Page 1 of 3 123 LastLast
Results 1 to 50 of 144
  1. #1
    mtbr member
    Reputation:
    Join Date
    Oct 2008
    Posts
    1,309

    MagicShine 1400 Deconstructed


















    XPE pod outer diameter

    MS900 vs MS1400

    MS900 vs MS1400

    MS900 vs MS1400

    Pod ends threaded and removable.

    Optics/reflector P7 retainer ring interal and external threaded. Reflector rim threaded and screws
    into retainer ring.

    XPE optic

    XPE optic

    XPE optic retainer ring inner diameter

    MS900 vs MS1400



    XPE Pod, ring, slug, xpe

    XPE Pod, ring, slug, xpe

    Remote Switch Aluminum Body. Cap housing Aluminum threaded. Seperate driver component

    3x XPG Carclo 10417 860mA

    MS900 P7 High

    MS1400 Dual XPE only High

    MS1400 P7 only High

    MS1400 P7 and dual XPE High
    Last edited by odtexas; 03-19-2010 at 08:50 PM.

  2. #2
    mtbr member
    Reputation:
    Join Date
    Apr 2006
    Posts
    596
    Thanks, Texas, I wuz wondering about that mouse eared light. I enjoyed the disassemble. How was soldering, swarf, your over all impression.

  3. #3
    mtbr member
    Reputation:
    Join Date
    Oct 2008
    Posts
    1,309
    No swarf. Solder was first rate. Thing is extremely well sealed. Thermal paste was a little thin on the xpe slug. Must desolder to remove and add more paste. PITA. Probably not necessary, but I already had the thing apart.
    It is much more complicated than the MS900.
    Body seems to be single unit cast aluminum. Pods connect via 18 mm by 2.5 mm truss. Not going to break them off very easily.
    Last edited by odtexas; 03-19-2010 at 09:09 PM.

  4. #4
    mtbr member
    Reputation: yetibetty's Avatar
    Join Date
    Dec 2007
    Posts
    1,781
    Thanks odtexas for doing all those pics, it does look like good value and not as big as I thought it would be. Is the inductor bonded to the pcb or just soldered with no suport like in the 900?

  5. #5
    mtbr member
    Reputation: znomit's Avatar
    Join Date
    Dec 2007
    Posts
    1,837
    Thanks very much for this! Any measurements of drive current?

  6. #6
    GeoMan
    Reputation: GEOMAN's Avatar
    Join Date
    May 2006
    Posts
    1,706
    Great thread and effort! Very impressed.

    Appreciate it,

    Geo
    GeoMan
    We Ride!
    www.geomangear.com

  7. #7
    mtbr member
    Reputation:
    Join Date
    Oct 2008
    Posts
    1,309
    Yeti - Inductor is about twice as big and is still unsupported. A blob of silicone might be nice for support. It does feel alot more solid than the MS900 inductor.

    Znomit - Forgot to grab currents when I had it all apart. I am hoping to switch out the XPE optics so will be taking it apart again real soon. Will get current readings at that time if someone doesn't beat me to it.

    No worries Geoman. This light is very well thought out. Quality control is top notch. Looks like a real PITA to assemble and wire. Heads and shoulders over the MS900.

    Mount is different material/plastic, and design of mount looks stronger.

    Only shortcomings was small amount of thermal paste on XPE stars.
    Allen set screws on bottom of pods, control pod rotation, both stripped out.
    The cast aluminum is just not strong enough for that few number of thread. Not a big deal if you aren't planning on taking the thing apart alot. But I could see that maybe someone might recieve one with the pods being a little loose in their collars and the customer would be unable to tighten. I just pushed some JB Weld into the set screw hole, let it set a little, and then put the set screws in through the JB Weld. Holding just fine.
    Water can get in through the small opening between the main body and the XPE pods where the wires leave main body and run to XPE stars. Only way to truly seal it would be to silicone the small rubber elbows which the wires are run in.
    If it wasn't for those little wire pass throught this light would be water tight to a couple of meters. All optics and pods have o-ring seals.
    Last thing would bet the star retaining rings are plastic. MS900 used alumimum rings that threaded in to hold the stars down. Was mildly surprised to see plastic ones in the light. Not a problem for your average user of course, but the aluminum ones would aid in thermal transfer from the front surface of the star to the light body.
    I still think this light has more than enough surface/transfer area to keep it cool.

    Fit and Finish is excellent. I was impressed. Like the sealed battery unit as well.
    And it is the same battery as the MS900. Just a new and improved housing.

  8. #8
    GeoMan
    Reputation: GEOMAN's Avatar
    Join Date
    May 2006
    Posts
    1,706
    Quote Originally Posted by odtexas
    No worries Geoman. This light is very well thought out. Quality control is top notch. Looks like a real PITA to assemble and wire. Heads and shoulders over the MS900.

    Fit and Finish is excellent. I was impressed. Like the sealed battery unit as well.
    And it is the same battery as the MS900. Just a new and improved housing.
    We had to establish a new production line to handle the added complexities of the Magicshine 1400 assembly. QC is being watched very carefully.

    Geo
    GeoMan
    We Ride!
    www.geomangear.com

  9. #9
    Rolling
    Reputation: lidarman's Avatar
    Join Date
    Jan 2004
    Posts
    11,110
    What is the capacity of that battery pack?

  10. #10
    mtbr member
    Reputation:
    Join Date
    Oct 2008
    Posts
    1,309
    Battery is a 7.4 v lithium ion 2p2s. Capacity probably around 4400 mAh if using cheap 18650s.
    Best guess on optics are the Carclo 10197, Luxeon 1 20mm.
    Can someone confirm height of those Carclo optics. Can't find a data sheet on it. I have never used them in my builds and don't have any on hand to measure either.

  11. #11
    mtbr member
    Reputation: mojojojoaf's Avatar
    Join Date
    Sep 2008
    Posts
    761
    Spill and penetration look great on that last photo- not bad at all.

  12. #12
    mtbr member
    Reputation: znomit's Avatar
    Join Date
    Dec 2007
    Posts
    1,837
    Quote Originally Posted by mojojojoaf
    Spill and penetration look great on that last photo- not bad at all.
    With just the P7 on it doesn't seem to have the punch of the 900. Is it a different reflector?


  13. #13
    www.hahntronix.com
    Reputation: mhahn@hvc.rr.com's Avatar
    Join Date
    Mar 2007
    Posts
    252
    odtexas,

    Very nice initial dissection of the light.

    A few things I'm curious about, and some requests.

    Are there separate drivers for the XPEs? The could be hiding in the little black ring the XPE thermal clad board was attached to.

    What gauge wire runs into the main light body. The original magicshine used something like 22 gauge which is kind of thin, in my opinion. Can you tell if the wire is PVC or Silicone coated? Based on where the mode switch connects to the cable, I'd guess there are at least 3 and maybe 4 separate wires in the cable that runs into the main body of the light. My hunch is they are using thin wire again. Easy to break, and fatigues easily after a lot of flexing.

    It looks like there is only one inductor on the main electronics board. Can you tell if that board is only for driving the P7 or does it drive the XPEs as well?

    Does the battery come apart easily? I'm curious what the protection circuit board looks like and if the Ii-Ion cells look like they are good quality.

    Request #1: If your camera has a macro function, could you take a close up of both sides of the main electronics board. I'm curious to see if they used the same 6 pin driver chip with an external mosfet transistor, like they used in the original magicshine.

    Request #2: If you haven't siliconed down your inductor yet, could you bend it a bit one way and then the other when taking pictures of that side of the board? It would make more detail visible.

    Request #3: If there are separate drives for the XPEs a picture of one of those would be nice.

    Oh, don't use something like silicone window caulking to mount your inductor. As it cures it gives off a vinegar-like smell that comes from acetic acid. The acid can eat away your circuit board. Use a neutral-cure silicone if you can fine one. Something that claims it is good for coating electronics is best.

    Or if you intend to use some artica alumina expoxt adhesive to make better thermal connections inside the light, that would also probably work fine for holding the inductor in place.

    Thanks much,

    Mark
    Last edited by mhahn@hvc.rr.com; 03-20-2010 at 04:06 PM.
    Nimium est melior!

  14. #14
    mtbr member
    Reputation:
    Join Date
    Oct 2008
    Posts
    1,309
    Znomit -
    Reflectors are different.

    Left is MS1400 right is MS900
    MS1400 15.3 mm at star base external diameter to 33.21 mm at flange base
    MS900 17.94 mm at star base external diameter to 35.85 mm at flange base
    Both 19.75 mm high
    mA gets interesting though
    MS900 high 2.36 amps P7
    MS1400 high 1.95 amps P7 .12 amps low
    MS1400 XPE high .62 amps .02 low
    MS1400 XPE and P7 high, all on, XPE .52 amps, P7 1.71 amps
    Interesting isn't it??

  15. #15
    Rolling
    Reputation: lidarman's Avatar
    Join Date
    Jan 2004
    Posts
    11,110
    ....................

  16. #16
    mtbr member
    Reputation:
    Join Date
    Oct 2008
    Posts
    1,309
    mhahn -

    Are there separate drivers for the XPEs? The could be hiding in the little black ring the XPE thermal clad board was attached to.

    Can't tell if there are two drivers in there or not. The P7 star and driver are stacked together and can only be removed if all wires to both XPE pods are removed. There is room in there for components on the back side of the board and all wires go into the space between the back of the P7 star and driver.

    What gauge wire runs into the main light body.

    20 or 22 gauge silicone.
    Wire running to XPE pods probably 28 or 30 gauge. Very thin and easy to break. Another reason to not recommend taking it apart unless you are careful and good at soldering.

    Based on where the mode switch connects to the cable, I'd guess there are at least 3 and maybe 4 separate wires in the cable that runs into the main body of the light. My hunch is they are using thin wire again. Easy to break, and fatigues easily after a lot of flexing.

    Can't tell how many wires come out of the main cable since it exits into that small space I didn't get into.

    It looks like there is only one inductor on the main electronics board. Can you tell if that board is only for driving the P7 or does it drive the XPEs as well?
    Back side of board is a mystery.

    Does the battery come apart easily? I'm curious what the protection circuit board looks like and if the Ii-Ion cells look like they are good quality.

    Yes. The housing is actually aluminum. Threaded end cap. Guessing same exact battery as MS900



    Request #1: If your camera has a macro function, could you take a close up of both sides of the main electronics board. I'm curious to see if they used the same 6 pin driver chip with an external mosfet transistor, like they used in the original magicshine.

    Top of driver only. 2 black rectangle components have MCC SK548 on them.

    Request #2: If you haven't siliconed down your inductor yet, could you bend it a bit one way and then the other when taking pictures of that side of the board? It would make more detail visible.

    What I can get of the board






    Request #3: If there are separate drives for the XPEs a picture of one of those would be nice.

    Couldn't get that one.

    Hope some of that helps.

  17. #17
    mtbr member
    Reputation: znomit's Avatar
    Join Date
    Dec 2007
    Posts
    1,837
    Quote Originally Posted by odtexas
    MS1400 XPE and P7 high, all on, XPE .52 amps, P7 1.71 amps
    Interesting isn't it??
    So each XPE is putting out 160lm and the P7 500lm ...

    So ~800lm vs 670 for the 900. Bigger porkies.

  18. #18
    mtbr member
    Reputation: yetibetty's Avatar
    Join Date
    Dec 2007
    Posts
    1,781
    There is quite a lot going on in that switch, does this thing have lots of strobe modes?
    Attached Thumbnails Attached Thumbnails MagicShine 1400 Deconstructed-magicshine1400096-small-.jpg  


  19. #19
    mtbr member
    Reputation:
    Join Date
    Oct 2008
    Posts
    1,309
    No strobe modes. 4 little colored leds.
    First push XPE pods on
    Second push P7 on XPE off
    Third push all on.
    Holding button down in any mode causes gradual dimming.
    Neat feature if you are into that sort of thing.

  20. #20
    mtbr member
    Reputation: troutie-mtb's Avatar
    Join Date
    Sep 2007
    Posts
    3,060
    Nice pull down Odtexas , thanks , man you must have been fast getting hold of one .

    battery pack looks a huge improvement .

    It is not to my liking the mickey mouse ears but I guess it will sell in its thousands like its brother .

    cant they make a light that does what it says on the tin lumen wise .

    We should have a special section on here for comercial light strip downs and testing maybe that would scare the big boys in to telling the truth .

    bit of extra space could you slot in a small driver and push the xpes harder

  21. #21
    www.hahntronix.com
    Reputation: mhahn@hvc.rr.com's Avatar
    Join Date
    Mar 2007
    Posts
    252
    odtexas,

    Thank you very much for the pictures. You've saved me from wasting $90 and ordering a litehead one from the SingleTrack store. I wanted to see if the new light had a better designed driver than the magicshine.

    Your first picture of the P7 driver has convinced me that they are using nearly the same circuit that drove the magicshine light. The driver chip is most likely a sm5241, I can make out "524*" on the 6 pin chip visible in the bottom center of your first photo. Can you confirm that the part is labeled 5241?

    The transistor labeled Q1, on the PCB to the right of the SM5241, is most likely a mosfet in a SOT23-3 package. The magicshine used an 8 pin SOIC. I've used SOT23-3 mosfets in LED drivers I've designed, so that might be OK. But it is typically easier to find a low Ron (on resistance) mosfet in a SOIC package. A low Ron mosfet will not get as hot as one that has a higher Ron. Low Ron mosfets are typically more expensive than high Ron ones. Have you tried holding your finger on Q1 while the light is running? Be careful, it may be hot enough to burn you. If it feels painfully hot, it probably won't last a long time. It will run hotter the lower your battery voltage is.

    And to the left of the SM5241are two resistors labeled R100 and R500. They sure look like 0805 sized resistors, and can't be rated for more than 1/8 a watt (could be as low as 1/10 a watt). Gee, that's the same size and same value resistors used in the magicshine. So this light has the same problem. These resistors, used to sense current, have the same amount of current flowing thru the two of them wired in parallel as the P7 has running thru it. Power is calculated by the formula Current squared times Resistance. Assuming they run the P7 at something close to 2.5 amps, the resistor labeled R100 has to dissipate something like 1/2 a watt, and it's only rated for 1/8 of a watt. It will burn out soon enuff. If you are handy with a soldering iron, you can replace these with two 1206 or 1210 sized resistors, maybe around 0.16 ohms.

    It would be interesting to see what's on the other side of the PCB to see if they do any kind of thermal regulation of the LED's brightness. It'd be nice to dim the LED when the temperature gets too high. But given that folks on this forum have dissed the magicshine driver's electrical design and the manufacturer has made the same mistakes they made before, I very much doubt they do any thermal regulation.

    My guess is they are still cutting corners where most folks won't see it. They have some new cosmetic features, like the aluminum can to hold the Li-Ion cells. The nice thing about that is if they still use low quality cells and one catches fire, the aluminum can may help contain the fire. My impression is it's the same innards as the magicshine, with a slightly improved user interface, and a different beam pattern because of the 2 XPEs.

    These new lights come with the same 90 day warranty the magicshine did. Hey if it was a quality light the manufacturer would offer a longer warranty.

    Mark
    Nimium est melior!

  22. #22
    mtbr member
    Reputation:
    Join Date
    Oct 2008
    Posts
    1,309
    Troutie.
    The lumen thing is aggravating. Just part of the game. This light is bright and beam pattern looks good. Great light for getting into biking on the cheap.
    Would never consider putting this on my helmet just due to its size.
    But then again it only seems big because of the little lights that many of us build here.

  23. #23
    mtbr member
    Reputation:
    Join Date
    Oct 2008
    Posts
    1,309
    One chip does have 5241B on it.
    Q1 does get hot. So many burns from soldering though has desensitized me so I may not be a good judge of "painfully" hot. Unless the flesh sizzles I tend not to notice anymore.
    Major improvements over the MS900, both in quality and beam pattern.
    The battery container is also really nice.
    Just put this in the same category as a low end mountain bike. Gets the job done and gives you an idea if you enjoy the experience enough to get something a lot nicer.
    Last edited by odtexas; 03-21-2010 at 10:19 AM.

  24. #24
    mtbr member
    Reputation:
    Join Date
    Apr 2006
    Posts
    596
    Yep, I get a kick out of "painfully hot". My finger tips are so seared by years of holding parts while soldering, I have seen the puffs of smoke before the I feel the pain. lol Good write up Texas, thanks again. You SAVED me from having to go buy a 1400 to disassemble. James

  25. #25
    mtbr member
    Reputation:
    Join Date
    Jan 2007
    Posts
    752
    Quote Originally Posted by mhahn@hvc.rr.com
    odtexas,

    The transistor labeled Q1, on the PCB to the right of the SM5241, is most likely a mosfet in a SOT23-3 package. The magicshine used an 8 pin SOIC. I've used SOT23-3 mosfets in LED drivers I've designed, so that might be OK. But it is typically easier to find a low Ron (on resistance) mosfet in a SOIC package.

    Mark
    Later Magicshine 900's switched from the 8 pin SOIC mosfet to a smaller SOT23-3 mosfet as well. It doesn't have as low of a RDSon as the SOIC part and it does run hotter and makes the driver less efficient.

    I'm amazed they won't fix the under spec'ed resistors. It's not like this is really subjective. Any first year EE major could figure this out and the datasheet for the resistor gives very clear guidelines for what is in and out of spec. They aren't even close. What's crazy is the pattern on the PCB board is clearly setup for larger resistors and we're talking about parts that cost less than a penny in quantity. The one good thing about the low drive current on high is the resistors won't be as far out of spec. But they sure are giving up a lot of brightness out of the LEDs.

    Of course they haven't seen a lot of warranty issues with a 90 day warranty and this is the kind of problem that could become a serious issue when the lights are a year to two old. They may get away with them running over spec for a while, but why take a chance for something that is so cheap and easy to correct.

  26. #26
    mtbr member
    Reputation:
    Join Date
    Oct 2008
    Posts
    1,309

  27. #27
    www.hahntronix.com
    Reputation: mhahn@hvc.rr.com's Avatar
    Join Date
    Mar 2007
    Posts
    252
    Quote Originally Posted by MtbMacgyver
    Later Magicshine 900's switched from the 8 pin SOIC mosfet to a smaller SOT23-3 mosfet as well. It doesn't have as low of a RDSon as the SOIC part and it does run hotter and makes the driver less efficient.

    I'm amazed they won't fix the under spec'ed resistors. It's not like this is really subjective. Any first year EE major could figure this out and the datasheet for the resistor gives very clear guidelines for what is in and out of spec. They aren't even close. What's crazy is the pattern on the PCB board is clearly setup for larger resistors and we're talking about parts that cost less than a penny in quantity. The one good thing about the low drive current on high is the resistors won't be as far out of spec. But they sure are giving up a lot of brightness out of the LEDs.

    Of course they haven't seen a lot of warranty issues with a 90 day warranty and this is the kind of problem that could become a serious issue when the lights are a year to two old. They may get away with them running over spec for a while, but why take a chance for something that is so cheap and easy to correct.
    Maybe we've been looking at their design standards all wrong. Maybe they know exactly what they are doing!

    Make a light cheap enuff that it wipes out your competition. Design it so that it will definitely burn out some critical component AFTER the 90 day warranty period. I mean we know that a well designed LED light, one that doesn't use under spec parts, and regulates the temperature of the LED, should last for what? Tens of thousands of hours. Even somebody who commutes 2 hours a day (unless they work very long hours) will only use the light for half of the year. Let's be pessimistic and say 9 months out of the year. That's at most 3000 hours a year. And those are folks who ride a lot. Most folks who buy these are going to ride what 2 hours a few nights a week for less than half the year. Maybe 1000 hours a year.

    You won't get a lot of repeat sales from a customer if they only buy a new light every 10 years. But if you design your light to fail after say 1000 hours, even the hard core commuters won't fry the light till well after their warranty runs out, and most will make it a whole year since they are unlikely to really use a light for 9 months out of the year.

    And the lack of thermal regulation, just decreases the lifetime of the LED. But odds are the 0805 resistors or the SOT23-3 mosfet will go zorch long before that. And using a 0.1 and a 0.5 ohm resistor even has a twisted logic to it. The 0.1 will burn out first, and will dim the output, so the user will still have some light, and probably survive long enuff to buy another Magicshine. I mean if you buy a light and it lasts a couple of years before it dies, or suddenly goes real dim, and the light costs 1/4 the price of the next cheapest competitor, you'll probably just buy another cheap light.

    If I had a decent power supply in my lab that let you set the current output, I'd be tempted to try over-currenting a bunch of different 0805 0.1 ohm resistors with say around 2 amps. Run the setup on a timer for say an hour, then give the resistors 1/2 an hour to cool off and repeat. Anybody want to set that experiment up? I'll bet if you started tomorrow, you'll find the median date that your resistors burn out is around May 24th.

    Wow, if that's what they are doing it is a truly devilish design

    Nah! That couldn't be it. They're probably just trying to save a measly 15 cents on every light they sell by using cheaper crappy parts.

    Mark
    Nimium est melior!

  28. #28
    mtbr member
    Reputation: oldassracer's Avatar
    Join Date
    Mar 2009
    Posts
    254

    Please upgrade the resistors...

    Quote Originally Posted by MtbMacgyver
    Later Magicshine 900's switched from the 8 pin SOIC mosfet to a smaller SOT23-3 mosfet as well. It doesn't have as low of a RDSon as the SOIC part and it does run hotter and makes the driver less efficient.

    I'm amazed they won't fix the under spec'ed resistors. It's not like this is really subjective. Any first year EE major could figure this out and the datasheet for the resistor gives very clear guidelines for what is in and out of spec. They aren't even close. What's crazy is the pattern on the PCB board is clearly setup for larger resistors and we're talking about parts that cost less than a penny in quantity. The one good thing about the low drive current on high is the resistors won't be as far out of spec. But they sure are giving up a lot of brightness out of the LEDs.

    Of course they haven't seen a lot of warranty issues with a 90 day warranty and this is the kind of problem that could become a serious issue when the lights are a year to two old. They may get away with them running over spec for a while, but why take a chance for something that is so cheap and easy to correct.
    Geoman are you listening?

    I won't buy any Magicshine lights unless this is fixed and I suspect there are others who are not buying either...One might as well buy a light that costs two or even three times as much and lasts more than 2 years than one magicshine which will probably burn out in a year or so.

    Of course, as was pointed out, within 90 days it is unlikely for the magicshine factory to hear of any failures.

    Pity- I was about to buy two of these 1400 lights...and suggest them to my riding club for a group buy of 10+ lights.
    It is possible for rice to absorb other odors in storage. Or could be the quality of water in it was prepared. Mahatma Rice

  29. #29
    mtbr member
    Reputation:
    Join Date
    Oct 2008
    Posts
    1,309
    Measured this board at 1.14 amp output.


  30. #30
    www.hahntronix.com
    Reputation: mhahn@hvc.rr.com's Avatar
    Join Date
    Mar 2007
    Posts
    252
    *** removed because I was confused about what the picture meant ***
    Last edited by mhahn@hvc.rr.com; 03-21-2010 at 03:53 PM.
    Nimium est melior!

  31. #31
    mtbr member
    Reputation: tamen00's Avatar
    Join Date
    Mar 2004
    Posts
    1,025
    That is a the DX3256 driver I think... the ones that are like 4 for 6 bucks. I have made a lot of lights out of these drivers, however only about 1 in 4 of them actually work for any length of time.

  32. #32
    mtbr member
    Reputation:
    Join Date
    Oct 2008
    Posts
    1,309
    Sorry, was interested in component ratings honestly. No EE here, but there is spec and then there is real world application.
    This board isn't out of the MS1400. It is used in many many lights here for years now. Never heard a report of component failure with its use.
    Tamen is correct though that many show up DOA or shortly after being fired up.
    I would agree that none of these boards are anywhere near the quality of anything George over at taskled makes.
    Point is that the components in the MS1400 and MS900 will do the needed job even if they are out of spec. Only proof of this is the real world use of similar spec'd drivers in DIY lights for years now.
    I would recommend this light to anyone interested in getting into biking at night, anyone wanting a second light, or anyone wanting to upgrade from an HID or Halogen.
    And no matter how cheap or expensive your lights are, always have a second for backup.

    BTW. Its great getting input from the EE types here who know these driver boards and components. Over driving cheap components cuts efficiency and life expectancy. So does over volting halogen bulbs.
    Every system will have its weak point whether its fragile HID bulbs, Hid ballasts, battery fade due to age, charge cycles, cold temperatures. Not to mention catastrophic failure due to crashes.
    But for our application here, with the understanding that one should always have a backup, these lights seem to hold up well.

  33. #33
    mtbr member
    Reputation:
    Join Date
    Jan 2007
    Posts
    752
    Quote Originally Posted by odtexas
    Measured this board at 1.14 amp output.

    I'm not following your point. Just because a board is inexpensive doesn't mean it's using improperly spec'ed components that will be more likely to fail. I've never looked at the driver you measured at 1.14 amps, so I don't know if it's a good or bad design. But you can't assume it's a bad design just because it's cheap and some have failed. Most early failures are caused by assembly problems and not by component selection.

    Yes, people over-drive halogen bulbs to get more light. But there is a purpose for doing that, it actually improves the light. And that's not actually quite the same thing because bulb makers do a lot of testing at elevated voltages and publish enough data so you can predict the bulb lifespan at elevated voltage.

    There is no purpose for driving a resistor over spec and the resistor maker do not publish data that'll let you predict the lifespan when it's driven over spec. So the only benefit is a very tiny cost savings. Just as an experiment, I searched digikey for the cheapest resistors both in the size they are using and a size that would be in spec.

    805 1/4 watt resistor - 3 cents at qty 5000
    http://search.digikey.com/scripts/Dk...R10EZHJLR10-ND

    1206 1/2 watt resistor - 8 cents at qty 5000
    http://search.digikey.com/scripts/Dk...CSR1/20.1FR-ND

    So, we're talking 5 cents difference. And either resistor will go on the PCB without any modification. So, I just don't get the reason for using the under spec resistors.

    Overall I'm not a Magicshine hater and I've recommended them to lots of friends that want to night ride without spending a lot of money. But I do warn them about the problem areas and will fix the driver for them if they burn out. But, why not make a simple change like this to avoid the possibility of a problem. And I have fixed 2 lights already where the resistors burned out. Both of those cases were because the folks let them get really hot by running them without airflow, but others could start failing with age. Only time will tell.
    Last edited by MtbMacgyver; 03-21-2010 at 03:05 PM.

  34. #34
    mtbr member
    Reputation:
    Join Date
    Oct 2008
    Posts
    1,309
    I can't remember any of the EE guys here majorly hate on any product. Critique and dissect, yes. But that is what we do here.

    Point is we have people now thinking that there are major issues with a light due to a misunderstanding of currently used components spec. vs real world application.

    The MS lights are what they are. Suggesting better components will aid MS if they have a reason to redesign. Application seems to suggest that redesign isn't necessary. If they were having significant failures and lost demand the changes would then probably be made.
    Good news is they know which components to look at and what to replace them with.
    Good work...........

  35. #35
    www.hahntronix.com
    Reputation: mhahn@hvc.rr.com's Avatar
    Join Date
    Mar 2007
    Posts
    252
    Your point about spec vs. real world has some validity. Chip makers typically recommend a range of voltages for a part, and have an absolute, never go above this voltage spec as well.

    You can typically get away with violating the recommended range, for a while. But if you go over the max rating, well who knows.

    A company I used to work for had an owner who was pretty tight-fisted, and thought he was an engineer (no degree that I knew of). He made me use a 5 cell NiCad battery in a $2000 datalogger without adding a $0.75 voltage regulator. The processor in the datalogger would get 6 volts, most of the time (top of recommended range), except when the batteries were fresh off the charger, then it might get 7 volts or more. I kept telling him he was nuts (used politer word than nuts, he signed my paycheck), we needed a regulator. He just replied that all the units we had tested in the lab seemed to work just fine, shut up about the voltage regulator and start shipping product. He was the boss, we started shipping.

    About 6 months later, we realized that we were getting a 20% customer service return rate on our datalogger product. Looking into the issue, I found the batteries we had used for testing had come out of our stockroom and had been sitting on a shelf for over a year before we took them to the engineering lab. Since they had sat for a while they had a higher internal resistance than batteries fresh from the factory. Fully charged they put out a bit over 7 volts, but dropped to a bit over 6.5 volts when the datalogger was turned on. This was outside the recommended range but under the 7 volt maximum voltage for the processor. When we started shipping dataloggers with factory fresh batteries the processor saw about 7.1 volts for just a few seconds after it turned on. This caused a 20% failure rate. The units didn't all die at once, it sometimes took several months before the came back for servicing. I think it depended on how often the user charged the battery pack.

    Best part of the story is I figured all this out, found a voltage regulator that worked, added it to the battery packs. Our return rate dropped to less than 1% in a couple of months, and most of those were customer abuse of the product. At my next performance review I got dinged for not implementing the voltage regulator from the start, like my boss claimed he had wanted me to.

    My lesson from all this: don't run a part outside the suggested range & don't work for assh*les. Life too short.

    I don't hate the magicshine. It just reminds me of products my old boss used to make. Though the price is better

    Mark
    Nimium est melior!

  36. #36
    mtbr member
    Reputation:
    Join Date
    Jan 2007
    Posts
    752
    Ironically, if the current readings you posted earlier are correct. They are low enough on high with all the LEDs running that the resistors will actually be just barely in spec. Problem solved at the expense of lumens....

    But, the current readings you posted don't really add up to the runtimes others have posted so I think there is a good chance they are incorrect and too low. Reade has posted that he consistently gets runtimes of 160 minutes. That would imply a load of about 12.25 watts on the battery at full power. The current reading you measured only add up to a little over 9 watts. So that could be true if the driver efficiency was really low in the 75% range. But the efficiency of the driver on MS 900 was in the 90s. So that seems unlikely.

    Which means the light may actually be a good bit brighter than the approximately 800 lumens that would results from the current values you posted. I suspect your current reading may be incorrect because some of the LEDs may be in parallel. If that's the case, then inserting a meter into the LED wiring will actually change the current flow.

    The other reason folks may not see as many resistor failure on the MS 900 as could be the case is because a lot of folks run the light on medium. On medium the resistors will be in spec and won't run really hot like they do when the light is on high.

    People that do run the light on high may also be having problems and not even be aware. When resistors run that hot their resistance value can start to degrade. That'll result in the light getting dimmer. So folks that have a MS with some hours on it may be running dimmer and not realize that's the case. And maybe they don't even care as long is it still gets the job done.....

  37. #37
    mtbr member
    Reputation:
    Join Date
    Oct 2008
    Posts
    1,309
    Don't forget that the battery cells are the cheapest Chinese made ones. CPF had a whole section on variabilty of cell voltage and runtimes back in the day ( a year ago)

  38. #38
    Rolling
    Reputation: lidarman's Avatar
    Join Date
    Jan 2004
    Posts
    11,110
    Quote Originally Posted by mhahn@hvc.rr.com
    Maybe we've been looking at their design standards all wrong. Maybe they know exactly what they are doing!

    Make a light cheap enuff that it wipes out your competition. Design it so that it will definitely burn out some critical component AFTER the 90 day warranty period. I mean we know that a well designed LED light, one that doesn't use under spec parts, and regulates the temperature of the LED, should last for what? Tens of thousands of hours. Even somebody who commutes 2 hours a day (unless they work very long hours) will only use the light for half of the year. Let's be pessimistic and say 9 months out of the year. That's at most 3000 hours a year. And those are folks who ride a lot. Most folks who buy these are going to ride what 2 hours a few nights a week for less than half the year. Maybe 1000 hours a year.

    You won't get a lot of repeat sales from a customer if they only buy a new light every 10 years. But if you design your light to fail after say 1000 hours, even the hard core commuters won't fry the light till well after their warranty runs out, and most will make it a whole year since they are unlikely to really use a light for 9 months out of the year.

    And the lack of thermal regulation, just decreases the lifetime of the LED. But odds are the 0805 resistors or the SOT23-3 mosfet will go zorch long before that. And using a 0.1 and a 0.5 ohm resistor even has a twisted logic to it. The 0.1 will burn out first, and will dim the output, so the user will still have some light, and probably survive long enuff to buy another Magicshine. I mean if you buy a light and it lasts a couple of years before it dies, or suddenly goes real dim, and the light costs 1/4 the price of the next cheapest competitor, you'll probably just buy another cheap light.

    If I had a decent power supply in my lab that let you set the current output, I'd be tempted to try over-currenting a bunch of different 0805 0.1 ohm resistors with say around 2 amps. Run the setup on a timer for say an hour, then give the resistors 1/2 an hour to cool off and repeat. Anybody want to set that experiment up? I'll bet if you started tomorrow, you'll find the median date that your resistors burn out is around May 24th.

    Wow, if that's what they are doing it is a truly devilish design

    Nah! That couldn't be it. They're probably just trying to save a measly 15 cents on every light they sell by using cheaper crappy parts.

    Mark
    I love conspiracy theories.

  39. #39
    www.hahntronix.com
    Reputation: mhahn@hvc.rr.com's Avatar
    Join Date
    Mar 2007
    Posts
    252
    Quote Originally Posted by MtbMacgyver
    Ironically, if the current readings you posted earlier are correct. They are low enough on high with all the LEDs running that the resistors will actually be just barely in spec. Problem solved at the expense of lumens....
    Current thru resistors is 0.4 amps for the 0.5 ohm resistor, and 2 amps for the 0.1 ohm resistor. That gives a total of 2.4 amps running thru the LED. That value comes right out of the SM5241 datasheet: ILED = 0.2/ RSENSE (one of the few parts of it that was in english).

    That is over spec for the 0.1 ohm resistor when run on high by a factor of 3.2, assuming a 1/8 watt 0805 package. There are 1/4 and 1/3 watt 0805 resistors available but they cost more than using a 1206 package resistor on the board (which there is ample room for), so my guess is they use a cheaper 1/8 watt part. It's only a matter of time, but the resistor will burn out. People have seen this part fail.

    I doubt the efficiency gets to 90%. At 2.4 amps the LED is consuming 8.4 watts (on average, according to SSC-P7 data sheet). 2.4 * 0.2 volts = 0.48 watts consumed by RSENSE. If the freewheel diode drops 0.4 volts (depends on how good diode is, but they show a tendency to use cheap parts), then they lose another 0.4 * 2.4 * 0.5 = 0.48 watts (0.5 is for 50% duty cycle of current thru freewheel). So they have to be wasting at least 0.96 watts. Best they could do is 89.7% efficiency and they must be losing something in the inductor and mosfet.

    As to lumens for the P7, the best they could do with a C bin would be 730, and 823 if they used a D bin (but they are probably shipping C or B binned parts). They'll lose a bit of light because the reflector and glass aren't 100% transmissive.

    I'd be surprised if the whole unit, with all LEDs on full, puts out anything near 1100 lumens. Which is still a decent amount of light.

    Mark
    Nimium est melior!

  40. #40
    mtbr member
    Reputation:
    Join Date
    Jan 2007
    Posts
    752
    Quote Originally Posted by mhahn@hvc.rr.com
    Current thru resistors is 0.4 amps for the 0.5 ohm resistor, and 2 amps for the 0.1 ohm resistor. That gives a total of 2.4 amps running thru the LED. That value comes right out of the SM5241 datasheet: ILED = 0.2/ RSENSE (one of the few parts of it that was in english).

    That is over spec for the 0.1 ohm resistor when run on high by a factor of 3.2, assuming a 1/8 watt 0805 package. There are 1/4 and 1/3 watt 0805 resistors available but they cost more than using a 1206 package resistor on the board (which there is ample room for), so my guess is they use a cheaper 1/8 watt part. It's only a matter of time, but the resistor will burn out. People have seen this part fail.

    I doubt the efficiency gets to 90%. At 2.4 amps the LED is consuming 8.4 watts (on average, according to SSC-P7 data sheet). 2.4 * 0.2 volts = 0.48 watts consumed by RSENSE. If the freewheel diode drops 0.4 volts (depends on how good diode is, but they show a tendency to use cheap parts), then they lose another 0.4 * 2.4 * 0.5 = 0.48 watts (0.5 is for 50% duty cycle of current thru freewheel). So they have to be wasting at least 0.96 watts. Best they could do is 89.7% efficiency and they must be losing something in the inductor and mosfet.

    As to lumens for the P7, the best they could do with a C bin would be 730, and 823 if they used a D bin (but they are probably shipping C or B binned parts). They'll lose a bit of light because the reflector and glass aren't 100% transmissive.

    I'd be surprised if the whole unit, with all LEDs on full, puts out anything near 1100 lumens. Which is still a decent amount of light.

    Mark
    Your calculations match my calculations for the MS 900 running on high.

    I was referring to odtexas's post that the current through the P7 was 1.71 amps when everything was on high and also making the assumption that the XPE's are driven from an alternate driver. I was also giving them the benefit of the doubt that they did use a 1/4 watt 805 resistor. Some of those assumptions may not be true, but you could make a case that this light is potentially in spec if the LED drive currents are really so low.

    If the average current through the P7 is really only 1.71 amps on high, then 1.41 amps will go through the .1 ohm resistor. That means it would dissipate 199mw. That would be within the spec for a 1/4 watt (250mw) resistor.

    I have my doubts if the 1.71 current through the P7 is correct. Those currents don't match up with the runtimes that have been reported. I suspect the current readings may be wrong either due to inserting a meter in the circuit or because there is PWM even when this light is on high. I'll know more when I get my hands on one myself.

    If the P7 drive current is really 1.71 amps on the highest setting, then they're only getting about 480lm out of the main light on that setting.

  41. #41
    Magicshine Light Dealer
    Reputation: The Singletrack Store's Avatar
    Join Date
    Oct 2009
    Posts
    171
    I agree, I've had great fun using this light. Thanks for all the work and time you fellas put into this thread. Cheers Reade

    Quote Originally Posted by mojojojoaf
    Spill and penetration look great on that last photo- not bad at all.
    www.thesingletrackstore.com
    Magicshine 900 & 1400 Lumen Cycling Light Dealer in Arizona

  42. #42
    mtbr member
    Reputation:
    Join Date
    Aug 2009
    Posts
    11
    Can you upload a better picture of charger? Does it work in Europe (220V)? Does it use this cord link?

  43. #43
    www.hahntronix.com
    Reputation: mhahn@hvc.rr.com's Avatar
    Join Date
    Mar 2007
    Posts
    252
    I would suspect from the picture that it just has the plug molded into the case, so it's probably set up for 'merikan plugs, not those funny round things you Brits use

    But odtexas, could you upload a picture of the plug side of the charger, and use your macro lens, pretty please. I'm curious what regulatory markings it has. I've seen bogus ones on Chinese chargers before.

    If you really feel brave, please open the charger and take a couple of pics of the innards.

    My MagicShine 900 used a very bogus charger design. It had a fuse, but I'd hardly call it safe.

    Mark
    Nimium est melior!

  44. #44
    mtbr member
    Reputation:
    Join Date
    Aug 2009
    Posts
    11
    I am not from Britain, I am from Estonia(North-East Europe). We have two round pins, which are a bit wider apart than american flat pins. I am more worried about the voltages than the plug actually

  45. #45
    mtbr member
    Reputation:
    Join Date
    Jan 2007
    Posts
    752
    I got my hands on a MS1400 and am in the process of analyzing how the electronics work. I'll do a more complete post shortly with pictures and details, but here's a sneak peak at some of the more interesting details.

    The circuit board in the center section of the light housing contains two complete switching buck current regulators. One for the main light and one for the side lights. The driver for the side lights is set for a base drive current of 0.6 amps and the one for the main light is 2.4 amps. The board in the main housing is just the base regulators; it doesn't provide any of the logic for user control of the light.

    The command control button provides all the user control for the light. It's connected to the driver pcb via a four wire cable. Two of the wires are power and ground and the other two carry pulse width modulated (PWM) control signal for the main and side lights. This pcb basically has a microcontroller, a button, some LEDs for the battery status, and a voltage regulator.

    As I suspected, the PWM control signals to the light are never driven at 100% (constant on) even when the light is on its highest setting. The duty cycle is in the 95% range when the lights are on high. It is interesting that the high setting for main and side lights is the same 95% regardless of whether the two sets of LEDs are on individually or together. That means in theory that the drive current to the main LED should be the same whether it's on by itself on combined with the side lights. This is different from what was implied by the current measurements odtexas posed earlier. I think the electrical noise from having both drivers running may have caused low readings when both sets of LEDs were on.

    So is it bad that the lights are never on 100% of the time... Not necessarily. In theory that means the max average current for the main P7 LED should be 95% of 2.4 amps and 95% of the 0.6 amps for the side LEDs. In practice, that may not be true because it depends on how long it takes the driver to ramp back up to full power after being off for an instant.

    It does mean that you won't get accurate current measurements for the LEDs with a normal multimeter. The current flow in this case isn't technically DC due to the PWM and most multmeters won't give an accurate measure of the true average current for a switched waveform. It's actually quite difficult to get good numbers for a circuit like this. Switching regulators generate a lot of electrical noise, which makes it hard to get accurate reading even using an oscilloscope. And these drivers are definitely on the noisy side. It's good they are inside a metal housing because they would certainly cause problem for other wireless accessories if they were in a plastic case. That may still be an issue, but it's not something I've tried.

    Here is what I measure for power consumed from the battery at different light levels.

    Side lights (high) - 4 watts
    Main Light (high) - 8.3 watts
    Both (high) - 12.3 watts

    I'm still trying to get precise measurements for the LED currents. It looks to be less than 95% of the base current settings for the drivers. From preliminary data, it looks like the true average current is about 90% of 2.4 and 0.6 amps for the main and side LEDs. If thatís correct, that puts the average LED currents at

    P7 LED current on high - 2.2 amps
    XPE LED current on high - 0.55 amps

    Based on the datasheets for the LEDs and 90% efficiency of optics, I would estimate the light output as:

    P7 Main Light - 550 lumens
    XPE Side Lights - 300 lumens
    Total - 850 lumens

    All these numbers are preliminary and I will update them as I get more detailed measurements. One thing I'm still investigating is whether the electrical noise when both drivers are running is actually messing up the efficiency of the drivers and hence the output or if it's mostly just causing causing problems with getting accurate measurements.

  46. #46
    mtbr member
    Reputation:
    Join Date
    Aug 2009
    Posts
    11
    Do you have any info about the charger(input voltage, charging times) or you only have lighthead?

  47. #47
    mtbr member
    Reputation:
    Join Date
    Oct 2008
    Posts
    1,309
    Will take, and post, pictures of the charger when I get home if no one beats me to it.

    MtbMacgyver -

    Thanks for doing a proper EE assessment of the drive circuits. Stuff like that is still beyond me. Since you do have one now did you notice the high pitch whine from the boards. It isn't that bad, but I can notice it when its on in a quiet room.
    Probably not a chance of hearing it on a bike.
    Looking forward to seeing where the more accurate drive information puts this light regarding lumens out the front.


    ODT

  48. #48
    mtbr member
    Reputation:
    Join Date
    Jan 2007
    Posts
    752
    Quote Originally Posted by odtexas
    Will take, and post, pictures of the charger when I get home if no one beats me to it.

    MtbMacgyver -

    Thanks for doing a proper EE assessment of the drive circuits. Stuff like that is still beyond me. Since you do have one now did you notice the high pitch whine from the boards. It isn't that bad, but I can notice it when its on in a quiet room.
    Probably not a chance of hearing it on a bike.
    Looking forward to seeing where the more accurate drive information puts this light regarding lumens out the front.


    ODT
    Yeah, I did notice that and wondered if it was just the one I have. It's actually pretty noticeable in a quiet room. I'm pretty sure it's from the pulse width modulation turning the driver on and off rapidly. It's most noticeable on my light when both sets of LEDs are on high.

  49. #49
    Magicshine Light Dealer
    Reputation: The Singletrack Store's Avatar
    Join Date
    Oct 2009
    Posts
    171
    Magicshine 1400 vendors will have access to European and Australian chargers very soon. Cheers Reade

    Quote Originally Posted by GeniusGun
    Can you upload a better picture of charger? Does it work in Europe (220V)? Does it use this cord link?
    www.thesingletrackstore.com
    Magicshine 900 & 1400 Lumen Cycling Light Dealer in Arizona

  50. #50
    mtbr member
    Reputation:
    Join Date
    Jan 2010
    Posts
    291
    I'll put in my 2c worth regarding the drivers etc.

    These lights are very likely made to a price point. Within that price point there is a certain $ amount that can be assigned to the design/assembly of the electronics.

    From just looking at the boards, they have taken some care in the design/layout, but there's lots of refinements that obviously have been ignored or are not within the knowledge base of the designers.

    Just looking at via diameters (rather large) shows a lack of knowing the tradeoffs for carrying current versus more smaller vias. Also the thermal properties of large diameter vias versus small.

    The fact that the inductor is a toroid, unshielded and not even fixed down to the PCB - that's likely the cheapest inductor they could find, definitely not a quality unit.

    The board has a thermal path that relies on the outside rim - very poor, but easy to assemble and they can put components on both sides.

    One fact that apparently has been missed in the above discussions of resistor ratings etc, is that all the components need to be de-rated for the ambient temperature of the board electronics. I'm betting that the driver board is stinking hot when bottled up inside the case - especially with the minimal thermal path to the body of the light. I'd hate to imagine the junction temperatures of the power semiconductors.

    Anyhow, it's really quite simple. The light is inexpensive, lots of folk won't spend much more than $100 (or even $50) for a bike light including battery & charger. Using that metric, the light will sell in large volumes and that will allow them to drive the prices down by being able to work with lower margins.

    If folk want a light that is built to the best possible standards (mechanically and electronically) it will cost more, volumes will be less so margins need to be higher.

    If you buy a big box/department store bike (<$100) then one of these Chinese lights will be the ticket and even then expensive compare to the bike. If you're the person that will spend $150+ on some good pedals and shoes or $200 for a hand made saddle, then you likely aren't looking to save $100+ on a bike light and instead are looking at getting a quality light made by folk that ride bikes and won't be happy selling a product that they themselves wouldn't use.

    So, just like there's a market for <$100 bikes and $500 - $5000 bikes, there's room for made in China lights and the high end lights where a lot of the profits go into R&D and support and hiring local talent.

    One last point (since I'm already partially off topic here)....

    Other than the distributors/reps have we ever seen the Magicshine (etc) folk on the forums - versus the folk that design/fabricate the electronics/lights of the $expensive lights? That would say a bit about the dedication to listening to the customer and taking direct input to improve the next generation product.

    cheers,
    george.

Page 1 of 3 123 LastLast

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •