An update on the original board.

I just went over the board again with the same results as last time. Thought I would post them in hopes someone could give me a hint of how to fix it.

Here is what I found before and just found again now.

The fuse was blown. I jumped in a similar .5 amp fuse. The fuse did not blow but the light did not work either. I did notice that there was one small transistor that gets very hot. It is an S9013 and it's collector is connected to the gate of a F9540N power Mosfet. I believe this is the MOSFet that switches the solar panel off after the battery is charged. The solar panel is connected via a resistor to the source of the mosfet.


And if this circuit is malfunctioning, it might make sense that the light does not light because the board senses the voltage on the panel. It turns the the light on when the voltage is zero (or whatever low output they picked) from the panel.

Could that S9013 be shorted?

I think it's more likely that your MOSFET is blown. They are very sensitive to static electricity. When they blow, the gate can get shorted to the drain or the source. If shorted to the source, that would explain your S9013 getting really hot.

 

Have not pulled the LED yet but I hooked up the whole array to a current limiting resistor and my variable supply.

I increased the voltage starting from about 1.5 volts. The LEDs started to come on around 8 volts . There was one stubborn one that did not come on till about 11 volts. The array was at ideal brightness at around 12V. I did not want to push it beyond that.

What is very strange is there are 4 LED that are much brighter than all of the rest. They come on long before the rest of the LEDs at around 6.5V.

And there is that stubborn one I mentioned that barely lights.

Why is this? Is it possible these LEDs are past their life? Can their VF increase with age?

 

The s9013 could very well be shorted.

More likely that the MOSFET is shorted. Transistors are pretty rugged compared to MOSFETs.

The unwise method of powering could have fried your components.

Tell me precisely how the method of powering was unwise?

You would think that with about a 2v drop across each led, you would need quite almost 65v to light the set. Can you tell if they are wired in series or parallel?

It has already been established that their are a large number of strings wired in parallel; each string comprised of three white LEDs. Since white LEDs generally have a Vf of 3.4 to 4v at their rated current, they could be driven from a 12v SLA battery with proper current limiting.

Can you easily pull one led? That way you can have the numbers you need and get on with the build.

Not so fast! The Vf of LED's varies considerably, even in a single lot. I bought a lot of blue LEDs; the Vf at a given current was fairly consistent for about 75% of them; however there were a good number at the high and low ends that were +/-10% out of tolerance from the others.

One single LED is not a good enough statistical sample to determine the Vf of those in the array.

Also, draw out the wiring diagram for the leds. That will show how they are wired and in turn how they should be powered.

That's already known.

Strings of 3, no current limiting resistors, lots of strings.

 

I think it's more likely that your MOSFET is blown. They are very sensitive to static electricity. When they blow, the gate can get shorted to the drain or the source. If shorted to the source, that would explain your S9013 getting really hot.

Thanks that was my second thought. How would I test an FET? Would it just show short, if I removed it?


BTW I did remove the transistor and I get a HFE of around 124.

 

Thanks for all the help!!

I have a 330 ohm resistor on hand.

As mentioned, I also have a variable 0-32V 1A bench supply that I have built.


I think I will pull the LED tomorrow. And do some testing.

What is your test plan?

You better write it up beforehand, post it for peer review, just in case you are off base.

I sure hope we can get this working with something off the shelf or at least easy to solder up. As I said the nice thing about the buck puck is it provides me 5 VDC regulated.

I don't know where you are getting the 5VDC regulated figure from.

BuckPucks do not regulate by voltage; they regulate by current. The output voltage changes to keep the current flow constant.

 

What is your test plan?

You better write it up beforehand, post it for peer review, just in case you are off base.

I was a bad student. See above I already went ahead with my plan.

I don't know where you are getting the 5VDC regulated figure from.

BuckPucks do not regulate by voltage; they regulate by current. The output voltage changes to keep the current flow constant.

From the datasheet.

Page 1 Features:

Built in 5V to power logic circuitry or MCU.

 

Have not pulled the LED yet but I hooked up the whole array to a current limiting resistor and my variable supply.

I increased the voltage starting from about 1.5 volts. The LEDs started to come on around 8 volts . There was one stubborn one that did not come on till about 11 volts. The array was at ideal brightness at around 12V. I did not want to push it beyond that.

What is very strange is there are 4 LED that are much brighter than all of the rest. They come on long before the rest of the LEDs at around 6.5V.

And there is that stubborn one I mentioned that barely lights.

Why is this? Is it possible these LEDs are past their life? Can their VF increase with age?

You've found a pair of strings that have significantly higher Vf (those that are stubborn) and another string that were much brighter.

Those are examples of what I mentioned earlier; those that have significantly higher or lower Vf's than the other LEDs.

Remove those LEDs from the array, and measure their Vf at a known current; 20mA is preferred. Label each one. Report your results.

Hint: What is the current through a 330 Ohm resistor when there is a measured 6.6v across it?

 

You've found a pair of strings that have significantly higher Vf (those that are stubborn) and another string that were much brighter.

Those are examples of what I mentioned earlier; those that have significantly higher or lower Vf's than the other LEDs.

Remove those LEDs from the array, and measure their Vf at a known current; 20mA is preferred. Label each one. Report your results.

When you say "those" . I assume you mean test two, one of my superman LEDs and one of my weaklings? Should I use 9VDC?

Hint: What is the current through a 330 Ohm resistor when there is a measured 6.6v across it?

Hey it has been a long time since I looked at Ohms law. In fact I think it was pretty new back then. And I am suffering from a cold but 20 ma?


But now I am going to bed which I should have done an hour ago. More testing tomorrow.

 

Tell me precisely how the method of powering was unwise?

My "The" should have been an "An". ie. An unwise method of powering could have fried your components.

It has already been established that their are a large number of strings wired in parallel; each string comprised of three white LEDs. Since white LEDs generally have a Vf of 3.4 to 4v at their rated current, they could be driven from a 12v SLA battery with proper current limiting.

True. The op did state the wiring. However, it was never noted in this thread the general Vf of white LEDs. I didn't want to assume the OP knew that as of this point.

Not so fast! The Vf of LED's varies considerably, even in a single lot. I bought a lot of blue LEDs; the Vf at a given current was fairly consistent for about 75% of them; however there were a good number at the high and low ends that were +/-10% out of tolerance from the others.

Being the unit was OTS, and already being used in an array arrangement, even a +/-10% measurement of a single LED would have given a starting point to research the likely Vf range.

One single LED is not a good enough statistical sample to determine the Vf of those in the array.

If our OP was pulling the LEDs to use in 36 separate devices, it would not be a good sample. For an overall average, after lighting the array, pulling a LED of average brightness in the array, would prove logical. And would also help in a design using measurements for the LEDs taking in account age.

I'm trying, sarge.

 

Sarge,

I understand what you wanted me to do now.



At 20ma, I measure 3.18 V across one of my "dim" LEDs. I am measuring 3.16 across one of the "bright" LEDs.


And here is something curious.

With one of the "bright" LEDs disconnected, the whole panel lights starts to light around 8.7 VDC.

I'd bet I could do better if I disconnected those other "bright" LEDs. I would be willing to do that or replace them if it meant light the whole panel at a lower voltage.

 

One cheap Chinese LED is bright and another cheap Chinese LED is dim.
In a couple of days the bright one will also be dim.
In a week or two they will not produce any light.

 

One cheap Chinese LED is bright and another cheap Chinese LED is dim.
In a couple of days the bright one will also be dim.
In a week or two they will not produce any light.

Ahh, the sad, true fact of poorly made goods. (poorly / goods...ironic)

 

I'd bet I could do better if I disconnected those other "bright" LEDs. I would be willing to do that or replace them if it meant light the whole panel at a lower voltage.

You would. The fewer LEDs, the fewer voltage drops. The less power needed to light the array. However, since you are willing to pull more LEDs or replace them, you may want to do just that. You can pick up higher quality LEDs for cheap, and have a more reliable product. You could even use the same PCB. You can cut the traces and make it into two 18 LED arrays.

Lots of options. What does your finished design require?

 

You would. The fewer LEDs, the fewer voltage drops. The less power needed to light the array. However, since you are willing to pull more LEDs or replace them, you may want to do just that. You can pick up higher quality LEDs for cheap, and have a more reliable product. You could even use the same PCB. You can cut the traces and make it into two 18 LED arrays.

Lots of options. What does your finished design require?

This is my overall plan.

We have a sign in front of our townhouse community. We want to illuminate it but we do not have power to that area and it would be very expensive to run power there. So that is why we purchased the solar light.

As mentioned the controller / driver board is fried. And worse I just noticed a Mosfet (not the one that was for the panel switch and was suspected of being shorted but the one for the light) is now gone! I just noticed that one of the legs was not soldered properly. What I can't figure out is how the other legs broke. Perhaps they were previously weakened through careless handling. It has to be around here somewhere I hope it shows up.

Anyway due to no schematic and the condition of the board, it is probably a lost cause getting it to work.

So I guess I am starting from scratch. I would like to illuminate that sign. It is (and I am guessing here) 6 ft. x 3 ft.

I really don't want to build my own buck regulator (if I can get away with it) and would like to use one of these.

It would be nice to save the light but I am not against rebuilding that too. I could use the current fixture but with new lights. If they are brighter and more efficient, then that would be great.

I would like to power the light with a 12V lead acid battery (I have a couple UB1250 batteries) but I am not against going to a battery pack.


And then there is charging. I am hoping the current panel will charge the battery. I did some testing and calculations and it looks like it might take 2 days to charge from a dead battery. If that is what it takes then I can live with that for now but if the project all works then this would be an upgrade in the near future.


Then there is actually charging the battery. I was considering an off the shelf charge controller but also might use a PIC to control the charging.
I want to use a PIC anyway to control the light. So I can turn it on after a settable timer after "dark", turn it off after a certain time period and perhaps turn the light on and off via some kind of sensor like a motion or light sensor (for headlights).

But for now the plan is to simply have a way to drive those lights. What I would like to do is to get the current light working. Even if I need to remove some LEDs. Keeping in mind that I will be replace all or part of them with something better that also works with the driver that I select.

Once I get the whole project working, I can start upgrading. Upgrade one would probably be improving the light.

 

Ok, we can do this. Let's design the load, then we will design the control circuit. It could be as easy as a LDR (light dependent resistor). That way, it turns on automatically at night and off during the day. That means no need to reprogram the PIC for daylight savings times and season changes. As well as overcast days and the such.

 

Ok. Lets get moving. The amount and type of leds(the LOAD) is going to determine the battery life and charging requirements. If you can get 9 hi-power leds, in your reflector, you should be able to light the 6x3 sign from 3 feet away or so, with no problem. The Leds would be run in strips of 3, 3 wide. That will use about 60ma of current.
A 12v 5Ah battery would give you ~80hours of use before needing recharge(10.5v). So even a few cloudy days wouldn't affect you.

Here is a link to the type of battery you want:
http://www.batterymart.com/p-12v-5ah-sealed-lead-acid-battery.html

Ive attached a wiring diagram to show you how to build the 9 led array.

You may wish to bread board it first to make sure you have enough light coverage on the sign. If you need to add some leds, we will.

Now we need to get some info on the solar panel to get the charging under control.

The attachment is based on 3.5v 20ma LEDs. Some white ultra brights 5mm.

 

Ok. Lets get moving. The amount and type of leds(the LOAD) is going to determine the battery life and charging requirements. If you can get 9 hi-power leds, in your reflector, you should be able to light the 6x3 sign from 3 feet away or so, with no problem. The Leds would be run in strips of 3, 3 wide. That will use about 60ma of current.
A 12v 5Ah battery would give you ~80hours of use before needing recharge(10.5v). So even a few cloudy days wouldn't affect you.

Here is a link to the type of battery you want:
http://www.batterymart.com/p-12v-5ah-sealed-lead-acid-battery.html

Ive attached a wiring diagram to show you how to build the 9 led array.

You may wish to bread board it first to make sure you have enough light coverage on the sign. If you need to add some leds, we will.

Now we need to get some info on the solar panel to get the charging under control.

The attachment is based on 3.5v 20ma LEDs. Some white ultra brights 5mm.

Why would I need current limiting resistors, if I am using an LED driver?

The new lights would only use 60MA?

What is my load of the exiting light? I have 2 sensing resistors on the board. They are .33 ohms each in parallel giving me a total of .165 ohms. The formula in the datasheet says Iout =.1 / Rs so that is 606 milliamps. Why the big difference?


I ran my array through the same wizard and it suggested the same array that I have but added the current limiting resistors. It says it will draw 600ma which is consistent with what I calculated from my sensing resistors.

Why the huge difference in the number of LEDs needed? Are the ones you suggested that much brighter?

 

The datasheet is using 1w and 3w leds, the leds you have are approx 1/4 watt. You will need some heavy duty LEDs.

You will not need an LED driver for the project. you can use it just like I said with this attached circuit. You would attach the led array to the area where the single led is. (we need to use a transistor that can handle 60ma)

This will light the sign when it is dark outside, and off when it is light outside.

There is a million ways to design this. If you want to use a buckpuck, it can be done. There is not a need however. The attached circuit is less than $1 in parts. Then a $13 battery. and $4 in ultrabright leds. The Solar panel can likely be attached to the battery easily for daylight recharge.

 

If you want to use the higher wattage LEDs, we can still do so.

 

If you want to use the higher wattage LEDs, we can still do so.

What LEDs in the 2-3 watt range would you recommend?

Say from DigiKey. Also something easy to solder up. Surface mount would be difficult.