How do I drive LEDs without flicker ?

[OM222O]

Mechanical design is mostly finished:
https://a360.co/3BbbbPL



you just need to finalize the dimensions of the angle pieces and the bars on top.
It might be a pain in the ass to drill that many holes, especially in the heat sinks (you can ignore the unused holes on the left and right-hand sides. They're there because I wanted to copy the bar 5 times).
You also need to tap the heat sink holes which is not ideal. Are you sure you can't find a larger heat sink?

I'm not sure how efficient the LEDs are but tbh this seems a bit overkill. If you can measure their maximum heat output, you may find that using a 400x400mm aluminum plate is enough. If it wasn't enough, maybe adding a few heat sinks to the plate could be enough (I can't imagine you'd need more than 4 heat sinks total).

Edit: I just realized I'm an idiot ... it would be impossible to put the glass in with this design. But it should be fixed now (I added 2 more bars that can be screwed to the frame with some L brackets. The 5 support bars can then be welded onto those 2). I still think you should reconsider using 16 heat sinks though.


What I mean is that instead of having the 5 supporting bars, you can have a 400x400mm aluminum plate (around 3mm thick) as a heat spreader. If this surface area wasn't enough, then you can use thermal adhesive to glue heat sinks to the plate. If you can figure out the efficiency of the LEDs or measure how many watts of heat each one outputs at 250mA, then you can even simulate it without spending any money.

[DeadlyChemist]

You also need to tap the heat sink holes which is not ideal. Are you sure you can't find a larger heat sink?

i can't or they get ridiculessly expensive, plate idea below

What I mean is that instead of having the 5 supporting bars, you can have a 400x400mm aluminum plate (around 3mm thick) as a heat spreader. If this surface area wasn't enough, then you can use thermal adhesive to glue heat sinks to the plate. If you can figure out the efficiency of the LEDs or measure how many watts of heat each one outputs at 250mA, then you can even simulate it without spending any money.

basically the math i did was:
the max power cool white can use would be 16*7*0.85 basically 16 leds at 7W at 85% max power (255mA instead of 300mA) about 95.2W
for warm white it's also 95.2W
rbgwyu would be 6*18W (worst case, everything turned on)
so total is about 300W, assuming 0% efficiency of the led, that's 300W i need to get rid of
the heatsinks are 2K/W
using 16 of them, that's 0.125K/W at 300W that's about delta 37.5K
so the heatsink would rise by about 37.5C with EVERYTHING turned on with 0% efficiency

leds have actual efficiency of about 30%, 30% is light, 70% is heat
so that's about delta 26.25C with a room temp of 25C thats a chilly 50C at 300W i'll probably never use lol

my goal is under 60C at about 200W (140W if you count 30% efficiency)
so 3x3 heatsink grid could defiently work

I like the 400x400 heat spreader with few heatsinks at the back, but i can't really do any math on that, how much can that handle? on ebay 400x400x3 aluminium plate is about 25€ plus like 5-10€ for shipping
way cheaper than 120€ + work for a heatsink, but i have no clue how much it can handle.
I defiently prefer a big plate, but need some help regarding thermal capability


as for taping the holes, that's fine
I'm thinking to go the "smart" way and mount the leds on a small carrier pcb (alumiunium) which gets screwed into the heatsink with some thermal paste inbetween (thermal glue between the led and board)
this is worse thermal contact than direct led on heatsink, but the cleanup/replacement is way easier
im unsure, might go lazy and just mount them without carrier board.

[OM222O]

I found this: https://www.heatsinkcalculator.com/flat-plate-heat-sink-calculator.html
A single large plate won't be enough as I guessed (doing the math for 1 LED: 100x100x3m flat plate gives a thermal resistance of  7.84°C/W which at 10 watts and 25c ambient is around 100c).

I suspect adding 4 or 6 heat sinks to the plate should keep it below 60c, but it's hard to tell given how much of it is guesswork (thermal paste thickness, conductivity, etc.)

Do you really need a 300-watt LED panel? that seems a bit too much for any indoor application.
If you wanna be sure, go with the 16 heat sinks, but it's gonna be painful (and you'll need a drill press, calipers, and taps which I assume you have access to).
You may also want to reconsider using fans because there is too much heat in a small area.

Edit: I'm gonna do a proper simulation in fusion later when I have time.
Can you take some pictures of the COB module from the top, side, and bottom next to a ruler? the aliexpress listing shows different packages which is confusing.

[DeadlyChemist]

I found this: https://www.heatsinkcalculator.com/flat-plate-heat-sink-calculator.html
A single large plate won't be enough as I guessed (doing the math for 1 LED: 100x100x3m flat plate gives a thermal resistance of  7.84°C/W which at 10 watts and 25c ambient is around 100c).

I suspect adding 4 or 6 heat sinks to the plate should keep it below 60c, but it's hard to tell given how much of it is guesswork (thermal paste thickness, conductivity, etc.)

yeah no, 6 heatsinks are about 6*7=42€ + 5€ shipping plus 25€ plate + 10€ shipping
thats 77€
i'd rather pay 16*7€ = 112€ + free shipping for 16 heatsinks

Do you really need a 300-watt LED panel? that seems a bit too much for any indoor application.
If you wanna be sure, go with the 16 heat sinks, but it's gonna be painful (and you'll need a drill press, calipers, and taps which I assume you have access to).
You may also want to reconsider using fans because there is too much heat in a small area.

300W is simply ridiculess, agreed, but could be very useful for slow motion
but it's technically 100W warm 100W cold, 100W rgbwyu

as for heatsinks, i intended to give my friend just the 16 heatsinks and tell him to have fun welding it all up lol + would have a big solid heatsink
but the connectors/straps sound also good, i don't mind doing it myself

as for fans, no, to much space (simply thiccness) and I hate any noise (even the barely audiable one)

Edit: I'm gonna do a proper simulation in fusion later when I have time.
Can you take some pictures of the COB module from the top, side, and bottom next to a ruler? the aliexpress listing shows different packages which is confusing.

Thanks, and I'll next week once i unpacked some of my stuff from boxes

[OM222O]

I ran a simulation using the plate and heat sinks with convection and emission:


I really doubt this is correct though ... I have to investigate a bit more (need a more detailed model of the LEDs; right now they are 13mm aluminum cylinders) and remove the heat sinks to see if the temperature goes up or if I messed something up.

Again, just based on pure guess I'd say a large plate and 4 heat sinks should be enough cooling and much easier than stripping paint from heat sinks and welding them together. I'll post another update when I make sure the simulation results are correct

Edit: I figured out what I did wrong. For heat load, I used 10 watts (14 watts LED at 30% efficiency) and selected all the LEDs thinking it applies 10 watts per LED, but fusion meant total power. After increasing it to 160 watts (10*16) this is what I get:


This is assuming perfect transfer between the LEDs and plate to heat sinks (no thermal interface material / assume the entire thing is machined from a solid piece)

4 heat sinks were definitely a no-go. Funny enough increasing them to 9 didn't improve things by a huge margin:



The inner LEDs are just on the edge but the outermost LEDs seem to be fine. Now if we add the thermal paste inefficiencies back into the equation, even 9 heat sinks won't be enough.


Edit 2: I have some bad news for you: after playing around with the simulation and using BOTH the heat spreader plate AND 16 heat sinks, the LEDs are at 50c with an ambient of 25c (25c delta). Again this is assuming no thermal paste (perfect transfer) so in reality, they'll be right around the 60C mark. Unless you want to run the LEDs at a lower power level, this design is very impractical on its own (ignoring RGB).  If you want, you can send me your fusion 360 account so I can add you to the project to play around with the model yourself (everything is parametric and easy to change).




I'm not sure how they measured the 2K/W (maybe with a decent amount of airflow?), but 16 heat sinks will definitely not be enough on their own. Increasing the convection value from 10 to 100 (aka using a fan) does improve things a lot but you said you don't want to do that.





As a side note: heat sinks are extruded aluminum so you may not be able to easily find them wider than 100mm, but I'm sure you should be able to get longer ones (for example 100x500mm). Try contacting the seller to see if they can send you longer pieces (without cutting them down to 100mm).


Edit 3: using the spreader plate and 4 heat sinks + airflow (fans) results in a max temp of 46c which is only 10c warmer than using 16 heat sinks! Increasing the plate thickness from 3 to 5 drops temps by another 3 or 4 degrees! This should tell you that the heat sinks you chose are a very bad choice for passive cooling and when using fans, just 4 should be enough.




As you can see the innermost LEDs get a bit hotter than the outermost LEDs so you might want to run them at a bit lower power which is easy as you have full control over each individual LED from the driver board (it's a good idea to actually check the temperatures using a thermocouple after you build the whole thing).
 You can also watch this video :

where they add a "massive" (compared to what's already in the steam deck) heat sink, but after about 30 mins, it reaches the same temperature as it did without the heat sink! but adding a small fan makes a huge difference in the end so they decide to keep the fan on there.

[DeadlyChemist]

which is easy as you have full control over each individual LED from the driver board (it's a good idea to actually check the temperatures using a thermocouple after you build the whole thing).

my leds will be pairs (1 pot for 2 leds (2x cool or 2x warm)

where they add a "massive" (compared to what's already in the steam deck) heat sink, but after about 30 mins, it reaches the same temperature as it did without the heat sink! but adding a small fan makes a huge difference in the end so they decide to keep the fan on there.

I know that, but heatsink should radiate some power passively, the heatpipe just redirects the heat

I'm not sure how they measured the 2K/W (maybe with a decent amount of airflow?), but 16 heat sinks will definitely not be enough on their own. Increasing the convection value from 10 to 100 (aka using a fan) does improve things a lot but you said you don't want to do that.

i assume with fan it's gonna be way less than 2C per watt
and yes, i don't want fans


here are the measurments:




Im considering adding a tiny tiny fan and use software to ramp it up barely only at like 50-60C
but i simply don't want it, space/cost/moving parts is another factor

i'll look for 400x100 aluminium heatsinks or something
but for now im not sure, not sure how much 2K/W is lies or not

EDIT: found these as well
https://www.conrad.de/de/p/fischer-elektronik-sk-42-100-sa-strangkuehlkoerper-1-2-k-w-l-x-b-x-h-200-x-100-x-25-mm-188026.html
basically 200 x 100 x 25 mm
so would need 8 for 400x400
they are 1.2 K/W instead of theoretical 1.0K/W
lookming for alternatives, don't like the fan idea

[OM222O]

i assume with fan it's gonna be way less than 2C per watt
and yes, i don't want fans

Trust me, manufacturers always advertise their best possible number, so it won't be any better than 2C/W,  besides the design does not seem like a passive one. I mean just have a look at these passive heat sinks for CPUs:



They have much thicker, longer, and spaced-out fins.

Im considering adding a tiny tiny fan and use software to ramp it up barely only at like 50-60C
but i simply don't want it, space/cost/moving parts is another factor

i'll look for 400x100 aluminium heatsinks or something
but for now im not sure, not sure how much 2K/W is lies or not

Nope, that's a terrible idea: smaller fans have to spin much faster to move the same amount of air, so they consume more power and produce A LOT more noise. If you want to add fans, go with 140mm from either Noctua or be quiet (I think they're called silent wings). Adding temperature measurements to that many LEDs would be a challenge on its own, so you should base your fan curve on power draw. Obviously, you can test the temps to optimize the fan curve for noise while keeping things below 60C (if you're measuring the temperature of the heat sink or spreader plate, you should aim for a max of 50C).

2C/W is not a lie, but I'm fairly certain they came up with that using a ton of airflow.

The new heat sinks are wider, not longer (wrong direction (I don't recommend using 200mm fans)).

Again, from all the testing I did, a 3mm plate, 4 heat sinks, and 4 140mm fans seemed like the best choice and simplest construction (and cheap), but the final decision is up to you.
You only need the fans to come on at higher powers(I will try to find the max power level with 4 passive heat sinks) and either Noctua or be quiet fans are optimized for noise, so you should be fine.

[OM222O]

I ran the numbers again using the most optimal layout I could come up with from previous simulations and at full power and without fans, the max temp was 63.5C (add about 10C to all the numbers because of thermal paste):



I then reduced the power in 10% (16 watts) increments which resulted in:

As you can see, it's roughly 4C per increment (0.25C/W), meaning you can probably go to 60% power without running the fans (I'd go with 50% just to be safe). I finally ran the simulation again using fans which resulted in:



To mount the fans, you can easily use 16 standoffs that are mounted to the 3mm plate. Are you happy with this as the final version? If yes I can finally finish the PCBs (stackable design would work without any issues here) or do you want to make some changes?

[DeadlyChemist]

questioning my life atm buuut

yeah i guess i gotta go 3-5mm plate and 4 or preferably more heatsinks + a slow large, mostly turned off fan (i assume as big as i can fit noctua)

can you make a more or less visual of how it would look like? and how much clearance do i need from the celling to the fan so it can propearly "grab" air

the only problem i have now is that it will get much more expensive because of shipping 3x instead of free shipping

yes, you can design the power boards

[OM222O]

questioning my life atm

This definitely got out of hand for a simple light (500x500m frame weighing around 10KGs  ) You may want to consider breaking this up into 4 smaller panels, each having 4 LEDs. That way you can eliminate the heat spreader plate as well saving some cost there.

But as it stands, here is the final design:
https://a360.co/3vGPL9p

I only included 1 fan as the model is really detailed and starts lagging the computer if I added 4. The fans need about 5cm of clearance to properly cool, so the overall stack height is about 15cm.

I also didn't realize how expensive fans have become (look for Noctua NF-A14 or its industrial PPC version which has better build quality). I still doubt you need that much LED power in one panel. If you break it into 4 panels, then the added height can even look good! Something similar to these:


Although I'm not sure how easy it is to find 140mm or 120mm square metal tubes or if you have access to a CNC (you need to cut out slots for ventilation, could be done by hand, but it'll be messy and look weird).

Alternatively, you can run the LEDs at 50% maximum brightness and avoid using fans, which should extend their life significantly.

Do you want to go with the large panel idea (500x500m frame, 400x400mm LED area) or do you want to consider a redesign for 4 smaller panels?

Edit:
here is a simulation of 4 LEDs directly on a heat sink with a fan:

even adding 10c for thermal paste, it's fairly cool at max power.

[amaschas]

Take it from someone who has just spent the last few years trying to design an LED application that utilized PWM dimming at a high enough frequency that it won't flicker on camera/video: use analog dimming for your implementation. Also the guy who recommended a DC/DC switching constant current LED driver on the first page had the right idea. The switching frequency of a SMPS is so high that there is no chance you will be able to see a noticeable flicker. In fact, the human eye can't see flicker above a few hundred hertz. On top of that, while a SMPS switches at a certain frequency, unless its designed really poorly (lots of voltage ripple) you are going to get a constant voltage out the other size. That's the whole point of a SMPS. There's really no need for a linear power supply for this application.

The recommendation for designing LED lighting is that your PWM frequency (for dimming) has to be above 20khz. Generally these applications target 30khz, since frequencies < 30khz are in the audible range, and can produce a noticeable whine. Even the slowest SMPS operate a frequencies of a few hundred kHz, so even if voltage ripple on the output were large enough to produce a noticeable change in the LED light output, you would need special equipment with a super high frame rate to detect it.

[tooki]

Take it from someone who has just spent the last few years trying to design an LED application that utilized PWM dimming at a high enough frequency that it won't flicker on camera/video: use analog dimming for your implementation.

What’s weird is how it must depend on the camera, too: my homemade dimmer runs at something like 25kHz, and I could not produce any flicker whatsoever with my DSLR (a Nikon D7000), no matter how hard I tried, even at the maximum shutter speed of 1/8000s.

In fact, the human eye can't see flicker above a few hundred hertz.

Just to be clear, that’s a commonly held fallacy. I’ve written about this before in other threads, and somewhere I have a big post with sources, but the absolute worst-case scenario requires 35kHz to absolutely eliminate flicker and related effects (especially the “phantom array” effect).

For me, I need at least 1.5kHz to not detect flicker in everyday lighting. Some people readily detect it at over 3kHz.

The recommendation for designing LED lighting is that your PWM frequency (for dimming) has to be above 20khz. Generally these applications target 30khz, since frequencies < 30khz are in the audible range, and can produce a noticeable whine.

Little by little, at least. Many people still hold onto the confusion about the sufficiency of a 60Hz frame rate and design lighting with PWM in the hundreds of Hz, which is nightmarish for flicker-sensitive folks like me.

FYI, the audible range for human hearing is 20Hz-20kHz best-case (children with undamaged hearing), but typically drops to 20Hz-17kHz (or less) in adults, with the upper range dropping with age. No human on earth can hear anything like e.g. 28kHz. If a switching supply produces audible noise when switching at 28kHz, it’s because there are sub-20kHz harmonics.

[DeadlyChemist]

so even if voltage ripple on the output were large enough to produce a noticeable change in the LED light output, you would need special equipment with a super high frame rate to detect it.

yeah this is the plan, sooner or later im getting a high speed camera, either something rented or might buy a chronos
i want to be free in whatever camera, framerate, shutter angle i choose so sadly linear is my only option for long therm
I know my powersupply (which i havent picked yet) might have flicker which would be detectable by the camera, but thats a problem for later

This definitely got out of hand for a simple light (500x500m frame weighing around 10KGs  ) You may want to consider breaking this up into 4 smaller panels, each having 4 LEDs. That way you can eliminate the heat spreader plate as well saving some cost there.

meh, i dont want 4 panels, it would be more work + im not able to make 4 panels look good
why does it have to be 500x500? cant it be 400x400? like the metal?

But as it stands, here is the final design:
https://a360.co/3vGPL9p

I only included 1 fan as the model is really detailed and starts lagging the computer if I added 4. The fans need about 5cm of clearance to properly cool, so the overall stack height is about 15cm.

I also didn't realize how expensive fans have become (look for Noctua NF-A14 or its industrial PPC version which has better build quality).

how about i add 5 heatsinks (cross) or 8 or 9 and just use larger than 100x100 fan? like 140mm or whatever in the middle?
i dont fancy spending like 100€ on fans (another reason why i didn't want fans (cost), plus fans just fail or get louder and louder)

[OM222O]

The frame is 500x500 because I chose 50x50x3mm angles which is one of the most common sizes (I thought you meant the window / light output area should be 400x400). If you want the frame to be 400x400, then the actual light output area would be 300x300. If you're ok with that, it's a really easy change.

Like I said before, adding more heat sinks without fans doesn't do much because they get saturated by natural convection and if you want to go down the path of using fans, 4 heat sinks would be enough. If you don't care about even spacing, you can fit everything under a single 200mm fan, but all of the light would be coming from the center of the panel (edges might look dimmer) but you can shrink the panel to be about 200x200 as well.

Do you want a redesign with a single 200mm fan and a smaller frame? or a 400x400 frame with everything in the middle?

You can also consider water cooling with something like this:
https://www.amazon.co.uk/sourcing-map-Aluminum-40x120mm-Polished/dp/B08JLNFJMN/ref=sr_1_3?crid=3ANDEO3EP9EFI&keywords=40x120+water+aluminum+water+cooling&qid=1659806267&s=electronics&sprefix=40x120+water+aluminum+water+cooling%2Celectronics%2C61&sr=1-3

you still need to buy fans and a radiator so it won't be cheaper, but it can be quieter.
It's VERY difficult to cool 160 watts passively; Like I said before, you can build the panel and limit the LEDs to 50% brightness, then later on if you really needed the extra output, you can always slap a few fans on top of the heat sinks. Let me know what the final decision is.

[DeadlyChemist]

Like I said before, adding more heat sinks without fans doesn't do much because they get saturated by natural convection and if you want to go down the path of using fans, 4 heat sinks would be enough. If you don't care about even spacing, you can fit everything under a single 200mm fan, but all of the light would be coming from the center of the panel (edges might look dimmer) but you can shrink the panel to be about 200x200 as well.

Do you want a redesign with a single 200mm fan and a smaller frame? or a 400x400 frame with everything in the middle?

preferably 400x400 frame, and the 400x400 plate beeing a structual part if possible?
the light i guess 350x350 or 300x300, depends what can fit, but 450^2 or 500^2 and the light 400^2 works too
what would you recommend?

You can also consider water cooling with something like this:
https://www.amazon.co.uk/sourcing-map-Aluminum-40x120mm-Polished/dp/B08JLNFJMN/ref=sr_1_3?crid=3ANDEO3EP9EFI&keywords=40x120+water+aluminum+water+cooling&qid=1659806267&s=electronics&sprefix=40x120+water+aluminum+water+cooling%2Celectronics%2C61&sr=1-3

sounds perfect, but not gonna use that due to maintance and other things needed... water is just a mess, especially 3m above me on the celling

you still need to buy fans and a radiator so it won't be cheaper, but it can be quieter.
It's VERY difficult to cool 160 watts passively; Like I said before, you can build the panel and limit the LEDs to 50% brightness, then later on if you really needed the extra output, you can always slap a few fans on top of the heat sinks. Let me know what the final decision is.

i think 1x 200mm or whatever fan and 4 heatinks with a 400^2 plate 3mm or 5mm
and just spin it up once i reach some target power or temperature (pretty easy todo in software)

[OM222O]

I just realized that the simulations don't match the advertised 2C/W of the heat sinks (sim shows around 0.3C/W) so the cooling design might have to wait for later, but knowing you want a 400x400mm frame and you're ok with adding a fan, I think I'll just finish up the PCBs. If you have one of those heat sinks on hand, you can do some thermal tests later to determine the number of heat sinks/fans required. Have you made an account for EasyEda? how about fusion 360?

[MrAl]

Hi,

Just thought i would add a little getting here late to this thread.

Yes one of the drawbacks to LEDs is that they have to have the right heatsinking and that can get rather elaborate.  There have been some advancements in the lab where some of the heat is added back to the electrical current that is used to power the LED, but i dont see any of that in any LEDs yet, maybe someone else here has.

The main thing though is that the main operating mode of an LED is that it requires a particular current unlike a regular incandescent bulb which requires a particular voltage.  To understand this better simply swap the word "voltage" with the word "current" and the word "current" with the word "voltage" in the following description of the operation of a light bulb.

A light bulb requires a certain voltage to operate correctly and with decent efficiency.  If you have a 120v light bulb you apply 120 volts to it and you get proper operation, and more importantly, it 'draws' the current it needs to operate correctly when supplied with this 120 volts.  By saying that it 'draws' the right current that means that we dont have to calculate or think about how much current it draws, as long as the power source can supply the voltage that we apply.  In a circuit analysis that would mean we would use a light bulb and a constant voltage source.

An LED is the same except it is best to drive it with a particular current.  The LED then automatically goes to the correct voltage it requires to operate, as long as the current source is capable of supplying that voltage.

There is a however a major difference between the two.  It is not as good to 'dim' a light bulb as it is to 'dim' an LED.  To dim a light bulb you reduce the voltage and the current goes down and the light bulb becomes less bright.  To dim an LED you reduce the current and the voltage goes down (slightly) and the LED becomes less bright.  So what's the big difference?
The difference is that when we reduce the voltage to the bulb the light intensity drops more than it would if the current and voltage followed a linear curve.  That means the efficacy drops by a large amount.  We get less light out per electrical watt than we did at full brightness.
When we reduce the  current to the LED the intensity drops approximately in proportion to the current level and thus the efficacy does not drop.  In fact, the efficacy actually goes up meaning we get more light output per watt.
This means that if we drop the voltage to the bulb by 25 percent we get possibly only 50 percent light output, but when we drop the current to the LED by 25 percent we get around 25 percent less light output.  That's a big difference, and remember the light output actually does not drop exactly in proportion the drop in light output is a little less, so when we drop the current by 25 percent the light output may only drop by 20 percent.  That means we get 80 percent light output for a 75 percent current level.  We also get a lot longer life.  With the bulb we get a somewhat longer life but not as much as with the LED.

There is one other difference worth mentioning.  That is, that some LEDs are meant to be run at full current in order to ensure the color temperature does not change.  That means that it is best to operate the LED at the full current if you want the right light color output.  The nice thing though is that if we use a true PWM current drive, the LED always receives the max current, just less often, and so the light dims but the LED color stays the same.
What this means is that the LED will be turning on and off rapidly which means it will flicker.  That in turn means  we have to use a high frequency, high enough that the human or pet eyes can not detect any change and it does not hurt any of us physiologically.
If we wanted to be super sure that we get no flicker and no bother to any animal eyes or brain, we could simply filter the heck out of the output current. Filtering means that the LED will get a very smooth current level no matter what the current level is set for.  The drawback though is that the color 'may' change although not all LEDs do change much, and the change may not be enough to bother us anyway. The advantage is higher efficacy because we are not driving (or banging) the LED at full current.

Hope this helps some and good luck with the project!

[amaschas]

What’s weird is how it must depend on the camera, too: my homemade dimmer runs at something like 25kHz, and I could not produce any flicker whatsoever with my DSLR (a Nikon D7000), no matter how hard I tried, even at the maximum shutter speed of 1/8000s.

If your PWM signal pulses every 1/25000th of a second, the light is going to turn on and off 3 times while the shutter moves across the image. I could see you experiencing some banding at low duty cycles.

yeah this is the plan, sooner or later im getting a high speed camera, either something rented or might buy a chronos
i want to be free in whatever camera, framerate, shutter angle i choose so sadly linear is my only option for long therm
I know my powersupply (which i havent picked yet) might have flicker which would be detectable by the camera, but thats a problem for later

So the Chronos captures ~25,000 frames in low res mode. A switching power supply turns on and off several hundred thousand times per second, minimum. Often they run at megahertz frequencies. And again, the only way you are even going to see a change in the light output is if the SWMP has really bad output voltage ripple problems, in which case I would recommend obtaining a better power supply. Trust me, you don't need a linear power supply for this use case.

[DeadlyChemist]

hey,

Im still alive, just super busy with work

ill let you know of the easyeda account and maybe fusion360

and yes, im still going linear route. im aware there are ways to make it work with high frequency but im not willing to try for this project and later possibly find out i can put my light or my camera into garbage because of banding/flicker (i had this issues too many times guys...)

I dont have a heatsink on hand, i planned to buy them all at once (to save shipping)

[tooki]

What’s weird is how it must depend on the camera, too: my homemade dimmer runs at something like 25kHz, and I could not produce any flicker whatsoever with my DSLR (a Nikon D7000), no matter how hard I tried, even at the maximum shutter speed of 1/8000s.

If your PWM signal pulses every 1/25000th of a second, the light is going to turn on and off 3 times while the shutter moves across the image. I could see you experiencing some banding at low duty cycles.

In testing, that’s exactly what I expected to see, but didn’t!!

[DeadlyChemist]

im not dead, moving in 2-3 weeks in a pretty nice apartment (42sqm)
after i buy a table i'll continue on the project.

however got this email:
Part Life Notification

Dear Valued Digi-Key Customer,

You have purchased the following part number from Digi-Key within the last two years. The manufacturer has announced an update to the part status.

Part Affected
Manufacturer   ONSEMI (VA)
Description   MOSFET N-CH 100V 5.8A DPAK
Manufacturer Part Number   FQD7N10LTM
Digi-Key Part Number   FQD7N10LTMCT-ND
Status   End Of Life
Last Time Buy Date   12/30/2022
Substitutes   Please click here

bit sad ngl but oh well, i have 50 of them, should be enough i hope

[MrAl]

im not dead, moving in 2-3 weeks in a pretty nice apartment (42sqm)
after i buy a table i'll continue on the project.

however got this email:
Part Life Notification

Dear Valued Digi-Key Customer,

You have purchased the following part number from Digi-Key within the last two years. The manufacturer has announced an update to the part status.

Part Affected
Manufacturer   ONSEMI (VA)
Description   MOSFET N-CH 100V 5.8A DPAK
Manufacturer Part Number   FQD7N10LTM
Digi-Key Part Number   FQD7N10LTMCT-ND
Status   End Of Life
Last Time Buy Date   12/30/2022
Substitutes   Please click here

bit sad ngl but oh well, i have 50 of them, should be enough i hope

Hi,

I would think there would be a thousand replacements by now made and sold.  Check around maybe find another part similar.

[John B]

There is one other difference worth mentioning.  That is, that some LEDs are meant to be run at full current in order to ensure the color temperature does not change.  That means that it is best to operate the LED at the full current if you want the right light color output.  The nice thing though is that if we use a true PWM current drive, the LED always receives the max current, just less often, and so the light dims but the LED color stays the same.

I experimented around with this colour temperature issue with my lighting controller project and found it to be of little to no practical concern. First step would be to purchase good quality high CRI discrete LEDs or LED panels.

My understanding is that the colour spectrum would skew slightly towards the red direction of the spectrum for low currents (not red in colour necessarily, but e.g. you could have more green, which is more to the red end compared to blue.)

The effect is perhaps more noticeable with 6000K LEDs vs 3000K warm white LEDs, but this is at very low currents.

FWIW I settled on using low dropout linear regulation as it's very easy to have ripple free and accurate regulation from 1-700mA.