Need help validating this HiPower LED COB solution from a safety and heat perspective

Thread Starter

vhaz

Joined Nov 15, 2022
7
Dear all,
Winter is coming and I'm building my own daylight-therapy light/lamp,.
With only basic electrical theoretical knowledge, I have some practical concerns I need help validating. I will not run this light for more that 45 minutes at a time.

Construction (see attached picture):
I will run four 'slings' of 11 36V LED COBS in series, making a total of almost 400V on 175 mA constant current per 'sling'.
Totaling a constant current of 4x175mA = 700 mA, which is powered with a MeanWell HLG-320H-C700A.
The LED COBS are Cree CMU1006-0000-000N0H0A65G (44 of them in total).

The LED COBS will be glued with heat conducting glue (1 W/mK) to a 600mm X 800mm 2mm thick aluminum plate/sheet, which serves as the lamps structural back pane.
The interconnections will be with 18 AWG (0.75 mm2) wire, soldered to the LED COBS.

What I would like to get validated by you guys:

1. High voltage concerns
A. Is is ridiculously dangerous to have 400V inside? (even with low current of max 700 mA). There will be acrylic front cover and wooden sides, so no accidental access to the interior will be possible.

B. The LED COBS soldering points are just a mm above the large shared aluminum sheet. Is there any risk of short circuit arc/sparking?
Here I do not understand if I have a 400V 'arc/spark'-risk situation or if the voltage potential is more like 36V (as the individual LED COB).

C. I will add on/off button and a Fuse Holder+1 A fuse. Do these components need to specified for 400V, or is my low current allowing me to use almost any 12V/car components?

2. Heat management concerns
A. Will the large aluminum plate be a sufficient heat sink for a total of 300W LEDs?

3. General electric circuit and LED operation
A. I hope my power assumptions (constant current and the 400 V range), parallels and series solution is correct.

Would love to hear you experts comments on this.
Thank you very much from a pitch dark Sweden.
 

Attachments

MisterBill2

Joined Jan 23, 2018
18,176
I suggest re-arranging the circuit to have ten series strings with a max voltage of 40 volts. This will be much safer as far as shock haardsThe power should be about the same, but the total current will be ten times greater.
 

MisterBill2

Joined Jan 23, 2018
18,176
If they are used all in series then the current does not add, only the voltages add. So the series current will be 175 milliamps, probably a much cheaper power supply. But still 400 volts DC is a shock hazard
However, since they are in sets that run on 36 volts, use that voltage and wiring will be simpler AND more reliable because of being redundant. If one section fails the other ten keep lighting.
36 volts is not a shock hazard, and three 12 volt supplies in series will work very well. or two eighteen volt supplies. The current would be about 11 x 0.175 amps, almost 2 amps. So you will have a safer array voltage and probably get away with a cheaper power supply.
Your heat sink will need to dissipate about 75 watts of heat without much temperature rise and so that will be some work.
A 24 volt transformer with a bridge rectifier and a fairly large filter capacitor will give close to 36 volts, and with a bit of experimenting you may not need a current regulator, if the supply has poor enough regulation. That will be the most efficient and cheapest scheme.
 
Last edited:

Thread Starter

vhaz

Joined Nov 15, 2022
7
Thank you, appreciate your advice.
Rearranging to a significantly lower voltage would mean almost 20-40 parallel slings using very much wiring. Also my Meanwell driver would need to be replaced. I hope I can use the actual design, making good soldering on the COBs that would not get in contact with the aluminum base plate.

Thinking about components a fuse holder and on/off switch, as there is no potential/voltage drop over them, I assume the can go by any voltage classification (i.g. 12V stuff, as long as the current is allowed)?
Guess then its only the DC-plug from the Driver to the lamp that needs to be spec'ed for 400V..?
 

MisterBill2

Joined Jan 23, 2018
18,176
OK, reading the first post a third time I see that there are strings of 11 of the chip on board LEDs, each powered rated for 36 volts. Originally I was thinking that each string of 11 devices used 36 volts, hence my suggestions about lower voltages. So instead, each string of 11 of the 36 volt LEDs would be powered by 396 volts. If the supply is adjustable you will be able to set it to a slightly lower voltage and gain much longer LED life times.
There is a possible safety issue because each assembly MUST HAVE A HEAT SINK, and those heat sinks may possibly be connected to one side of the supply voltage. So there might be a shock hazard from the heat sinks. But that is unknown presently.
 

Sensacell

Joined Jun 19, 2012
3,432
Just use the series strings as you planned. The COB strings in parallel might not share current well, especially if they are not at the same temperature.
This might require some balast resistance inserted in series with each string to balance the current.
In this context, the 400V is not a big problem, the circuit is never broken under power, and the output should be floating.
Just pay careful attention to insulating the connections.

GROUND THE CASE.
 

MisterBill2

Joined Jan 23, 2018
18,176
Once again I will remind that these COB devices must have an adequate heat sink properly attached. And also very important, verify that neither connection to the COB is connected to the heat sink.
 

Thread Starter

vhaz

Joined Nov 15, 2022
7
OK, reading the first post a third time I see that there are strings of 11 of the chip on board LEDs, each powered rated for 36 volts. Originally I was thinking that each string of 11 devices used 36 volts, hence my suggestions about lower voltages. So instead, each string of 11 of the 36 volt LEDs would be powered by 396 volts. If the supply is adjustable you will be able to set it to a slightly lower voltage and gain much longer LED life times.
There is a possible safety issue because each assembly MUST HAVE A HEAT SINK, and those heat sinks may possibly be connected to one side of the supply voltage. So there might be a shock hazard from the heat sinks. But that is unknown presently.
Great, thank you.
The COBs heatsink is not electrically connected, so there should be no voltage potential in the base plate chassis.
 

Thread Starter

vhaz

Joined Nov 15, 2022
7
Once again I will remind that these COB devices must have an adequate heat sink properly attached. And also very important, verify that neither connection to the COB is connected to the heat sink.
Yes, my aluminum chassis is the heatsink, initial calculations indicate its area will be able to dissipate at least 50% of the heat. I will measure temp an if necessary add additional heatsinks/fins to the backside of the chassis to extend the surface area.
 

MisterBill2

Joined Jan 23, 2018
18,176
I am aware that there is no intentional connection through the COB assembly. But I have no knowledge of the TS soldering skill level, and in addition I have seen quite a few terrible soldering effort results over the years. And as 400 volts DC is a hazardous voltage, it is appropriate to remind about checking for unintended connections.
 

Thread Starter

vhaz

Joined Nov 15, 2022
7
Just use the series strings as you planned. The COB strings in parallel might not share current well, especially if they are not at the same temperature.
This might require some balast resistance inserted in series with each string to balance the current.
In this context, the 400V is not a big problem, the circuit is never broken under power, and the output should be floating.
Just pay careful attention to insulating the connections.

GROUND THE CASE.
Great news, thanks for securing me about the 400V.

Resistor to protect from over current is a great idea. My calculations say a 2,2K resistor in each string would be right.
175mA * 2200 ohm = 385 V. (typical Vf of each LED is actually 34V @ 85°C, 11* 34 = 385 V).
Is my calculation right here?

How on earth (pun intended) do I ground the case?
My external driver only have to leads (+,-) as output.
Should it be grounded to my residential 230V grid, or the external drivers metal chassis (which is the same, actually).
 

Sensacell

Joined Jun 19, 2012
3,432
Great news, thanks for securing me about the 400V.

Resistor to protect from over current is a great idea. My calculations say a 2,2K resistor in each string would be right.
175mA * 2200 ohm = 385 V. (typical Vf of each LED is actually 34V @ 85°C, 11* 34 = 385 V).
Is my calculation right here?

How on earth (pun intended) do I ground the case?
My external driver only have to leads (+,-) as output.
Should it be grounded to my residential 230V grid, or the external drivers metal chassis (which is the same, actually).
No, totally wrong idea. It has nothing to do with "limiting current".

The resistor should be chosen to drop only a few volts at the design current, it's there to allow slight differences in the COB Vf from causing current hogging.
A guess would be ~ 12 ohms, this will drop about 2.1V volts at 175 mA.


Make NO connections between your driver and Chassis Gnd.
Bolt the Driver to the metal.
Make a solid connection between the metal chassis and the external earth ground. (use 3 wire cord)

These drivers provide galvanic isolation between the mains input and the output, it's floating.
Even if there was a direct short between the output and the chassis, not much would happen.

A fuse and proper earthing of the metallic chassis is still mandatory.
 
Last edited:

Thread Starter

vhaz

Joined Nov 15, 2022
7
No, totally wrong idea. It has nothing to do with "limiting current".

The resistor should be chosen to drop only a few volts at the design current, it's there to allow slight differences in the COB Vf from causing current hogging.
A guess would be ~ 12 ohms, this will drop about 2.1V volts at 175 mA.


Make NO connections between your driver and Chassis Gnd.
Bolt the Driver to the metal.
Make a solid connection between the metal chassis and the external earth ground. (use 3 wire cord)
Thanks for the resistor guidance!

I do not fully follow the idea of grounding.
I have planned to have my external MeanWell driver away from the lamp (on the floor below), as its quite bulky. Hence not bolted to the back plate. Probably need to reconsider this, to bolt it on to the back plate. Guess there are few decent alternative ways having it apart from the lamp?

Edit:
Make a solid connection between the metal chassis and the external earth ground. (use 3 wire cord)
As the external driver chassis is externally grounded (by 3-wire), I should be able to have an earth-lead from the drivers chassis to my backplate chassis, if I keep it apart on the floor. Correct or not?
 
Last edited:

Sensacell

Joined Jun 19, 2012
3,432
Having the driver external makes it sketchy.

If it's got a connector, nasty arcing will happen if you pull it under power.
The wiring now has to be all rated for 400 V.

I think that makes for too many problems.
 

Thread Starter

vhaz

Joined Nov 15, 2022
7
Having the driver external makes it sketchy.
The wiring now has to be all rated for 400 V.
Agree it would be sketchy.
So the wiring need to be rated for 400V now also with the driver inside? All small wires between each LED also?

Thank you for your valuable help!
 
Top