Why not - Convert your mains LED floodlights to solar? (actually there are several reasons not to, but I did it anyway).

[Crossy]

A spot of history

We have a number of these baby 220V 10W LED floodlights, their main advantage is that they are small and hang on the wall almost invisibly.

 

 

Over the years we have had a number of them fail, sucumbing to power surges and simple old-age. They are on all night, every night controlled by a photo-switch.

 

Being a lazy sod when one failed I've simply bought another, the net result being we have a box of dead fittings looking to be resurected.

 

With Madam in "green" mode everything that can be has to be solar powered.

 

Enter this little project.

 

Here we go

It's the work of a few minutes with a screwdriver to pull the beast apart.

 

 

In all the dead ones I've looked at both the LED driver and the LED (actually a COB panel) have been dead.

 

So a bit of shopping is needed:-

• 10W COB units - https://www.lazada.co.th/products/i1764906459-s5121368195.html 30 Baht a pop.
• DC LED driver (actually it's a universal step down converter so you can use it as a battery charger too) https://www.lazada.co.th/products/i1591488632-s4324320455.html 79 Baht a time.

 

Replace the fried COB in the light with a new one, ensure you get the polarity right and use heatsink ointment or a heat transfer pad.

 

Before going mad connecting it all together a little setup is needed.

Connect the driver input to a power supply (a 12V battery will do) and wind the Constant Voltage pot fully clockwise (it will click). Then with your multimeter on DC amps connect it across the output terminals. These units seem to come set to 3A or so, so adjust the Constant Current pot until the current is down to 0.7A (700mA) or thereabouts.

 

Now you can connect the COB to the output terminals.

 

 

Turn on the power and the light should light!

 

Again with your meter on DC amps check the current through the COB and adjust to about 700mA. The COBs I have seem to be at maximum brightness at 680-700mA so there's no real point going any higher despite the spec. saying 750mA. and every mA we save is extra battery life.

 

Slide the driver into an insulating tube, I used some battery tubing that was in-stock but anything insulating will do.

 

 

Add a zip-tie to the cable as strain-relief, button it all back together and test again.

 

Bingo we have an LED floodlight that will work on any DC voltage between 12V and 30V.

 

 

In the next exciting episode we will look at the power source for these chaps.

 

[Thaifish]

 

2 hours ago, Crossy said:

every night controlled by a photo-switch.

 

Do they still have a sensor switch??

[Crossy]

 

1 hour ago, Thaifish said:

Do they still have a sensor switch??

 

Patience young Grasshopper. Patience ???? 

[Crossy]

OK onwards and upwards.

 

I'm running two of the modified lamps, illuminating our parking area.

 

 

So we have a 17Ah (measured) x 12V LiFePO4 (4S3P) pack with a 30A BMS.

 

 

Batteries were from stock, BMS is this one https://www.lazada.co.th/products/i2294264016-s7728105060.html at 90 Baht plus the optional 15 Baht heatsink, not really needed in this application.

 

And a 12V photo switch https://www.lazada.co.th/products/i1650986683-s4583006216.html at 49 Baht.

 

 

There's an 80W panel on the verandah roof https://www.lazada.co.th/products/i1341840835-s7487117650.html 1,100 Baht.

 

 

The panel is currently connected directly to the battery pack and charges reasonably well (relying on the BMS to prevent over charging/discharging), but it gets shaded later in the day so I have a MPPT board on order (coming from China).

 

Currently in "prototype" (spead all over the table).

 

With the lights on they pull about 1.2A from the pack, so 17Ah should last about 14 hours, in reality the lights are on for about 12 hours. The main issue is whether the partly shaded panel can get that much charge back in during the day, hence the MPPT board on order.

[Crossy]

Whilst we are being green I may add the camera to the load too. That will need more batteries and certainly another panel. 

[maxpower]

The challenge with DIY solar lights is integrating the light control, charge and battery management without consuming too much current in the process. 

 

Most of the control chips employed in budget Chinese lights operate in the micro amp range. These products get the job done but without much consideration for long term reliability or battery condition.

 

 

 

We have some solar security lights around our facility all built in house using mostly off the shelf modules tweaked to fit.

20W led arrays are driven at around 10W through a constant current driver PT4115 powered up by separate control MOSFET. 

15Ah LiFePO4 3P4S packs provide the stored power with each having its own BMS and charge controller using the basic MPPT module (CN3722) connected to 50W 18V solar panel.

A very simple low power comparator circuit monitors the solar panel voltage. When the output reaches dark level it powers up a ATtiny85 processor which in turn operates the the LED's and monitors the battery voltage via analogue input.

The processor makes no PWM adjustment based on battery condition and simply cuts the light out when the voltage starts to dive. This is because the flat line discharge curve of LiFePO4 is difficult to monitor for capacity. 

 

 

Photo of the original test prototype before PCB design.

Left green light is MPPT module showing charge complete. Center is the comparator board right is the ATtiny85 board. 

 

 

[Crossy]

3 minutes ago, maxpower said:

A very simple low power comparator circuit monitors the solar panel voltage.

 

Yeah, the photo switch uses a 555 (a well known power hog) plus a relay (another power hog), but it's quick and simple and really doesn't draw that much juice.

 

Definitely one of the things on the "improve" list.

 

My aim here is to do something our non-engineer members can do and feel confident that it will actually work ???? 

 

[maxpower]

40 minutes ago, Crossy said:

 

Yeah, the photo switch uses a 555 (a well known power hog) plus a relay (another power hog), but it's quick and simple and really doesn't draw that much juice.

 

Definitely one of the things on the "improve" list.

 

My aim here is to do something our non-engineer members can do and feel confident that it will actually work ???? 

 

 

If you want to continue with the 555 in a light level circuit consider using the CMOS version and a MOSFET to switch the light.  Very simple arrangement most can follow.

[Crossy]

1 minute ago, maxpower said:

If you want to continue with the 555 in a light level circuit consider using the CMOS version and a MOSFET to switch the light.  Very simple arrangement most can follow.

 

Yeah, I thought about the CMOS 555.

 

I've got a CD4001B based, mains powered, switch being modified on the bench at the moment, still got a relay but the CMOS device is a lot less thirsty than the 555.

 

[maxpower]

9 minutes ago, Crossy said:

 

Yeah, I thought about the CMOS 555.

 

I've got a CD4001B based, mains powered, switch being modified on the bench at the moment, still got a relay but the CMOS device is a lot less thirsty than the 555.

 

 

There are complete all in one packages available for those wishing to build high power solar lighting without too much effort. Just add custom light and battery.

 

Edited May 9, 2021 by maxpower

[Crossy]

19 minutes ago, maxpower said:

 

There are complete all in one packages available for those wishing to build high power solar lighting without too much effort. Just add custom light and battery.

 

... and a suitable panel.

 

22 USD (700 Baht or so) each on AliBaba.

 

But it really takes the fun out of the game.

 

[Crossy]

Been doing a bit more testing of the cheapo photo-switch.

 

Idle current at 12V is 4.0mA increasing to 65mA at 20V (panel would be open-circuit, cells fully charged) it has a simple zener/resistor regulator, 12V 1W zener + 10 ohm resistor. With the panel open the 65mA would mean the 12V zener is dissipating .78W (it's a 1W zener).

 

I have some 15V 1W zeners, I may try one instead of the 12V beast (555 is good to 16V supply).

 

With the relay operating and 15V supply it's pulling 50mA = 750mW.

 

Looking at replacing the relay with something less sucky.

 

But in reality the effect on the battery life is small.

 

 

[maxpower]

3 hours ago, Crossy said:

Been doing a bit more testing of the cheapo photo-switch.

 

Idle current at 12V is 4.0mA increasing to 65mA at 20V (panel would be open-circuit, cells fully charged) it has a simple zener/resistor regulator, 12V 1W zener + 10 ohm resistor. With the panel open the 65mA would mean the 12V zener is dissipating .78W (it's a 1W zener).

 

I have some 15V 1W zeners, I may try one instead of the 12V beast (555 is good to 16V supply).

 

With the relay operating and 15V supply it's pulling 50mA = 750mW.

 

Looking at replacing the relay with something less sucky.

 

But in reality the effect on the battery life is small.

 

 

 

I am guessing the zener resistor arrangement comes from when a the sensor was powered via capacitor dropper.

 

With the 555 having absolute max rating of 18V and being across the battery (4s)  I cant see why you need a zener.

[Crossy]

8 hours ago, maxpower said:

I am guessing the zener resistor arrangement comes from when a the sensor was powered via capacitor dropper.

With the 555 having absolute max rating of 18V and being across the battery (4s)  I cant see why you need a zener.

 

Yeah, replacing the 10R with a suitable capacitor would make it into the mains powered version (it has a bridge already too). The zener/resistor does give it protection from over voltage, the panel is 20V when open circuit (BMS has cut off due to full charge). 

 

I'll stick a 15V zener in instead that will save a few mA and get out the text books (ok, Google) to learn about FETs as switches instead of the relay, my education is strictly bipolar, it was a loooong time ago (even JFETs were new).

 

I do like the idea of using the panel as the light sense element, low power comparitors here we come ???? 

 

[maxpower]

4 hours ago, Crossy said:

 

Yeah, replacing the 10R with a suitable capacitor would make it into the mains powered version (it has a bridge already too). The zener/resistor does give it protection from over voltage, the panel is 20V when open circuit (BMS has cut off due to full charge). 

 

I'll stick a 15V zener in instead that will save a few mA and get out the text books (ok, Google) to learn about FETs as switches instead of the relay, my education is strictly bipolar, it was a loooong time ago (even JFETs were new).

 

I do like the idea of using the panel as the light sense element, low power comparitors here we come ???? 

 

 

Are you connecting the solar panel directly to the BMS without any regulation?

[Crossy]

59 minutes ago, maxpower said:

Are you connecting the solar panel directly to the BMS without any regulation?

 

At present yes, peak current is about 3A and the BMS stops any overcharge. Not ideal but the MPPT regulator is somewhere in transit. 

[maxpower]

1 hour ago, Crossy said:

 

At present yes, peak current is about 3A and the BMS stops any overcharge. Not ideal but the MPPT regulator is somewhere in transit. 

 

Yes its not ideal because when one cell goes over voltage the whole pack gets disconnected from the charge circuit.

 

Having the correct supply voltage to a BMS ensures the balance dump circuit across each cell can operate without too much overshoot.

 

Supply the BMS with a voltage too low and the balance might never operate until a cell really goes south.

 

The MPPT module in my previous post has its output voltage trimmed to suit the 4S pack and BMS (approx 14.6V). For most of the time the module will disconnect the charge current before BMS gets to disconnect.

 

The big question is, does the pack really need a balance circuit if all is equal from the start and charge/discharge continues in a stable window.

 

I am told the 16S pack on our test bench has been happy cycling without balance for over two years.

 

[Crossy]

The pack has just (12:15) decided it's full on the BMS, the 80W panel may be a little over-sized but that would make up for the dull days (pretty sunny today). 

 

Cells range between 3.63 and 3.66V, close enough.

 

Obviously this is very much a work-in-progress, more updates as things develop.

[Crossy]

The baby MPPT controller has arrived! This one from China 

https://www.lazada.co.th/products/i1949962010-s6191428424.html 

 

Took a few minutes and some Googleing to set it up (no instructions of course) a quick tutorial is here

https://robotzero.one/5a-mppt-charger-instructions/

 

Here it is charging at about 2.5A (80W panel at 9AM), to the right is the Mk1 version of my "dark detector" which uses the panel as the sensing device, more of that later (we are already at Mk3).

 

[maxpower]

4 hours ago, Crossy said:

Took a few minutes and some Googleing to set it up (no instructions of course) a quick tutorial is here

https://robotzero.one/5a-mppt-charger-instructions/

 

The setup link provided suggests the writer could not find the CN3722 manufacturers data sheet.

 

http://www.consonance-elec.com/pdf/datasheet/DSE-CN3722.pdf

The first procedure with this module is to apply source voltage and set the output to max charge voltage (approx 14.4 for 4S LiFePO4) This will be the point at which charge ends and LED outputs change from charging to charged.

 

Small adjustments can be made to the voltage when the battery charge complete state is known.

 

Charging and charged LEDs will toggle (flicker) if there is no battery load. This can be useful in identifying if a BMS has disconnected the battery before CN3722 charge cycle has completed.

 

With charge load applied the MPPT initial setting can be made by monitoring panel input voltage and adjusting the CN3722 until it is at the panels optimum voltage. Tweaks can then be made by monitoring the output power.

 

These devices have no smart MPPT control beyond a basic adjustment window based around what the user has set at the MPPT divider input. Efficiency of these modules varies and is often down to how well the output inductor matches average charge current.

 

It would be interesting to compare performance of these basic MPPT modules to one of the very efficient PWM controllers kicking around the DIY scene.

[Crossy]

Yeah ^^^ that's pretty much what I did once I'd worked how to get the output to actually come on (so his instructions were handy).

 

The max charge current is with the panel at about 16.5V I shall leave it like that.

[Tronx]

Thanks for detailing all your steps, I really enjoyed your car port stuff. I made a charging controller for my 10Ah LiFePo4 with the ICL7665, and I'm charging via a 20W panel. In this way, the current will never exceed the battery happieness level. A 100W panel gives 6A which is too much.

My claim is: The lead acid or LiFePo4 4s battery voltages are very close to the MPP of 12V rated solar panels which have open circuit 24V or similar. The battery pulls the voltage down to 13V  which is where the solar panel has its maximum power output, and with the efficiency of the MPPT device, there will be no improvement.

Could you run a test with your MPPT on the same 12V battery, and without the MPPT and post the results here?

[Crossy]

8 hours ago, Tronx said:

My claim is: The lead acid or LiFePo4 4s battery voltages are very close to the MPP of 12V rated solar panels which have open circuit 24V or similar. The battery pulls the voltage down to 13V  which is where the solar panel has its maximum power output, and with the efficiency of the MPPT device, there will be no improvement.

 

Whilst I haven't done any formal testing I don't think you are far off the mark.

• With no controller (direct panel to battery connection) the battery was full (BMS turned off) at about 12.15PM.
• With a 20A schottky diode in the loop (part of my Mk-1 "dark detector") the result was very similar.
• With the MPPT it actually took slightly longer to go to full (MPPT indicating charged).

Of course the "test light source" (the sun and intervening clouds) is uncontrollable and variable.

 

Letting the panel run at it's measured MP point (16.5V for the panel I have) has to be more efficient, but the losses in the controller could well offset that (it does get quite warm).

 

I strongly suspect that the actual gain in energy from using the MPPT in this case is small. But as @maxpower noted in one of his earlier posts limiting the charging voltage to below the BMS cut-off level will allow the BMS balance circuits to do their job more effectively.

[Crossy]

Now, what about that darkness switch?

 

Poking around on the net produced some designs for very simple methods to use the solar panel itself as the light sensor.

 

So the Mk-1 version came out like this. The opto is part of a FET switch module that I had in the bits box which switches the lights.

 

It actually works remarkably well turning on the light once the panel voltage gets to about 2V below the battery voltage.

 

The battery is charged via D1 (there was no charge controller at this point). When the panel voltage goes below the battery voltage (it gets dark), Q1 is turned on via R1 and D2 (R3 provides a path to -ve if the panel has a reverse blocking diode). Q1 provides current to the LED and opto, limited by R2.

 

It does have some problems, it's not adjustable and the lights do tend to come on a bit early and go off late, and there's a short time where the switch is "half on" which isn't good for the FET (gets warm).

 

With D1 removed it continues to work with the MPPT charge controller. 

 

EDIT I thought of adding a zener in series with R1 and D2 to move the threshold lower and maybe some positive feedback from the switched output to hurry the changeover, but it was starting to get complex so I never built it. The updated version coming soon.

[Tronx]

This slow turn on/off needs either huge oversized FETs, or maybe a more complex approach. As you have a few of these lights to fit, did you consider a solution with comparator?

I've recently used TS27L2C opamps for a similar purpose. They are LM358 pin compatible, draw 10uA and are with 35cents still affordable. I bought them even on Aliexpress. The advantage would be, that you can set the turn on point exactly where you want, and hysteresis should also be possible to prevent blinking lights and such ugly stuff.

ts27l2-1852517.pdf

[Crossy]

3 hours ago, Tronx said:

This slow turn on/off needs either huge oversized FETs, or maybe a more complex approach. As you have a few of these lights to fit, did you consider a solution with comparator?

 

Yes, the Mk-3 which is under test at present uses an LM358 (as I had a tube in stock). Watch this space ???? 

 

As with my other projects it's my intention to show the development process, warts and all, rather than just present the "look how clever I am" final solution. If my workshop doesn't smell of fried electronics it's not been a successful day (those MOSFET drive boards don't have short-circuit protection, ask me how I know).

 

It's a bit of fun and lets me exercise the electronics design brain cells which have not been used in many years (I let other people do that now).

 

[Muhendis]

On 5/15/2021 at 9:28 AM, Crossy said:

The baby MPPT controller has arrived! This one from China 

https://www.lazada.co.th/products/i1949962010-s6191428424.html 

Thanks for that one. I've been looking for something like this for ages. Got a couple coming from Alibaba. I'm doing something similar having discovered no street lights with IR detectors and separate solar panels. 

Please don't tell me they're everywhere. I'm committed to this project now.

My thoughts about the light detector are similar to others regarding using a comparator + 3 or 4 resistors and a zener/TL431

[Crossy]

Since I have your attention a quick preview of the Mk-3 sensor.

 

U1A is a schmitt-trigger centred around 1/2 the battery voltage (it's not regulated but we're not after ultimate accuracy), with the value of R4 at 4.7M the hysteresis is about +- 0.5V, enough to stop the light flickering at switch over. No I didn't remember how to calculate the hysteresis 

https://www.petervis.com/dictionary-of-digital-terms/schmitt-trigger-op-amp/schmitt-trigger-circuit-using-op-amp.html

 

U1B is a unity gain buffer, not really necessary but the op-amp was in the package.

 

R7 and D1 are there to prevent us ever taking the amp input outside the supply, it doesn't like that.

 

Current draw with an LM358 is about 1mA at 12V supply to just under 2mA with the lights on (most of the extra is lighting the LED). Not sure if it's really worth going to an ultra low-power device.

 

[Muhendis]

PWM between R9 and R3?

[Crossy]

12 minutes ago, Muhendis said:

PWM between R9 and R3?

 

It's the terminal name of the XY-MOS module, it's not actually PWMing.