Announcement

Collapse
No announcement yet.

repairing corn cob led bulbs

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #31
    This is very likely the way the LEDs are connected:

    Click image for larger version

Name:	LED_Array.jpg
Views:	57
Size:	309.1 KB
ID:	1905282
    http://pauleschoen.com/pix/PM08_P76_P54.png
    Paul , P S Technology, Inc. and MrTibbs
    USA Maryland 21030

    Comment


    • #32
      That setup does not seem to agree with the earlier statement that an open causes the whole thing to shut off.

      In the suggested circuit, there are so many parallel paths that the open would have to be in the copper connecting one bank to the next, or in the power connection. Depending on the actual physical layout, it may or may not actually be possible in such a way as to cause an overall open.

      The suggested circuit has 27 LEDs in series, and 5 parallel paths, interconnected such that a single open LED would not shut off ANY others.
      1601

      Keep eye on ball.
      Hashim Khan

      Comment


      • #33
        If you look closely you will see that D1-D5 is one bank of five in parallel, connected to another (D6-D10), and a third (D11-D15). I should have shown a single connection between banks, but the principle is the same. As I explained earlier, the failure mechanism is that one LED, say, D1, fails open. The current is now carried by D2-D5, and now instead of 20 mA, they each carry 25 mA. Thus they are more heavily loaded, and the weakest one, say, D2, will fail open. Now D3-D5 each have 33 mA. When D3 opens, D4-D5 see 50 mA, which quickly blows one of them, and the remaining 100 mA through D5 pops it, causing an open string. If the LEDs are not well matched, or at different temperature, the current imbalance will be worse, and self-destruction will be accelerated.
        http://pauleschoen.com/pix/PM08_P76_P54.png
        Paul , P S Technology, Inc. and MrTibbs
        USA Maryland 21030

        Comment


        • #34
          I saw how they are connected.

          The theory then is that the first failure leads to a "zipper failure" of the others. That may or may not actually happen, as the first failure may not automatically cause a second. Subsequent ones are more likely, due to larger increases.
          1601

          Keep eye on ball.
          Hashim Khan

          Comment


          • #35
            If there are actually parallel groups of five, and there have been several failures where all five have opened and caused total failure, this "zipper" failure mechanism seems to be confirmed. If the blown LEDs are identifiable by the dark spot noted by the OP, it should be possible to see other groups with one or more failed devices. If the failures always occur in a group at the top, bottom, or middle, it would point to possible manufacturing defect or heat concentration as a failure mechanism. But I think once more than one LED opens, the "zipper" will soon cause the rest to go.
            http://pauleschoen.com/pix/PM08_P76_P54.png
            Paul , P S Technology, Inc. and MrTibbs
            USA Maryland 21030

            Comment


            • #36
              im surprised that apparently not many people use these lamps. they used to be under $5 and are currently a little over $5. as mentioned they are roughly equivalent to 150w halogen and you get 6500k. what is the alternative for e27 sockets? simply nothing, at least over here. incandescents only come up to 40w i believe and leds have a few watts and give no light. crazy world. i have about 20 of these and similar installed and about a dozen have failed over time. no, they dont have a life span of 100 years. there are also"15w" spot lamps that are great, similar to the 50w/12v halogen spots.

              new info:

              - brown cap is 225j 400v, what is that 2.2 µf?
              - lamps show a consumption of 10-12 w on the digital watt meter.
              - there is no electronics in addition to whats on the picture. (a rectifier, two caps and two resistor as far as i can tell.)
              - the leds indeed fail in groups of five. mostly the whole array is dead, but there are cases where one group of five is dead and the other two goups work (in that case the lamp works).
              - the connections to the arrays are burrier in the round boards and are not visible.

              of course i have cut the lamps open, where else would the arrays i use to bridge the faulty ones come from?

              paul, the arrays themselves are as you show. but there are 11 and while it seems logical to assume they are all in series (because lamp stops working if one is dead) i should be able to measure steadily increasing voltages at the connection points which is not the case. the max i see is 66vdc as mentioned. so i really wonder how it is.

              a cree 5730 is specified as: 0.2w, 65 ma, 2,8v, 37 lm, 210 lm/w (why doesnt it add up?). so 11w/165 = 0.07w, 2000lm/165 = 12lm and 182lm/w are plausible values.

              and again: why cant be the leds be fed ac voltage? would make it simpler, no?
              Last edited by dian; 10-17-2020, 11:17 AM.

              Comment


              • #37
                The problem with failure of the OP's corncob is that they often over volt them to get big lumens out of them. You can see the brown spot on one of the LED's in his pic.

                Then the arrays are so close together they have no heat sinking and no air flow and subsequently cook. I've liked them for the output and of course have had them fail. On one enclosed lamp I took the plastic globe off of a couple of them for cooling and they lasted maybe 6 months longer.

                Comment


                • #38
                  I've been using Cree lamps ONLY for several years (since they came out) with zero failures in what is now perhaps 20 of them of various styles (they have evolved). The only bad one is bad only because I dropped it and the (then) glass enclosure broke. The silicone coating kept the pieces in place, but it cannot be screwed into a socket..

                  Small sample, but the previous 20 or so CF bulbs would often have failures within months, sometimes weeks, especially the Feit" ones. And those failed shooting out sparks. I posted pics of the exploded inside components years ago.
                  1601

                  Keep eye on ball.
                  Hashim Khan

                  Comment


                  • #39
                    Originally posted by dian View Post
                    sorry, no, they produce around 2000 lm, are as bright a 150 w halogen bulb. thats why i have been using them for several years now.
                    then something that you measured is wrong - either the layout or the current per string. If they're being driven at 105mA per string of 15 LEDs (3 series, 5 parallel) and you have 12 strings, then using the lumens/ watt of those LEDs (http://led-obzor.com/led/smd-led-352..._of_LED_diodes) you get at most 1000lm from that power consumption. There's no magic there, it's pretty simple stuff.

                    I'll also point out that our eyes don't perceive light intensity in a linear manner - a light that produces 1/2 the output is perceived at about 80% or so of the brightness.

                    Comment


                    • #40
                      no idea if my lux meter "perceives in linear manner", but its consistent. 0.105a @ 100v = 10.5w an thats what they consume. at crees 210 lm/w you get 2200lm. the cree is measured at 65 ma, btw, and you can overdrive them to 240 ma.
                      Last edited by dian; 10-17-2020, 01:56 PM.

                      Comment


                      • #41
                        you don't have Cree LEDs in that light, I can guarantee you of that. If it makes you happy to think you got a 2000lm 10W light for $5 knock yourself out, no skin off my nose

                        Comment


                        • #42
                          I very much doubt these are genuine Cree LEDs, so all bets are off. It is possible for an LED to fail shorted, which would allow the remaining clusters to remain lit. If they are indeed rated nominally at 65 mA they should not be overloaded until three or four have burned out.
                          http://pauleschoen.com/pix/PM08_P76_P54.png
                          Paul , P S Technology, Inc. and MrTibbs
                          USA Maryland 21030

                          Comment


                          • #43
                            who said they were cree leds? if they were they would put out 165 x 37 = 6'105 lm and 33w if driven @ 65 ma.

                            matt, your repeatedly saying those bulbs are yttihs. so what do you suggest as an alternative for strong lighting in regular lamp sockets (e.g. e27), preferably daylight colour?

                            Comment


                            • #44
                              With a 2.2 uF capacitor I find about 24 mA through each LED and 128 mA RMS input.With 96 volts on the LED array, that is 12.3 watts. At 210 lumens/watt that would be almost 2600 lumens. But Matt's link shows more like 100 lm/watt, so more like 1230 lumens. I question the 210 lm/w figure. However, Cree claims to have achieved 303 lumens/watt in 2014.

                              [edit] As to why the LEDs aren't driven directly from AC - they emit light only in the forward direction, and cannot withstand more than a few volts when reverse biased. With essentially 27 in series, anything more than 120 volts or so would cause damage. And they would light only during the positive portion of the input AC. It might be possible to add a rectifier in series to block the reverse voltage, and you could apply twice the current at 50% duty cycle. But then there would be DC current drawn from the supply, and you could not use a capacitor as a current limiting ballast.

                              You could use two strings connected in anti-parallel, such that each would draw current during alternate polarity of the AC voltage, but that may result in visible flicker. The present design is probably about as good as can be with only cheap passive components, but the series capacitor may cause high current surges when AC power is first applied to the bulb, especially if it occurs at a voltage peak. The 4.7 uF capacitor across the LEDs may absorb the surge, and there appears to be a power resistor of maybe 300 ohms in series that will limit surge current to an amp or so.
                              Last edited by PStechPaul; 10-18-2020, 04:03 AM.
                              http://pauleschoen.com/pix/PM08_P76_P54.png
                              Paul , P S Technology, Inc. and MrTibbs
                              USA Maryland 21030

                              Comment


                              • #45
                                i have connected leds to ac before and they ran without flickering (or maybe im not sensitive to it). they were not power leds though. i mentioned the crees only to show that the output was not impossible. your calculation supports my measurement.

                                i have one lamp with a more sophisticated driver, but the sophistication didnt help much, it burnt out after a few week (thats why i got not more of this type). currently im buying lamps for $7 that consume 22w and really have double the output of these. they will light a room pretty well.


                                Click image for larger version

Name:	0 118.jpg
Views:	25
Size:	2.84 MB
ID:	1905529
                                Last edited by dian; 10-18-2020, 05:10 AM.

                                Comment

                                Working...
                                X