You are looking to design for 650W power. It won't be trivial to divert that amount. Do you have a suitable load to handle that? Bear in mind the likely unregulated output of the inverter, such as shown in the second sim. It ain't a pretty sight. Also, it results in current spikes several times the amplitude of the un-modulated /un-chopped supply. Whether the internal semiconductors will cope with that is debatable.

Fair enough. Thanks for the insight.

 

Chiefrunningphist, looking a little more into this I found someone using your COB LEDs - http://ledgardener.com/samsung-lm561c-build-getting-started/

From there I referenced the model of Meanwell he is using and that led me to - https://www.ledsupply.com/led-drivers/mean-well-hlg-c

And on that page I found this, "Still unsure of what Mean Power Source is right for you? Take a look at our Mean Well Power Guide which explains the many variations and styles of LED power supplies. Or, if you have you prefer we are always available via phone or email to answer your tecnical questions."

So instead of taking advice from many people on a forum, maybe go to the source?

 

What current? SSRs with a SPDT function are available (e.g. here or here), but generally for current less than 200mA.

 

There's a disconect somewhere in my explanation,

It's not a new thing in this thread.

 

@shortbus @Alec_t

Like this..
View attachment 171217

Not fast switching, spend at least 15min per LED array, could be hours at a time. What would be the best component combo that you'd recommend to achieve such a task?

So then, from that drawing you are just wanting to switch from panel "A" to panel "B" from time to time?

 

What is the supply voltage of your MCU?

 

What is the supply voltage of your MCU?

I was going to try to use the ESP32 which is 3.3v I believe. Figure a max of 3A on the load cct just to be safe.

 

Then isn't a "relay shield" what you need? Just one of the many out there - https://www.seeedstudio.com/Relay-Shield-v3-0-p-2440.html

Can I ask why you didn't post that drawing earlier? Like 3 pages ago? I thought you were after something totally different.

 

Then isn't a "relay shield" what you need? Just one of the many out there - https://www.seeedstudio.com/Relay-Shield-v3-0-p-2440.html

Can I ask why you didn't post that drawing earlier? Like 3 pages ago? I thought you were after something totally different.

Lol well I was after something totally different, but decided to squeeze all the utility from this thread with a quick question in a change of direction. Thanks, I'll look into what a relay shield is.

 

Here's one option for switching between two LED strings. Should work whether the MCU has a 3.3V or 5V supply.

The FETs must be 'logic-level' types, such as the IRL 530 or IRL540, rated for the highest CC driver voltage and able to handle the LED current comfortably. Zener diode D3 keeps the gate voltage of M2 within the rated range. The small current through D3 is unlikely to cause unwanted illumination of diode string 1, but could be reduced by increasing R2.

 

Here's one option for switching between two LED strings. Should work whether the MCU has a 3.3V or 5V supply.
View attachment 171338
The FETs must be 'logic-level' types, such as the IRL 530 or IRL540, rated for the highest CC driver voltage and able to handle the LED current comfortably. Zener diode D3 keeps the gate voltage of M2 within the rated range. The small current through D3 is unlikely to cause unwanted illumination of diode string 1, but could be reduced by increasing R2.

You da man!

 

Here's one option for switching between two LED strings. Should work whether the MCU has a 3.3V or 5V supply.
View attachment 171338
The FETs must be 'logic-level' types, such as the IRL 530 or IRL540, rated for the highest CC driver voltage and able to handle the LED current comfortably. Zener diode D3 keeps the gate voltage of M2 within the rated range. The small current through D3 is unlikely to cause unwanted illumination of diode string 1, but could be reduced by increasing R2.

R2 reduces the current flow enough to redirect current to a greater path of least resistance of nMOS 2 and diode string 2 when nMOS1 is closed? I shouldn't have to worry about the load ohms of a max 24w array redirecting current to diode string 1 because theres no way the load ohms would be greater than 1k?

Reference example...
285 w LED array @ 108v & 2.64A
V=iR
108/2.64= 41ohms

So a similar LED array but only 24w should be nowhere near 1000ohms, right? That's why its able to switch between the 2 LED strings? The higher the LED resistance (ohms) the more leakage to string 1 will be noticeable? To be greater than 1000 ohms, the voltage has to be 1000× greater than amperage, right?

 

Don't think of LED strings as resistors; they aren't. LEDs are very non-linear devices. Just consider their total forward voltage drop and the current they are passing. For the 285W array example, 108V is the total forward voltage drop at 2.64A.
The circuit I posted is merely a switching arrangement. When the MCU output is high (say 3V), M1 turns on and its drain voltage drops to near zero, hence so does the gate voltage of M2, turning M2 off. Thus LED string 1 is on and LED string 2 is off. When the MCU output is low, M1 turns off, a few mA of current through LED string 1 and R2/R3/D3 raises the gate voltage of M2 to a value (~10V) limited by the Zener diode, so M2 turns on and so does LED string 2.
For the 285W LED string example, simulation shows that the current through the LED string and R2, with R2=1k, would be ~9mA when that LED string is supposedly turned off. This might well cause enough spurious light to be noticeable. Increasing R2 to 22k brings that current down to ~1mA.
Edit: If the two LED strings have different total forward voltage drops, the one with the higher voltage drop should be string 2, to ensure the gate voltage of M2 is adequate.

 

The voltage across a resistor is always strictly proportional to the current through it (Ohm's Law). This is not the case with LEDs, which have a distinct kink in their voltage-current characteristic. I have never studied the ins and outs of the way LEDs work, except to understand that light output and forward voltage drop increase non-linearly with current, so cannot comment on your theory of operation. The datasheets for LEDs should tell you how light output is likely to vary with current and/or forward voltage.
If all your LEDs are driven with a controlled current well below their rated maximum (which is how they should be driven for the sake of a long life) then there won't be thermal runaway.

 

Are you saying that due to LEDs different electrical properties that I can have 20 parallel strings, and if some strings have different ratios of components (still 20 components per string though) that the board power distribution, or the current supplied will split evenly throughout all 20 strings?

No. I'm saying that if you have two dissimilar strings, with different voltage drops but each requiring the same say 150mA, then the switching circuit I posted can be used to drive them alternately using a 150mA constant current source. If the two strings were connected in parallel it is most unlikely that they would pass the same current.

Could different string current ratings (if some strings had more high current rated chips than others, or more short WV chips than others) be an issue?

Yes, if you use a constant current source with a current set greater than the maximum rated current of the least powerful LEDs.

I thought I'd have to add resistors to the lower voltage drop strings so they didn't get more current than intended.

That's the solution if the strings are actually in parallel.

 

I'm trying to use one driver but split it into different LED arrays and have the LED arrays be independently dimmed.

That effectively creates two drivers. I don't see any advantage over just buying two drivers to start with.

 

Can I design a more efficient PS than 93.5%

I very much doubt it. The high frequency transformer used in a typical SMPS is a critical component and might prove difficult to source. Building one is something of a dark art.

 

I very much doubt it. The high frequency transformer used in a typical SMPS is a critical component and might prove difficult to source. Building one is something of a dark art.

Ahhh interesting...

Would this cct create better overall efficiency compared to using a seperate driver? Would it even work?

*I've been informed I need resistance between the zener and + source to achieve the zener properties desired...

 

Would this cct create better overall efficiency compared to using a seperate driver? Would it even work?

Probably not and probably not. If you siphon off, say, 100mA from a 500mA constant current supply, I don't see how you would get controllable dimming for one array independently of the other array.

 

If anyone has any other direction I should be looking blurt it out!

I hope you haven't already bought the devices to make this. You have changed your mind many times since the thread started. If this is a real project, shouldn't the parameters be drawn up first, then find what is needed to get the job done? Even a simple "block" diagram. Your way of approaching it seems to be backwards. Not trying to put you down at all just pointing out how it looks to some one not involved and trying to help.