont be a problem, as this is basically resin Blocks, thus none of the parts can move.

    Thanks 55p for helping me in realizing the mistake I made, received the new components today and will beaver away tonight to connect all up, will test and post my results.

    Any advice on converting 12v DC to 3v DC, current needed would be around 500mA so would design it to have a bit more spare. This would be form the new 12v 5A supply that would run
    14 of the arrays and the Arduino, then the 3v would supply the remaining 8 supply’s at 3v.

    The 3v regulator I used for testing was part of a stepper motor controller and don’t want to keep that in the project.
     
  19. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    25,412
    2,780
    Jan 21, 2010
    If you're running the LEDs from 12V then you are far better off putting several in series rather than all of them in parallel.
     
  20. varkpos

    varkpos

    13
    0
    May 14, 2010
    Steve,
    I did see that from the led calculator that I used, although due to the nature of the wire mesh, actually "knitting" and the location of the LEDS on the work this would be close to impossible to connect them in serie.

    I have completed the objects, and all is working fine. Tested it, correct current and all looks dandy. Will cast the object in resin this afternoon, and then work on the controller.
    Will probably start a new thread, regarding the controlling.
     
Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day

-