LED Driver Circut

Looking at PC817 datasheet it has a minimum current transfer ratio of 50%. From the scheme I see that with 0V on the mosfet gate the collector current of the transistor (PC817output) is about 9mA -> this means that the diode needs 18mA on worst case to be able to handle such output current! With 220Ohm series resistor however you're polarizing it with only 2mA, it has no chance to handle all the output current.

If you keep your 100k resistor and swap the 1k with 10k resistor the current that the optocoupler has to handle drops to about 1mA and should work (given that all other components are fine and not damaged by ESD as pointed out by others )

Beware that swapping 1k with 10k will slow down turn on transient, but I don't think it will be an issue...

Looking at PC817 datasheet it has a minimum current transfer ratio of 50%. From the scheme I see that with 0V on the mosfet gate the collector current of the transistor (PC817output) is about 9mA -> this means that the diode needs 18mA on worst case to be able to handle such output current! With 220Ohm series resistor however you're polarizing it with only 2mA, it has no chance to handle all the output current.

If you keep your 100k resistor and swap the 1k with 10k resistor the current that the optocoupler has to handle drops to about 1mA and should work (given that all other components are fine and not damaged by ESD as pointed out by others )

Beware that swapping 1k with 10k will slow down turn on transient, but I don't think it will be an issue...

thx. ill report back

I think your calculation has an error - there should be just over 8 mA through the optocoupler LED, not 2 mA. (3.3 - 1.4) / 220  = 8.6 mA.

So it's nowhere near as bad as you mention.

8 mA is a bit of a compromise, because it depends on how many connections there are on the Pi, since there's an overall limit. However, I guesstimated that 8 mA should be harmless. 

It's not a production circuit, I don't think it needs to be guaranteed to function for the worst case conditions, since CTR of 50% is very untypical.  

Having said that, for sure the component values can be adjusted. But I don't think from the symptom, that this is the issue. But it can be known 100% by simply shorting the 100k resistor. 

Oh, you're right, I messed up the calculation. My bad, my statement is wrong! With 8mA of led drive it should work indeed 

I put the original circuit on a breadboard and it works using pwm commands. MOSFETS are fine.. No leakage. (no sparksl). I'm thinking I should use a bigger perforated board to separate the components further away.  the boards are such crap, any suggestions on that.

Separate question. I will be using a start up service to activate the Py led driver,  I will use "try", "while true", and "except KeyboardInterrupt" to put my driver in a loop. there will be no keyboard connected.. will that be an issue? also can i add an "if" statement that will acknowledged input from a button? 

Read the comments posted by Scottiebabe about a possible wiring error and inconsistency between schematic and protoboard wiring.

Perfboard is about as decent a prototyping solution as there is, before moving to a full printed circuit board. 

Having said that, there is some slightly nicer perfboard than the low-cost Amazon ones. I like this one: https://www.newark.com/roth-elektronik/re200-c3/prototyping-brd-eurocard-100-2x160/dp/05M1075?ost=re200-c3 it's a decent price, and  is a large size (160 x 100mm) easy to break into a smaller piece if required, and the pads are not too large, which is occasionally helpful, plus quick to desolder and re-work, since the pads are not through-hole plated (the green board in your photo is through-hole plated which is inherently more robust, but non-through hole boards can be reliable too).

A custom PCB is low-cost, and would be worth trying, although there's a learning curve (but not much; a week of tutorials/practice for a couple of hours each day would be enough to learn say KiCad https://www.kicad.org/ ; I can recommend a video if you wish to go that route).

Just remembered I did a project with a Pi involving 24V lights in the past too, using perfboard, it is documented here:  Raspberry Pi enabled Christmas Lights   .

However mine was lower power (about 2.8W per chain of lights, and there were 4 chains), and it used lamps instead of LEDs. It definitely helps using a larger board. This was 85 x 70 mm approx:

Underside, using thin (30 AWG) bare wire, but it's possible to use solderable enamel wire etc. I used the perfboard mentioned above.

However now PCBs are so cheap, if I were to do it again, I'd make a custom board.

lol in that photographed circuit above when the opto is switched off,  the FET's gate voltage is determined by in-circuit leakage currents...

i caught the mistake two nights ago when i was checking the Msft for connection problems to the 100k. i spent some time remaking a new circuit with care to insulate everything, even after cleaning the board with rubbing alcohol, but when i was finished, it still didn't operate correctly. my continuity was off all over the board. i had pin holes with no surrounding components giving charge. 

looks good, the boards i have now are traceable on both sides. My plan was to connect all three and run grounds together and power together. I ruined my week by losing my galvanic isolation