led floodlight teardown
Sometimes this can be unfortunate, I\'m just another MTBF statistic outside of the histogram, and for those who understand these statements, you know where I am, anyway, the LED spotlight shown above is an example.
Originally I bought 3 of them and, to be fair, they have been going on for over a year and the remaining two are still strong.
Unfortunately, one of them gave up the ghost and faced getting another.
From the price point of view, their price in the UK is about 10 per car, and you can\'t complain about what you get.
They give out soft 10w light, ideal for porchlight, or just as I spread them out in the garden as a backlight.
They are subtle and pleasant.
Most people will accept this and move on. .
I mean, why would it fail? . . do we care. . . . . ?
For those who want to chase, jump to number 10 in this series.
If anyone else wants to read the ins and outs of a. . then read on. . . .
It is clear that the LED is not replaceable. .
It may be said that there are no parts that users can repair. . .
Well, it\'s a bad thing for me. . .
Anyway, we can have a look even if they are not. . .
Is it well designed?
Use a screwdriver right away. . .
It\'s time to look deeper. . . …. . .
Now, at this time, it\'s time to start my preaching, just like the other items I comment on that involve more than 60v power supply and voltage, be extra careful.
I\'m not responsible for anyone being hurt because of these things, don\'t unless you know what you\'re doing.
If you have to and you are curious enough, then always disconnect the item from the power supply and be wary of any undischarged voltage that may persist and they will still be hurt.
If you want to measure the absolute voltage, you must connect any meter, etc. , then turn off the power, connect the test meter and turn it back on.
Always working with one hand and working better with the right rccd.
Grounding is ground on the housing, but the ground on the pcb diagram is not isolated by a diode.
If the range is used, the range is no longer used by isolating the transformer floating test piece.
Don\'t float the range, and don\'t want to lift its nails, which is a bad practice for it, and it may be forgotten if it leaves.
You were warned. . . .
Death is fatal!
This is the performance.
Now, there is a screw type cable cover on the back of the accessory, and according to the photo, it shows 3 terminals below.
The corresponding landmark is marked as lne I removed the cable and connected to another main cable that I know must be charged. Connected up. . . . not a sausage. . . typical.
Turning the accessory over, I noticed that it was fixed with a safety screw that looked unsafe with 4x3 holes Phillips.
I think you can buy a tool for this, but needless to say I don\'t.
This may be part of the CE specification.
To get you to speak out loud, they were buried in the back radio fence!
Half an hour later, with some selection words, the front glass was removed, showing from the appearance that some LEDs had a reflector with a center diffuser at the top. . . funny that.
The reflector is held by several Phillips and the pcb is placed in an insulated cover.
Connect to the pcb by splicing and clamping 240V junction.
You can see this on the right side of the picture.
Also pay attention to the Earth that enters and is bolted to the die-casting shell.
I have highlighted the two areas on the diecast shell that shows good things.
The area of the blue border holds the led fixed with 2 screws and a layer of radiator, which I will describe more fully later.
The red area is the area where the pcb is located.
So why is this not working!
Therefore, the trunk line enters in the form of line and neutral [Red/Blue]
One end is connected to the pcb.
There are also two wires that hold the PCB to red and blue and connect to the led block.
Let\'s do some quick checks to see if this is a quick solution such as a blown fuse etc.
Meter check fuse on Ohm [red block]
Feed Red directly on the track after live broadcast.
The good news of the short flying.
Since the fuse is not blown, this means that there is no short circuit for any required power, or that the device that is switching power has no short circuit, or that any other device on the high voltage track may be fine, however, this may not be good if there is any semi-participation. What\'s next . . .
Enter the bridge rectifier of the magnetic filter block consisting of two caps and two magnetic coupling coils.
This is a two-coil generator in a common ferrite housing.
This works in two ways, isolating public noise from the outside and will keep any switch noise inside, which may also be required by UL or CE.
The lid sits on the power supply, so the rated voltage is 400 V, which you can see in the photo above.
That doesn\'t stop it from working anyway, so what happens next.
It looks like it hit some resistors and some sort of chip. . .
There is no doubt that since the direct power of the power supply to the led is quite high, the buck regulator.
Let\'s check the fat main power cap sitting in all these places.
Its rated voltage is 400V @ 105 °c, so the current track will be at 220 v ac rms or 220x1. 414[root2]=310V DC ISH.
Here is a short digress.
It\'s bad enough when AC power rolls at 50Hz, but at least it\'s elegant enough to pass through a zero point every 10 ms, 300 v dc is more than just very hard. . how do I know? . . . don\'t ask. .
So unless your heart stops beating at the moment of contact, this is very unpopular.
Think about it, if you do, I expect you might have one. . . .
Be safe at all times.
Now, we can install a light bulb on the lid at this point to see if it lights up, but it will run safely and check the diodes in the 1 m bridge rec and pass through the filter.
Everything is fine now. . . …. .
Now, in the past, I know that these main DC electrolysis will die, either because the heat is too large, resulting in a higher resistance, and then more heat.
The giveaway is an expanded jar, but it looks OK. . . see photo.
Yes, I may have overlooked the obvious here. . .
What if the led is broken? . .
Of course, they are incredibly bright and generate a lot of heat. . .
Let\'s take a look at this block, so here\'s the LED block.
It consists of 9 LEDs and I think the rated power per LED is 1 w, although I am not sure.
This is weird because the back panel shows 10 w, but I think they are referring to a small marketing loophole that refers to the power consumed rather than the equivalent LED power.
Anyway, it becomes a photon with a percentage. . . .
More efficient than incandescent lamps, so let\'s move on.
Please note the positive and negative connection to the board.
Now I know that the forward voltage of the high power led is very different from the bog standard led variety, look at the specifications of the Google 1w device and it looks like it will take at least 4V to get them to glow.
So I have a string of LEDs. 9]
In the series where I need to use the power test. 9x4 =36v. . .
My power supply is only 30v and there is wind in the back so need to test them separately.
Look at the structure of the top pcb stick to the board above.
I included some pictures of the sides.
It is aluminum and the pcb is glued directly to it to eliminate heat, but if it is not directly attached to the base plate of the led, I suspect it works too well.
We can put a hot gun on it later to see how hot it is.
Start with the first 5 lines at the bottom of the photo.
The current limits the power supply to 100 mA and they go off when we reach 16/20 vtoprow 4 No go. . . ah hah. . .
There is a damaged led.
By the way, when checking these diodes, I made a quick model of the 9v PP3 battery and put it on the diode for separate inspection.
Make sure the polarity is correct, of course.
Just thought of another project. . .
The led inspector can be switched. . . . . . stop it. . .
Let\'s Do led one by one before we find the culprit. . . .
Well, if we look at the picture, it seems to be familiar with it.
Now it happens that I have a spare led rated 0.
5 w, not sure if this works in the current settings.
I won\'t do anything fancy when removing the old gun with a hot gun, so just remove the old gun and glue it in the new one.
The New rated power is 100 mA and the maximum power is 150 mA, so we may have a chance if the other rated power is 150 mA.
What have we lost. . .
Other than another LED.
Hold on, though what set the parameters of the current in the led, and how we got about 45 v dc from 310 v dc to the led string. . . .
It must be a step-down regulator for a switching mode power supply, but also because it is cheap and cheerful, completely unisolated. . . . . be afraid. . . be very afraid! . Let\'s dig deeper.
From the early findings, we encountered some resistors after correction that seemed to be related to some kind of drive chip.
At this juncture, I usually try to find out the name of the chip, or in some cases, the components used around the chip are obvious.
It\'s easy for this one because its obvious mark is MT7812 sold by Maxictech or until it\'s replaced by other companies.
Interestingly, the mark for this board is version 2. 3 from 2015.
The data sheet is very comprehensive and provides you with some application information.
Let\'s see if I can relate this to what we have here.
From the data sheet, we have a full-wave rectifier directly connected to the storage capacitor c1.
In our example, before we reach C1, we first filter the network input.
The inductor filter leg I measured is about. 1.
82 mh of each leg is parallel to the two 220n and 15 N respectively.
The chip caps with resistance RST1 and RST2 each are 200 K and C2 are about 1. 5u.
On this board, the center junction of RST2 and C2 has a Zener ground with a rated voltage of 14 v.
RST1 and 2 set the Zener current for this and will be selected to maintain the 14v to pin 3 of the integrated circuit, even when VMIN is estimated to be 290 V
R1 and 2 are over-voltage protection resistors for 330K and 12 K.
Well, it seems good to check all of this and can be checked for zina by feeding with low pressure, it can be said, though considering death, I doubt it\'s a problem.
I may check later.
So what about the input o/ps and feedback?
Let\'s go back to our motherboard and see if there\'s something different from the schematic.
First of all, what\'s really interesting is the pin 8 resistor Rcs.
Reading the notes and looking at the internal structure, the resistance here seems to set the current in the LED path independently of any voltage consideration, and when looking at our board, it seems that there are two resistors sitting on pin 8 above.
One is 20 and the other is 1. 3 ohm.
This looks like 1.
22 ohms is 1.
3 Ohm resistance dominates.
The equation that inserts this into the peak inductor current can get 327 mA.
This will give the led current of 163 mA, slightly higher than the rated current of 0.
5w led so we can increase the resistance to reduce the current.
Perhaps the 120 mA target is safe.
If I find a 1 watt led on Ebay, maybe this is a better option?
It\'s 10p anyway, so let\'s flash it out.
This is my new board with led in it.
Not very elegant, but what do you expect :-)
Please note that it is much larger, which may help to eliminate some heat, but wonder if my contact with the lower side is particularly good.
Let\'s start it up with DC power to see if it will die.
So I connected 5 in the circuit and under 34 v I managed to get through 118 mA.
View the specifications of the 1w led. . .
This is not a COB device, but I highlighted the significant bits on VF and current in yellow.
VF is like all LEDs, wandering around depending on when in the day. . .
Not really. . .
More like how hot it is, how much current you are trying to transfer through it.
The 1w version likes 140 mA and will accept a peak of 260. . .
Wow, close to a 100% increase. . .
I don\'t think I will try my luck on this, as MTBF may plummet.
On the other hand, my poor little substitute ran 100 mA M, 150 mA M Max and 45lm.
Forward voltage is not displayed.
Other specs lost lm for some reason. I know. . . .
I have a cunning plan for us to measure the original led and try to push 150 mA through it.
I will increase the voltage before we have no brightness change. . . apparent ….
And measure the current.
This is very interesting. . . @6.
6v forward voltage we move 150 mA and are quite bright, you have to push it over 200 mA in order to get the obvious difference and then the damage starts to show.
The naked eye 120 mA does not look more than 150, so it may be better to run at a level of 125 mA.
The resistor is not set like this, so maybe we can modify it a little so that everything can last longer without damaging the level of light too much.
Let\'s turn it into excel and see what brand it is.
We can use the calculation formula in the data table.
Excel seems to think that given the operating current of 123 mA and the peak of 246, we should run at a resistor value of 1. 625 ohms.
What is this made. . .
Because I\'m lazy, I love the online calculator, which gives 30 ohms parallel to 1. 8=1. 7ohm. . that will do.
Give us about 120 of the operating current in the inductor. . .
Are we done? . . . .
We still have some checks to do. . .
We opened a can of bugs! !
Now, in another of my instructions, I may have mentioned that we brought a cocker, which was for the dog to go home.
Now she is as affectionate and tearful as a Nighthawk, but she is also very naughty.
Don\'t be fooled by appearance. . . .
If it was on the floor and lost the fair game, she would eat anything.
Push better through the mailbox.
Now, what does this have to do with fixing this light, and what does repentance have to do with it?
Now, during the testing process, it was found that the original duff led was the culprit, I did some testing and accidentally added the voltage on the other led and blew it away. . .
Yes, the dead died. . . no longer more.
You see, they don\'t like it, the smoke is gone, and it\'s gone.
So I thought I would buy another one from the 50 I bought from ebay some time ago.
As you have guessed, my wife told me that the dog ate the whole rope, well, chewed them up and caused the split.
The wife didn\'t tell me because she thought I wouldn\'t miss them. . . typical. . .
So back to Ebay I will order some 1 W.
That said, it won\'t distract our mission, so when we wait for them to arrive with the new resistor, let\'s take a look at the magnetic blocks that power this string of LEDs.
Keep in mind that with reference to the schematic diagram, the led is fed from the high voltage rail through the inductor to the drain of the fet inside the driver chip.
As an inductor, applying a complete Monti on it will generate a triangular current through an inductor.
Also keep in mind that this is not an isolated circuit, all we have to do is get the complete DC voltage from the high voltage rail.
So the current upgrade to IPk will be close to 250 mA in our revised version, with an average of 125 mA.
It can\'t go beyond that because the chip senses and turns off the fet. . .
So, what is the growth rate determined?
Now, if we have a fast slope, the frequency will increase, and the way to slow down the slope is to add some inductance. . .
Insisting on this must mean that the frequency is inversely proportional to the inductance.
Looking at the equation in the data sheet, the frequency is definitely inversely proportional to the inductance and IPk, but it also locks directly on the voltage on the LED string and input voltage. . . .
So if the input voltage drops, then the frequency drops. . .
Is this important?
The answer is not a lot, but there is a limit on the conduction time of the chip and the surrounding components (such as the recovery diode) and to avoid entering the interrupt mode.
In an ideal world, we want the waveform of a triangle, which is almost continuous, rising and falling on the slope.
Let\'s take a look at some possible numbers.
The frequency that the chip can handle is 30 to 80 Khz, which sets our boundaries.
It also sets the size of our input filter, although the scrolling point should not be affected too much.
Vin Min Can be 10% lower than our 310 V, so let\'s put it at 285 V.
What about our LED string? . .
We have 9 LEDs and I measured 6.
The 6v is going down positively, which, by the way, meets our specification for 1w LEDs at 5. 8 to 7V. . . so lets use 6. 6V.
What about L?
Where do we start? . .
I know let\'s start with the inductance value of 100uH, sweep it forward and see what kind of numbers we get the frequency, after all we have enough consts. . .
In terms of programming
Some people in the UK will remember the track that shows your age. . . .
Damn it, it shows me too. . . moving on.
So this thing will work at 55Khz frequency so is this the best policy?
For those who know it, it\'s equivalent to about 18 microseconds, or the amount of time my bank account is in black every month. . . no I joke . . .
The second of its skin :-)
So what is the constraint?
The Toff Min From the data sheet must be greater than 1.
5usToff Max is not greater than 400usMax on should not be greater than 55us.
Let\'s run the numbers.
It seems that only a few of these parameters are within the limit.
At the end of the day, you need to squeeze the magnetic material into a box, the higher the frequency of the drive, the lower the inductance, and the smaller the coil. . . hooray. . . .
That\'s in henrith too! So what we got. . . 2. 4mH ----->5. 8. . .
What if we give the chip a breather and let it run at 55 Khz? . . that\'s 3. 4mH.
At this point, I will be following the old ferroxcube program to spit out some numbers. . .
Let\'s take a look at the EFD series as they are small and low from 2.
4 mH, this will make the chip run at the top level of about 78 K.
This means that the inductance will be small.
However, if vin rises, the frequency of possible violations of some restrictions will also rise.
So in some numbers like EFD series/inductor value/current and click go!
Boom we have a suggestion that the EFD15 core material 3F3 has 125 laps.
It is also 15mm in diameter, the same as it is currently installed. This is an ETD for clearance and the wire size is 0.
RDC of 224 and 2 ohms. . . .
Now, this is interesting because it means the existing inductor on the [board]
I can\'t measure the inductance of it because that means pry it off the board]
But I can measure it with a resistance of about 5 ohms.
That means it has more turns.
OK, let\'s try the midpoint at 3. 4mH.
Do not increase the core size by one value.
Okay, let\'s try it at 2. 9mH. . .
Bingo EFD15 153 turns about 3 ohms 0. 2mm wire size.
So now let the bet, scan the input voltage, what happens to our 2. 9mH value.
Let\'s try at the nominal 310 V.
As predicted, the frequency has been increased to compensate, but only 66231 Hz tons, and within the limits.
OK, so the final check is overload at 341 V.
Frequency 67K is still within the limit.
What happens if we increase the Vfv of the led to 7 v?
As expected, our frequency climbed again to 70 Khz, but still within the limit of 11 useccs shutdown times.
It may be difficult to regroup this, but let\'s give it a try.
So, it\'s always faster to buy a new one, what else do we learn.
It looks like the root cause of the failure is definitely a LEDs in the series string.
If a person dies, they all die when the circuit is turned on.
You need to replace the correct LED because the design is critical for series current and voltage, especially when driving directly from the power supply.
In the case here, they are definitely 150 mA varieties of 6v to 7v 1 W.
Shopping around it looks like a backlit led for telecom devices.
They are not expensive but if you pay 5 then it is half the cost of the flood but I bought 50, this should allow me to fix these for a while, maybe on another design.
I\'m going to include here the closed schematic of the board I\'m tracking, inserting component values as much as possible.
At the moment, I don\'t have the inductance value of the on-board inductor, but I might try to get it out of the on-board.
It is welded on both sides, which is a pain and can be destroyed in the process.
We have proven that we have some wiggle room in terms of design, the inductance will automatically adjust as long as the led arrives, I will fix and insert it, grab some images with this range, so review it in a few weeks and I\'ll update it.
I have also attached the excel spreadsheet with some numbers for you to play.
Maxitech is said to have its own design program for this chip, but I can\'t find it.
It would be great if my numbers match their numbers!
I hope you enjoy this waste of time, but maybe you have learned something. I know I have.
If you think this is interesting and may not even be useful, send me a message as always.
Footnote: Well, I bit the bullet and decided to remove the inductor from the board.
Yes, you guess the worst.
The bottom winding on the coil breaks, which means I have to painfully remove all the wires from the shaft of the line and rewind them.
There\'s no winding machine right now, it\'s hard, I counted 300 tons or just shy.
The wire I measured is 0. 17mm .
It is also a killer to install in the winding window with a winding ratio of less than 50%, but I managed it and measured the inductance on this coil as 2. 49mH@5. 8 ohms.
It would be even more exciting if we put those numbers into the spreadsheet.
With Ipk of 250 mA, we had a problem with insufficient henna, resulting in a swing to 80 Khz. . .
I don\'t like this.
If we go back to the higher peak, 327 mA is normal according to the resistor and the frequency drops to the 60Khz mark.
This is a compact design and there is not much room for the current to pass through the LEDs unless we add more inductance such as 3mH labeled aim.
This requires a larger core, a larger size and space.
In any case, we don\'t have to worry about this now, because as long as the inductance stays the same. . . .
I probably won\'t restart it, and then we just need to replace the dead LEDs with some new LEDs with the correct forward voltage and current, and we\'re done.
Oh, of course, you need to install them in the right way, the diodes of course.
OMG for those who have followed some of the bashing without falling asleep, you may notice something if you pay attention. . .
I probably don\'t know.
When we measure the existing inductor, it is about 2.
According to my inductive meter, 49mH.
It\'s a cheap one now, so I have another measurement 2. 84. . .
Anyway, the key to the problem is that it has opened nearly 300. . . It\'s in an electron
Core packaging of about 14mm x 10.
I also noticed that the core is unassembled, two electrons.
The core is held with some insulting tapes [sic]. Yes . . . . and. . . . well 300T of 0.
17mm of the wires will heavily saturate the core of this very small proportion, especially ungapped.
Try this using the ferroxcube program, or sit down and enter the number of the flux density.
Obviously it takes time and effort to gapping and this thing is cheap so what happened.
Why there are so many turns when ferrocube proggie says using about 100 and gap the core.
Interestingly, if you do not gap, the number of turns of the wire core will decrease, but the size of the wire core will increase with the size of the wire core. . .
More cost and space.
Apparently this is a government conspiracy. . . . . .
That\'s another forum. . . . .
No, I think there is a distribution gap in this core due to its material.
Although it has a slight ferrite, it is absolutely inadequate in terms of permeability.
The tradeoff of this size is replaced by the core loss, as we need more coils to produce the required inductance due to the increase in reluctance.
, Which causes the wire size suitable for the window to become smaller and the resistance to increase accordingly.
The loss of this copper core is I _ r = 6R x 163 mA = 150 mW. . . . ah ha. . .
Know where some of my 1w are going now. . .
Heat the case and try to get rid of the led string and let me buy another one. . . . .
I think so.
How about I connect a 10w COB diode directly from the power supply to the back of this flood and power. . .
There is an idea now. . .
Look at this space.
On November 2018, from the front of this series, you will see the led blocks that are removed from the floodlight.
Test all LEDs for any faults and remove them with a very hot iron.
Looking at the top of the board, you will notice a small pad corresponding to the lower side of the replacement LEDs.
The replacement LEDs and specifications I purchased from ebay are noted here.
Use the iron and tin paste again to properly secure the new LEDs on the board.
They should look like the last picture.
They don\'t look very beautiful, but they do work.
If you have a 30v power supply but check them with at least five strings or you will blow them up.
Or use the 9v battery mentioned earlier in this manual.
Note that the lid diffuser will go through these, so don\'t worry Robinson if they look a little healthy.
If you want to make it neat and tidy, then put them in the oven and melt them under and heat the circuit board with some solder.
End with a hot gun.
So now we repair the boards with new LEDs and have time to light them up.
The photo above shows that they are under diffuserSo, and after applying some energy, the device suddenly appears.
Fortunately, the LEDs are all faulty here, so it can be fixed cheaply. . .
I bought 50 LEDs for less than 10, each less than 20 p.
I only had one blast before I detonated the other tests, but even if you replaced all 9 times, it was less than a few pounds.
Since I have a while, I think I will connect the range and current probe to see the current waveform and from the schematic pdf you will notice that I have added three points to some graphics of the waveform: current limiting resistance at the internal fet source of the ic, capacitance on the Led string.
And the current probe attached to the blue lead on the top of the inductor.
The waveform is above.
The ground reference is the ground of the decoupling capacitor of the main input power supply.
Please note that I have been floating on the ground of the lamp using an isolation transformer and using variable power supply.
Whether you do not connect the oscilloscope to the power supply here without isolating the transformer, the grounding is not grounded, and the resulting explosion is not good.
The current survey has aroused interest.
Here you will see the main switching frequency of the chip I measured as 50 KHZ.
The opening time is about 16 us and the closing time is about 4.
The aliasing of continuous mode switching is very good.
The 4 us reset of the core winding is within the range of chip parameters and there is no obvious sign of saturation.
Using a voltage probe on the limit resistor, you can see that the voltage goes up to about 450 mV until the internal reset and drop.
The current peak seems to be 50 mA, which is surprising, as you can see from the specification, which is also lower than the ability of LEDs.
Although efficiency looks good, they do look very bright.
Anyway, the work is done, so I might design the cob version. . . . .