LED Battern lighting is making my video flicker

[John B]

You say this is for a work/business context. There's a few things to keep in mind. I would be careful about modifying mains powered stuff (especially cheap stuff) then using them in an unattended situation. It's also a legal problem.

The safest way would be to buy an off the shelf AC/DC constant current LED driver which is capable of outputting sufficient voltage. This way all the mains isolation and certification is taken care of.

For rough figures, you'd be looking at $50 per driver, 1 per panel.

Another option is to cut the losses on these panels and invest in a more standardised system. Ie 1200mmx300mm passive panels running at around 36V, with external drivers. This way you can pick the panels you want, specifically the highest CRI ones you can get, then pair them with some brand name drivers. Panels you buy will come with drivers, but will most likely have the high AC ripple issue.

[Automate007]

You say this is for a work/business context. There's a few things to keep in mind. I would be careful about modifying mains powered stuff (especially cheap stuff) then using them in an unattended situation. It's also a legal problem.

The safest way would be to buy an off the shelf AC/DC constant current LED driver which is capable of outputting sufficient voltage. This way all the mains isolation and certification is taken care of.

For rough figures, you'd be looking at $50 per driver, 1 per panel.

Another option is to cut the losses on these panels and invest in a more standardised system. Ie 1200mmx300mm passive panels running at around 36V, with external drivers. This way you can pick the panels you want, specifically the highest CRI ones you can get, then pair them with some brand name drivers. Panels you buy will come with drivers, but will most likely have the high AC ripple issue.

John your 100% right. Stupidly I Ordered these lights and installed before I should of tried 1 and assess with the camera's when we were at the building stage. But I do have 2 jobs and time was not on my side and rushed into it. Lesson learnt. My issue is also I was stupid to think going LED that there would be no frequency flicker..... I assumed wrong! I'm now torn as the ceiling is a concrete slab with no where to really hide drivers and not easy to mount LED panels to concrete let alone rerun all the cabling. Aaargh I'm so annoyed with myself

My issue now is If I do buy drivers and tie it into the existing lights how can I be 100% sure they are ripple free dc output drivers?  Anyway this is a loosing battle and only have my self to blame.

Thanks guys for all the responses. Good advise given and hopefully this will help the next guy similar to my predicament!

Advice. Buy a few different options .... test.... then buy bulk.

[John B]

My issue now is If I do buy drivers and tie it into the existing lights how can I be 100% sure they are ripple free dc output drivers?  Anyway this is a loosing battle and only have my self to blame.

I have the most experience with the HLG series of drivers from Meanwell. They are more like AC/DC power bricks in their output waveform, ie clean DC, with switching noise. Also the datasheet for these types of drivers will give a Vpk-pk figure which you can gauge how much ripple you'll be getting, typically a few hundred mV. This will include the switching noise up to 20MHz, it wont be mains frequency ripple.

The problem is these integrated panels will often run fairly high voltage for efficiency. While there's nothing technically difficult about driving them, actually finding suitable equipment at a cost-effective price will be difficult. For example it wouldn't be too technically difficult to make your own, or buy a linear constant current module for each panel, then feed each panel with ~160V DC from a single AC/DC power supply, that's outside the extra low voltage specification, so there's much less choice out there in buying those power supplies. It's still legally a no-no at that voltage  Bloody red tape.

Then you have to weigh up sinking money into this specialised voltage setup, and whether you'd be able to repurpose the equipment for anything else. At least your typical 36V panels and drivers can be mixed and matched, and reused. You could even use the same principle of cheap, small low drop out linear drivers for each panel and a single AC/DC converter. Assuming you don't need dimming or colour control. That's real can of worms. I'm building something similar to your needs with dimming and colour control. It's a pain.

Also with passive panels, you could just run the wire to each panel and have drivers in a cabinet somewhere to simplify the mains wiring. Constant current drivers will be compensating for any voltage drop in the wires, not that there's much with LED panels, especially if you use 1.5mm2 wire or something.

It's a tough call. Can you salvage the existing lights for a shed or patio or something?

[John B]

Just another thought on the idea of a single AC/DC power supply feeding multiple constant current drivers - you don't even need 1 driver per panel. Large strings of LEDs actually balance themselves pretty well in parallel when run at modest powers and currents. Meaning if you had panels than ran at 700mA typically, a 1.4A driver could power 2x panels in parallel. You don't want to go much further though, just in case an LED ends up shorting on one string then that driver ends up dissipating too much heat. I've actually been running 4x 60V panels that way for several years and there's been no issues.

Colour temperature adjustable panels actually work in this way. There's 2x strings of LEDs, one at ~2700K, then one at ~6000K. Then a constant current driver either is applied to one or the other, or both strings.

[Automate007]

Just another thought on the idea of a single AC/DC power supply feeding multiple constant current drivers - you don't even need 1 driver per panel. Large strings of LEDs actually balance themselves pretty well in parallel when run at modest powers and currents. Meaning if you had panels than ran at 700mA typically, a 1.4A driver could power 2x panels in parallel. You don't want to go much further though, just in case an LED ends up shorting on one string then that driver ends up dissipating too much heat. I've actually been running 4x 60V panels that way for several years and there's been no issues.

Colour temperature adjustable panels actually work in this way. There's 2x strings of LEDs, one at ~2700K, then one at ~6000K. Then a constant current driver either is applied to one or the other, or both strings.

Ok now thinking out loud.

Forget the 3 colours I only need the 5700k LEDs  so this eliminates the secondary warm LED wire.

So the 320 White LEDs by themselves draw 0.380amps at 120VDC.

The shop front I can live with the lights on the concrete ceiling as they are...  in reality we won't be running any live streaming in there for a while so for now I can concentrate on the teaching room which has a a new gyprock ceiling and a cavity from the mezzanine above. Now that I have completely dismantled 1 light I can see a number of options.

1. Find 2 x 470uf (or higher) caps under 18mm diameter each and run them in parallel in the existing board space . It will be tight but should fit but not sure how low I can get the ripple

2. Remove the cap and extend with a 2 core cable out of the light fixture and into the ceiling cavity and mount the larger 1200uf cap in a sealed junction box. Crazy but do able

3. If I unscrew the ends I can slide out the diffuser in which I have a cavity between the diffuser and LED board of 21mm . I could in theory solder the caps directly to the DC plug feeding the White LEDs and glue(silicon blob) the caps in there on the LED aluminium panel and parallel 2 (or 3) x 470uf (18mm dia) in there not modifying the existing driver ..... do I Remove the original 100uf or keep it in there?

4. Or Remove the existing driver board altogether and extend the 0v and 120v wires from the panel to the outside and connect it to a good driver in the ceiling cavity.  This is the safest option I think.

Maybe next year if the business does ok then I'll rip em all out and get expensive CRI rate LED panels but for now it's all we can afford as we have literally put in all our savings to make this home hobby business into a reality. We will succeed but covid killed us for 2 years with no cakes or functions and all our stock from manufacturers stopped.... resulting in no income for 2 years...  so lucky my day job carried us through till now? I know get the violins out but this is my dilemma! Just trying to catch a break for once

[John B]

Understood, you also need to consider the time and effort, and even the expense of say 20 junction boxes. Let's say they're $10 each, that's already $200 of budget on useless plastic.

What would be your budget for fixing everything?

[Automate007]

True. Keep in mind that room has 8 lights. I can do about $400 to fix just that room for now

[John B]

There's one last option that I can think of, which Stray Electron mentioned at the start of the thread. Keep the existing drivers as they are and then power the lights with DC, or rectified mains.

I'll have to follow up on this later, but perhaps a 1.5 kVA step down transformer 240V -> 120V used for powering US equipment, then rectified and filtered. This will provide up to 170V DC with ripple depending on the output capacitance. Rectifying the mains is probably an option, but ~340V DC low impedance is getting a bit sketchy.

A 120V DC power supply is also an option, but it's going to be out of the budget.

[Automate007]

There's one last option that I can think of, which Stray Electron mentioned at the start of the thread. Keep the existing drivers as they are and then power the lights with DC, or rectified mains.

I'll have to follow up on this later, but perhaps a 1.5 kVA step down transformer 240V -> 120V used for powering US equipment, then rectified and filtered. This will provide up to 170V DC with ripple depending on the output capacitance. Rectifying the mains is probably an option, but ~340V DC low impedance is getting a bit sketchy.

A 120V DC power supply is also an option, but it's going to be out of the budget.

Thanks John for your time. I finally found a 680uf 160v cap that is 18x40 from elements14. I have worked out I can make 2 fit in a cavity beside the driver .... 1 on each side. I will try with 1 680uf + the 100uf on the original board might just be low enough that my cams can deal with it. If not I'll add a second 680uf on the other side. If that doesn't work. I'll remove these lights and will have to search for flicker free panels that will cover the holes of these units.

Hoping I will get thd caps delivered by early next week and will report back my findings.

[Automate007]

There's one last option that I can think of, which Stray Electron mentioned at the start of the thread. Keep the existing drivers as they are and then power the lights with DC, or rectified mains.

I'll have to follow up on this later, but perhaps a 1.5 kVA step down transformer 240V -> 120V used for powering US equipment, then rectified and filtered. This will provide up to 170V DC with ripple depending on the output capacitance. Rectifying the mains is probably an option, but ~340V DC low impedance is getting a bit sketchy.

A 120V DC power supply is also an option, but it's going to be out of the budget.

I have received the 680uf caps today and applied 1 in parallel with the 100uf and it takes 2 seconds to turn on and about 2 minutes when switching off to get down to 0vdc.

To make matters worse even at a 3v ripple it still flickers. Not as bad but still unacceptable.

I'm so confused as to my best option now that will work 100% . I have now resigned to the fact there is no way to fix this . Anyone can point to a safe solution that has no ripple and is hopefully under $100 per light. I'm all ears!

[John B]

Yeah that sounds like my experience, you're trying to modify an AC/DC topology that inherently has a fair bit of ripple. The best solution that I can think of is to look at using a CC LED driver, and remove the existing drivers from the panels, ie convert them to passive panels.

Consider something like this:
https://au.mouser.com/ProductDetail/MEAN-WELL/XLG-100-L-AB?qs=sGAEpiMZZMvV8Y9YugmIggXPjVR9iwOZ2ARXrvaN0alPoJ9X1te3Xw%3D%3D

$55 in single qty and $50 in tens. The current output has 3% ripple, note that your voltage ripple will result in a much larger current ripple due to the non-linear nature of LEDs, and current roughly equals light output. One of these drivers could be used to drive 2x panels in parallel. I would add a small amount of balancing resistance per panel, perhaps a 1R to equalise the current sharing between panels. The driver would be set to output around 700 - 750mA, best to be a little conservative on panel power.

Note the ending code on the meanwell models. They will often have an A, B, AB or another code. These denote the current output control / dimming function. Without any adjustment, the XLG 100 L will want to output 1A of current (500mA per panel )which would probably damage the LEDs. The "A" version has an integrated trimpot which you would use to set each driver to around the 700mA level. However this seems to be out of stock in most places. The AB model has a pair of wires to control the "3-in-1" dimming function. You can use a 0-10V DC signal, a PWM signal or use a potentiometer (probably the safer bet). The AB model seems to be fairly available at the moment.

0.75A at 120V equates to a 160R resistor, which you could use to set up the driver and test the current before connecting them to panels to avoid panel destroying mishaps.

I'd buy a single unit first just to test the concept, then buy another 10x @ $50.

It's a little tricky to find other suitable drivers with V/I characteristics, at a good price point, availability etc.

I also encourage you to check the datasheet for anything I might have missed.

[Automate007]

Yeah that sounds like my experience, you're trying to modify an AC/DC topology that inherently has a fair bit of ripple. The best solution that I can think of is to look at using a CC LED driver, and remove the existing drivers from the panels, ie convert them to passive panels.

Consider something like this:
https://au.mouser.com/ProductDetail/MEAN-WELL/XLG-100-L-AB?qs=sGAEpiMZZMvV8Y9YugmIggXPjVR9iwOZ2ARXrvaN0alPoJ9X1te3Xw%3D%3D

$55 in single qty and $50 in tens. The current output has 3% ripple, note that your voltage ripple will result in a much larger current ripple due to the non-linear nature of LEDs, and current roughly equals light output. One of these drivers could be used to drive 2x panels in parallel. I would add a small amount of balancing resistance per panel, perhaps a 1R to equalise the current sharing between panels. The driver would be set to output around 700 - 750mA, best to be a little conservative on panel power.

Note the ending code on the meanwell models. They will often have an A, B, AB or another code. These denote the current output control / dimming function. Without any adjustment, the XLG 100 L will want to output 1A of current (500mA per panel )which would probably damage the LEDs. The "A" version has an integrated trimpot which you would use to set each driver to around the 700mA level. However this seems to be out of stock in most places. The AB model has a pair of wires to control the "3-in-1" dimming function. You can use a 0-10V DC signal, a PWM signal or use a potentiometer (probably the safer bet). The AB model seems to be fairly available at the moment.

0.75A at 120V equates to a 160R resistor, which you could use to set up the driver and test the current before connecting them to panels to avoid panel destroying mishaps.

I'd buy a single unit first just to test the concept, then buy another 10x @ $50.

It's a little tricky to find other suitable drivers with V/I characteristics, at a good price point, availability etc.

I also encourage you to check the datasheet for anything I might have missed.

Thanks John that's a lot of info to take in but I agree I need to split the panel and drive it with something else. If I can find a driver that I can adjust the current to 350ma and Voltage to around 110vdc that would be perfect. Are there such things out there where both variables are adjustable

I won't be able to run cabling to 2 panels in parallel easily but am happy to get a driver for each light now.

1 thing that I'm very confused is constant voltage vs constant current drivers. Wouldn't dialing in the voltage (around 110v) to a fixed 350ma driver be the answer? Or it's a unicorn driver for my spec?

If I use the 700ma CC driver per panel could I put a resistor in series with the 1 panel , if yes I'm too dumb to figure it out what size and wattage resistor would do it?

[John B]

Most of the drivers have a combination of CC and CV control on the output. Meaning they will increase the output voltage until the maximum set current is reached through the load, or the max set  voltage is reached.

Connecting a 700mA constant current driver to a single panel would be a very bad idea. The linked driver could potentially output up to 142V, so it will raise the output voltage until it pushes 700mA through the single panel, probably destroying it in seconds. Adding a resistance isn't a solution since the driver will try to increase the output voltage to try and reach the 700mA target, it will only limited once it reaches the 142V max output voltage. But now the resistor will be dropping around 22V at 0.7A which gives you 15W dissipation in the resistor.

When you use a CC driver on 2 panels in parallel, the 700mA will be shared between the 2 panels, but the exact ratio is a little tricky since LEDs are non linear and the forward voltage lowers with an increase in temperature. Adding a small amount of resistance to each helps to match the ratio of currents, without dissipating too much power in the resistors. There is the potential to destroy a light if one goes open circuit, the full 700mA will be forced through the other light, destroying it. This is where an additional CV adjustment pot is ideal since you can set the max voltage to just above the voltage needed for CC operation. Alas the linked driver doesn't have it, only lower voltage ones do in the 24-54V range.

The driver could potentially work on a single panel if the dimming control was set to around 35% of its 1A output capability, but if you look at the datasheet, that is slightly outside the recommended I/V operating area. You'd need to ensure it starts in a failsafe way, ie the dimming control always works when the driver is on (like a simple potentiometer)

A 350mA driver for each panel would be ideal, but you need to find one that is capable of outputting up to 120V. I didn't see any suitable options, though something might be out there. Also even if you find something within those specs, they might have higher ripple anyways, bringing you back to square one.

You don't stand to lose much by buying a single test unit

[Automate007]

A 350mA driver for each panel would be ideal, but you need to find one that is capable of outputting up to 120V. I didn't see any suitable options, though something might be out there. Also even if you find something within those specs, they might have higher ripple anyways, bringing you back to square one.

John you've been amazing with your knowledge and help Thank you once again . OK I think if you can look at the spec of this it might be the unicorn I have been looking for .... It's 42W, Constant current setting of 350mA and 110V, I know it might be slightly dimmer than the stock 45W driver but I don't think a loss of 3 watts will make a difference.

Could you take a look at the spec and let me know if I'm getting excited for nothing or it's worth a shot. I couldn't see anywhere regarding ripple voltage. https://www.meanwellaustralia.com.au/products/lcm-40

[John B]

You said you measured 380mA at 120V DC, whereas the driver you linked has a maximum output of 100V when operating at 350mA. Open circuit voltage is 110V, so I would expect the panel to light extremely dimly or not at all. When I did a search on element14, mouser, and digikey I was looking for a max output voltage of >120V.

[Automate007]

You said you measured 380mA at 120V DC, whereas the driver you linked has a maximum output of 100V when operating at 350mA. Open circuit voltage is 110V, so I would expect the panel to light extremely dimly or not at all. When I did a search on element14, mouser, and digikey I was looking for a max output voltage of >120V.

Hmmm. I ordered 1 for testing. After I ordered I realised it has a 5% current ripple. So I guess that was a waste of money. Anyway curios to see what it does anyway.

I'm thinking maybe replace the lights with a high CRI >90 zero flicker free T8 tubes but hard to find at a decent price

[Automate007]

You said you measured 380mA at 120V DC, whereas the driver you linked has a maximum output of 100V when operating at 350mA. Open circuit voltage is 110V, so I would expect the panel to light extremely dimly or not at all. When I did a search on element14, mouser, and digikey I was looking for a max output voltage of >120V.

Hey John, You are 100% correct. it arrived late afternoon. I hooked it up this morning

I thought I'd sacrifice the warm LED's if it went wrong! Just like you said it was very dim (around 1Watt) the unit was setup for 350mA on the dipswitches.

Result output was a very dim light and 100Vdc and only 10mA getting to the LED'S. So I don't get it...... if its supposed to be a constant current why is it not 350mA at 100Vdc?

I have to go away next week for work but in the meantime I found this one and am pulling all my strings to be able to get one to test with when I get back and the specs say </= 3% ripple current which seems to be the best I have seen so far.

https://www.tci.it/prodotti/alimentatori-led/968/mp-50-350-slim/

[Automate007]

Datasheet

[John B]

Result output was a very dim light and 100Vdc and only 10mA getting to the LED'S. So I don't get it...... if its supposed to be a constant current why is it not 350mA at 100Vdc?

Most of the drivers have a combination of CC and CV control on the output. Meaning they will increase the output voltage until the maximum set current is reached through the load, or the max set  voltage is reached.

A real constant current device can only work within the constraints given by the load. You can test this idea with a couple of resistors. If you were to attach a 10R resistor to the driver, the theoretical figures would be given by ohms law: IR=V, or 0.35A * 10R = 3.5V. This is within the specified range of output voltages at 350mA, which is 2 - 100V. Also note that the resistor here is only dissipating around 1.2W and so isn't stressing the driver either.

On the other hand if you were to attach a 10 Megaohm resistor, an unlimited CC device would theoretically output 3.5 million volts to put 350mA through the resistor. Obviously this is not a realistic device. The driver in practice will only be able to increase the output voltage to around ~110V max as per the datasheet.

Likewise your panels have a conduction characteristic where they will only pass significant current once it reaches around the 120V mark. The conduction is gradual though, as you've found they will pass a few mA under the nominal operating voltage.

I would still stay with my recommendation of paralleling the panels in groups of 2. This is so that you only need to buy 10 drivers and not sink too much money into this system. Those LCM drivers still aren't any cheaper.

[Automate007]

I understand you now. Makes total sense. The LED haven't overcome their forward voltage and therefore not sucking the max current yet till the voltage goes higher in which it can't get over the 100V!

I have a 40V 40W very good panel I tested a while back with CRI over 90. Testing with the driver it came with and flickers so given it was twice the price at $160 each I gave up on them . I figured I can try that meanwell driver I just got on it and it's freaking perfect. A 1V ripple and even with slow motion there is no flicker. But will mean around $160 per light plus the $75 meanwell driver I just tested it with
 and a load more holes in the gyprock to mount it but at least I know I have a way out of this mess ..... just a very expensive fix If that TCI 50W driver that I get next week doesn't work also he told me BTW around $30 each (hopefully he was not pulling figures out of the air) as they buy bulk but he wasn't sure if they had many left till next shipment and needs the warehouse manager to find them next week. So for now I wait!

[Automate007]

Result output was a very dim light and 100Vdc and only 10mA getting to the LED'S. So I don't get it...... if its supposed to be a constant current why is it not 350mA at 100Vdc?

I got bored and decided I would sacrifice 1 panel in experimenting. I bridged 1 led and 8 in a group turned off but the rest lit up drawing 100ma. I then bridged another in the next group and went up even more. So I kept bridging groups till I got to 350ma and tge brightness was on slightly less than a fully lit panel using a light meter. Zero Flicker and driving with the meanwell LCM-40.

It's a cheap and nasty way and so far the easiest solution if all else fails!

[Automate007]

So gone from 320leds to 272. 15% loss. So from 45W to 38W. I can live with that considering I have 8 of these in a 25mtr sq room! I really hope the other driver from TCI works as I feel bad screwing with LED panels like this! Also please ignore the quick n dirty solder.... I Know I know I'm cleaning it all up now once my vision comes back from staring at these LED'S head on

[Automate007]

There's one last option.....

The last option of re working the light and adding that 50W TCI LED Driver turned out to be the best option. Worked perfectly and I have completed the conversion of 8 lights in 1 room. Zero flicker now. Thanks mate for all your help getting me through this hurdle. I tried to make it as clean and safe as possible . Even added labels everywhere on the light under the covers indicating 120VDC so at least in the future if anyone buys this premises can clearly see its not 240Vac and being driven by a LED driver

[Automate007]

Here are some pics of what I did to modify and label

[Automate007]

Some more