Chronos > Chronos User Discussion
Create light for high-speed shooting 1500w
Nikon1:
--- Quote from: ExlerElektronik on January 16, 2021, 08:08:36 AM ---A few notes about the ideas discussed here, as my business is designing electronics:
1) If the module has 6 individual LED chips you will have to supply them by individual drivers. Connected in parallel a LED which heats up will draw more current from the given voltage resulting in even more heat dissipated until is's gone :-[ .
2) I would not recommend to use a step up design if you want to get flicker free light. Without lots of filtering at the output a step up will produce strong flicker at it's PWM frequency.
3) Power dissipation in step up converter. The MOSFET or IGBT in an step up converter has to handle the input current and the output voltage. Thus converting 1500 W from 12 to 36 V will have an input current of 125 A. Power distribution is split in two components, static and dynamic loss.
Static loss can be estimated with the ON resistance RDSON for MOSFETs to be about (InputCurrent)^2*RDSON*(Vin/Vout). For the IXIS device referenced in an earlier post this would be about 125^2*0.0055*12/36 = 29 W (quite fine).
Using an IGBT this estimate would be saturation voltage VCEsat (typically in range of 2 V) * input current * Vin/Vout, resulting in about 83 W (need for good cooling).
Dynamic loss is caused by switching and could be estimated for both with SwitchingTime*Vout*InputCurrent*SwitchingFrequency. Trying to run the IXIS MOSFET at 125 kHz this leads to 200ns*36V*125A*125000 = 112.5 W.
To cool a total of >150 Wwill need water cooling or a pretty big heatsink for the MOSFET. Note the 1500 W shown in the datasheet is the maximum power dissipation if you manage to keep the case temperature at 25°C!
More notes:
a) As you can guess from the dynamic loss estimate above, switching at 5 MHz will definitely not work at this load.
b) Gate current during switching will be several A, so you need a good gate driver circuit to achieve the switching times shown in the datasheet.
c) In general switching converters will produce high electromagnetic noise emission. At this power there is a good chance you run into serious issues up to disturbing radio, TV, mobile and WiFi reception.
d) Stray inductance in the circuit caused by even short wires might produce strange voltage peaks which could easy destroy your electronics.
e) As of the voltage/current characteristic of a LED you won't get stable output current/power unless you put in a control circuit. Without this active control circuit there is a good chance you overload your LED with a slight change in PWM duty cycle. Warning: You won't get a linear relation between PWM duty and output current/power! The actual will be more like a threshold where the output starts to increase pretty steep without current control.
Conclusion: Don't try to build your own circuit unless you are familar with design of power electronics.
--- End quote ---
YES, i can mostly agree with everything you said. 5MHz will be theoretical, as is the thing about 1500W power Dissapation. Reality is far from that, just like the 20kHz for the Fat IGBT Bricks, those gates just have way to high Gate Capacity to work well for HF. i also would have assumed, that everyone trying to actually build a PWM for >1KW Load would use pretty beefy cooling on the Power Components without even asking, so i didnt even go as far as to even mention it, but yes, you are also very right about that, needs serious cooling, or it will fail from overheating in a matter of Seconds or Minutes.
In general i wouldnt use Pulsed Light anywhere where it isnt actually needed. So i am not quite sure why you would want to do that in the first place, since as you allready said EMV issues (not sure if that is an international Term or just a german thing) are a Problem at any frequency where it would be even remotely interesting for Highspeed Filming use.
So i get why feeding it with 36V a Power-Supply directly would be an very simple and attractive Solution, allready because simplicity (but please, as i allready said multiple times, and ExlerElektronik also said here, use Current Limiting for each of the Single Parts of the LED, they will fail Quickly if you just wire them together in parallel and feed them Constant Voltage @SergeyKashin), but really couldnt reccomend PWM for this thing.
I personally used Step Up-Converters like this for LED-Chips and i never had any issues with flickering even at 24kfps, so from my experience up to now, they are fine for this kind of use, are cheap and readily available, and dont need any additional filtering, whenever i used those bevore (thats also why i initially reccomended them, they worked well for me so far, have current limiting built-in and are dimmable). From using them, i dont really see an issue why they wouldnt work, but i see lot of Potential Problems with the PWM Idea. While this might even work just fine (if done Right. But as you dont seem to know about those IGBTs, i assume your knowledge about Elecronics is not at a Level where you would just go and build one yourself), i wouldnt do it. (@SergeyKashin) .
just my opinion, but if you want it as simple as possible, use your 36V PSU directly with at least some Resistors for the Most Primitive Current Limiting for each Chanel of the Chip, or do propper Step-Up / Step-Down Controllers for each chanel with current Limiting if you actually want it dimmable (or some Form of Active Linear Current Limiting, which would basically be a space heater and would draw tons of Power at all Times). (@SergeyKashin)
SergeyKashin:
I already have 2 PWMs of 1000 watts and they work perfectly, although I can not yet turn them on at full power because I only do water cooling for the LED. Therefore, I am looking for a more powerful one so that 2 wires do not lead to the diode, but only 1 two-wire one. To do this, I want to find a pwm at 2000 watts 150 kHz with an operating range from 12 to 60 volts
In PWM there is such a transistor
http://ncepower.com/Upload/SGT/NCEP15T14Tdatasheet-15470517689.pdf
There is also such a power source that allows you to adjust the power immediately in it. LM 40-75ER -SORENSEN
Nikon1:
--- Quote from: SergeyKashin on January 16, 2021, 01:58:46 PM ---There is also such a power source that allows you to adjust the power immediately in it. LM 40-75ER -SORENSEN
--- End quote ---
Yes, there are adjustable PSUs, but for the Kind of Power that Chip Needs, they get really Bulky and crazy Expensive, if you got that kind of Money to afford one of those, might as well look into buying some Finished Lamp with such chips inside, propably will cost about the Same.
.
If you actually just wired those Chanels on the Chip together as 3+3 Configuration without anything else directly to two PWMs, i would really be interrested How long that Chip will survive that, bevore some Part of the LED Fails...?
.
If you allready do Watercooling on the Main LED, might as well think about Also Watercooling the Power Mosfets on the PWM, there are Penty of Cheap Waterblocks available which would be pretty easy to add into your existing Loop:
https://www.ebay.de/itm/Aluminium-Wasserkuhlung-Block-40x40x12mm-Liquid-Kuhler-Wasserblock-Heizkorper/283495000224?hash=item42019e68a0:g:wzsAAOSwaKVc0BqI
Nikon1:
So finally i finished my Lamp. Was Planing to do that for quite a while now. I also considered to use one of those Big chips, but in the End Went with 150W LED Chips with integrated Drivers, which run directly from Mains Voltage. Those will flicker when used normally, but the ones that i came across until now, will Produce Flicker-Free Light when you feed them with well enough Filtered DC Voltage (they do rectify the Mains AC internally anyways, but dont have a big enough Capacitor built in, propably mostly because of size and also Cost, and because of that will cut out everytime time Voltage Crosses Zero, which causes the Flicker), which is still way simpler to do than building a full on Power Supply for any of the Lower Voltage Chips out there of similar Power. Only "Problem" of that is, that i cant really control the Brightness, apart from just switching off entire Chips if i wanted to, but i figured, when i get that Lamp out, i propably need all the Light i can get anyways, so just have it at full power or else use Some Smaller 300 or 200W Lamps (which i can control in Brightness).
.
Main Benefit of those Chips is, that they are really cheap compared to the 1500W ones, even those cheaper ones at ~500 each. I paid less than 300€ for all the Chips needed for this lamp (9 chips making it a 1350W Lamp in total), the 20mm Thick Aluminium Plate (including Cost for getting it cut to size and shipping), Thermal paste and Mounting Hardware all to gether.
.
Biggest downside on the other hand is drilling and Hand Taping (as its to big for any of my machines) a ton of small holes, and almost an Gurantee to break one or more Taps while you are at it.
.
I originally Planed on building two of those (i allready have the Chips for the Second one), but we will see how i like that one when actually using it, bevore i do that.
Right now the Chips are cooled mainly by thermal Mass of the huge Piece of Metal behind it and the Still pretty consideable Surface Area of the Whole thing, keeping it cool for like 10 to 12 Minutes bevore it gets uncomftably hot (calculated it Month ago to last some time, not sure anymore, but i guess i assumed ~10Min, To stay below 100°C on the Plate, only from Thermal Mass, again assuming the Plate Starting at 20°C and upright position of the Lamp with some Light airflow from the Hot air moving upwards), which is good enough for now, but i propably end up adding some better cooling to it, by either Bolting on some Big L-Bar Aluminium As Fins on the Back (would be like 100€ in Metal to add, and some Work), Watercooling it or something else to make it able to run longer or even constantly. Maybe even a big powerfull enough Fan would improove the Time by a lot, right now its just purely passive Cooling without any Forced Air movement.
.
Overall, this isnt a pretty/ideal solution, but it does Produce quite nice Light (and a lot of it also) very cheap.
Would not reccomend to build anything like this without some serious bevorehand experience, because this uses Mains Voltage all over it, and is actuall fairly dangerous and can potentially easily kill someone.
SergeyKashin:
1. These LEDs have a very low color rendering index, so you will get a terrible picture on processing (unless of course this is an experiment in which Cree is not so important). I have a 1000-watt LED, its price is $ 85, but it has Cri80 and then I looked at what he got on installation and was shocked, it turned out to be a mess of colors and shades
2. The outlet has 60 hertz and these LEDs can strobe
3. If they have such a power of 1350 watts, then your radiator would heat up in less than a minute, take a wattmeter and measure how much LEDs consume, you will see that it will be about 500 watts, maybe even less
Now I am assembling LED panels in the size of 30x20 with a capacity of 250 watts. This panel heats up in about 1 minute to a very hot state and to reduce the temperature, I installed 2 coolers 140x140mm 1200 rpm there
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version