karimwassef
Active member
Are there DCDC boost constant current converters at 100W with 30-60V output?
Looking for a specific DC CC driver for LEDs
- Thread starter karimwassef
- Start date
karimwassef
Active member
ok. the result of today's little experiment...
I put a 1.1 Ohm resistor (30W capable) in series with the current flow and used the voltage source directly. Then I slowly walked the voltage up from 40.5V, measuring the current in small increments...
40.5V 2.12A 86W
41.6V 2.34A 97W
42.9V 2.63A 113W
44.4V 2.92A 130W
45.6V 3.18A 145W
46.1V 3.29A 152W
46.6V 3.40A 158W
47.1V 3.51A 165W
That's where I decided to stop... clearly, very linear. The curve shows a resistance of about 4.76 Ohms and a cross at 30.44V.
This is supposed to be a 3A 34V chip, but it's nowhere near that.
I decided to get a 400W digital CC boost with up to 80V DC output and a digital current setpoint.
http://www.ebay.com/itm/111906927331
If nothing else, I should be able to investigate the operating limits around this chip.
I put a 1.1 Ohm resistor (30W capable) in series with the current flow and used the voltage source directly. Then I slowly walked the voltage up from 40.5V, measuring the current in small increments...
40.5V 2.12A 86W
41.6V 2.34A 97W
42.9V 2.63A 113W
44.4V 2.92A 130W
45.6V 3.18A 145W
46.1V 3.29A 152W
46.6V 3.40A 158W
47.1V 3.51A 165W
That's where I decided to stop... clearly, very linear. The curve shows a resistance of about 4.76 Ohms and a cross at 30.44V.
This is supposed to be a 3A 34V chip, but it's nowhere near that.
I decided to get a 400W digital CC boost with up to 80V DC output and a digital current setpoint.
http://www.ebay.com/itm/111906927331
If nothing else, I should be able to investigate the operating limits around this chip.
karimwassef
Active member
I guess one possibility is that this is a higher voltage/higher power LED that failed spec and so got scrapped and binned as a lower power chip?
Even 200W LEDs still run at 34V.
300W chips go to 48-50Vf with a peak current of 7A.
500W chips go to 70-78Vf with a peak current of 7A.
But those chips are physically larger:
Even 200W LEDs still run at 34V.
300W chips go to 48-50Vf with a peak current of 7A.
500W chips go to 70-78Vf with a peak current of 7A.
But those chips are physically larger:
oreo57
Well-known member
you do understand that was never a "hard" number....
I understand your surprise on the "out of spec" voltage but current draw increases till burnout..
3A is meaninless..except as a starting point. Just because the chart ends............:
so they "make "stuff" up.....
http://electronics.stackexchange.com/questions/151627/why-does-an-led-have-a-maximum-voltage
theatrus
100-mile-commuter
Right, once out of the knee and into the fully forward biased conducting region, current capability is thermally limiting (till melt down). Its really only interesting to check Vf at a rated drive current (set CC to 3A, what Vf is needed over the temp range to meet that)
karimwassef
Active member
Yes, I understand diode characteristics. I didn't expect that they'd be that far off on a "guesstimate" of forward voltage.
At 3A, the drop is 45V.
At 3.5A, the drop is 47V.
Since thermally, it looks like I can run the LED at 200W with my current heatsink and fan - I just need to determine my LED temperature. I know I have a thermocouple somewhere.
Here's what I was expecting - I was expecting a voltage drop of 34V "almost" independent of current. Then a PWM current from 0 to 3.5A to modulate output power. There was no hard voltage drop... it continues to go up and down with current. The resistance of the LED is significantly more than I expected.
At 3A, the drop is 45V.
At 3.5A, the drop is 47V.
Since thermally, it looks like I can run the LED at 200W with my current heatsink and fan - I just need to determine my LED temperature. I know I have a thermocouple somewhere.
Here's what I was expecting - I was expecting a voltage drop of 34V "almost" independent of current. Then a PWM current from 0 to 3.5A to modulate output power. There was no hard voltage drop... it continues to go up and down with current. The resistance of the LED is significantly more than I expected.
Last edited:
karimwassef
Active member
So. Next experiment was to see if I can use a PWM to dim. I used an old remote LED PWM driver on a 5V powers supply and a resistor to create a cheap PWM signal and drive it into the Sure driver. It works backwards - no signal or short to ground = on. +5V or Vin is off.
It worked and I can go from completely off to "full power" - depending on where I choose to set that.
However, running in PWM mode did change the voltage to current relationship... the voltage was lower for the same current.
It worked and I can go from completely off to "full power" - depending on where I choose to set that.
However, running in PWM mode did change the voltage to current relationship... the voltage was lower for the same current.
Last edited:
karimwassef
Active member
The jumpers on the Sure Buck work, as does the PWM function. I really like this little board.
It runs up to 130W (2.92A and 44.6V) with PWM dimming (well, on this chip).
0.00A 0.00A 0W
0.10A 2.70A 0W
0.30A 6.00A 2W
0.50A 9.00A 4W
0.60A 10.0A 6W
0.80A 15.0A 13W
1.06A 18.0A 19W
1.39A 22.0A 31W
1.64A 27.0A 44W
2.08A 32.7A 68W
2.16A 34.5A 75W
2.27A 36.7A 83W
2.45A 39.1A 96W
2.52A 39.9A 101W
2.67A 41.9A 112W
2.74A 42.7A 117W
2.89A 44.1A 127W
2.92A 44.6A 130W
and it will dim to zero.
I still can't find a CC boost up to 60V 200W with dimming. The CNC digital dimming boost is nice for experimenting, but not for ongoing use. It's also a lot more expensive ~$30.
There are plenty of cheap solutions up to 50V and 250W, but no PWM or 0-10V dimming:
http://www.ebay.com/itm/DC-DC-boost...ower-supply-250W-10A-LED-Driver-/181940997254
What do you guys think?
It runs up to 130W (2.92A and 44.6V) with PWM dimming (well, on this chip).
0.00A 0.00A 0W
0.10A 2.70A 0W
0.30A 6.00A 2W
0.50A 9.00A 4W
0.60A 10.0A 6W
0.80A 15.0A 13W
1.06A 18.0A 19W
1.39A 22.0A 31W
1.64A 27.0A 44W
2.08A 32.7A 68W
2.16A 34.5A 75W
2.27A 36.7A 83W
2.45A 39.1A 96W
2.52A 39.9A 101W
2.67A 41.9A 112W
2.74A 42.7A 117W
2.89A 44.1A 127W
2.92A 44.6A 130W
and it will dim to zero.
I still can't find a CC boost up to 60V 200W with dimming. The CNC digital dimming boost is nice for experimenting, but not for ongoing use. It's also a lot more expensive ~$30.
There are plenty of cheap solutions up to 50V and 250W, but no PWM or 0-10V dimming:
http://www.ebay.com/itm/DC-DC-boost...ower-supply-250W-10A-LED-Driver-/181940997254
What do you guys think?
Last edited:
karimwassef
Active member
you do understand that was never a "hard" number....
I understand your surprise on the "out of spec" voltage but current draw increases till burnout..
3A is meaninless..except as a starting point. Just because the chart ends............:
so they "make "stuff" up.....
http://electronics.stackexchange.com/questions/151627/why-does-an-led-have-a-maximum-voltage
Oreo - nice and simple explanations... I love it. I would derive from my IV curve that my thermal management is keeping thermal runaway at bay - so I'm basically extending the operating range of the chip.
Would you agree?
karimwassef
Active member
theatrus - what are the limits of your board? Max/Min input/output/power?
If I start at 60Vin (which I can), would it buck down to 50V and 4A (200W)?
I think I can thermally manage up to that power level with my conduction cooled/liquid cooled design.
If I start at 60Vin (which I can), would it buck down to 50V and 4A (200W)?
I think I can thermally manage up to that power level with my conduction cooled/liquid cooled design.
theatrus
100-mile-commuter
Looks like I fitted it with the LM3409HV, which puts the next limit on the FET (60V) and the input filter capacitors (50V). The input filter capacitors are the most concerning. Running at 60V should be ok for short periods of time - it would need to be restrapped for 100V caps (at a much higher relative cost) and larger footprint of those for long term operation.
Current limit is set with a 0.080 shunt which is ~ 3A. Going to ~5A is easy. This design still requires swapping the shunt (a 2010 footprint SMT part) to increase current. If it was re-strapped for 5A operation, you can use the analog (0-1.2V) dimming port to drop the peak current to other levels, which is a design I'm messing with right now.
Current limit is set with a 0.080 shunt which is ~ 3A. Going to ~5A is easy. This design still requires swapping the shunt (a 2010 footprint SMT part) to increase current. If it was re-strapped for 5A operation, you can use the analog (0-1.2V) dimming port to drop the peak current to other levels, which is a design I'm messing with right now.
karimwassef
Active member
What do you think the cost would be for the buck compared to a boost version starting from 24V (for maximum flexibility)?
theatrus
100-mile-commuter
Since both are external controller based designs, there are two contributing factors:
- Higher voltage parts (>40V) are much rarer, so if the input voltage is higher there is a significant limitation of what you can find.
- Boost converters will obviously run a lot more current through their magnetics, which can add a cost and the switch currents get very high.
In the end? It may be a toss up. A 50V output with a 24-36V (common) input boost design would probably be feasible. Thats about 10-15A peak through the main inductor, which does limit your off-the-shelf choices (i.e. http://www.digikey.com/product-detail/en/bourns-inc/2300LL-220-V-RC/2300LL-220-V-RC-ND/725902) . I would need to check a SEPIC design as well (which has other drawbacks).
- Higher voltage parts (>40V) are much rarer, so if the input voltage is higher there is a significant limitation of what you can find.
- Boost converters will obviously run a lot more current through their magnetics, which can add a cost and the switch currents get very high.
In the end? It may be a toss up. A 50V output with a 24-36V (common) input boost design would probably be feasible. Thats about 10-15A peak through the main inductor, which does limit your off-the-shelf choices (i.e. http://www.digikey.com/product-detail/en/bourns-inc/2300LL-220-V-RC/2300LL-220-V-RC-ND/725902) . I would need to check a SEPIC design as well (which has other drawbacks).
karimwassef
Active member
True, but with a buck boost, it would be the ultimate any input DC driver. I haven't seen a constant current sepic though. Do you have a controller that can do it?
karimwassef
Active member
So.. more experiments. I took an old 100W cool white chip on a thin intel heatsink/fan. Ran it through the Sure buck and it did exactly what I thought it would.
Voltage was 35V and current was 2.6A! Input DC source voltage didn't matter... 40V to 45V... It was rock solid at 35V. This is exactly what I was expecting from a 100W multi-chip. At least that justifies by surprise with the other chip.
Now... I turned it off.. and then on again. This time, it only ran at 2.3A... and again off/on.. 2.0A... again... 1.7A... for some reason, every time I powered this LED/thin HS, the current through it dropped. But the voltage was about the same 35-36V.
<a href="http://s1062.photobucket.com/user/karimwassef/media/0110407D-BAFC-423C-9977-F233DCAAB81D_zpshls7blic.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/0110407D-BAFC-423C-9977-F233DCAAB81D_zpshls7blic.jpg" border="0" alt=" photo 0110407D-BAFC-423C-9977-F233DCAAB81D_zpshls7blic.jpg"/></a>
Then I just let it run.. as it heated, the current went up.. 1.7A, 1.9A, 2.1A ... then it powered off. I assume the converter detected runaway and turned itself off.
I let it cool down a little and turned it back on and it was back at 2.6A.
I'm assuming the undersized heatsink was really not doing the job and allowed the chip to run away.
Interesting that the voltage drop was so steady compared to the linear response with the blue chip on the larger HS.
I have a much bigger HS/fan and I'll try that on a third blue chip.
I don't get the UV chips for a couple of weeks. If cooling will allow me to run the chips in "over-power" mode (like over-clock for CPUs), that would be a remarkable cost benefit.
Voltage was 35V and current was 2.6A! Input DC source voltage didn't matter... 40V to 45V... It was rock solid at 35V. This is exactly what I was expecting from a 100W multi-chip. At least that justifies by surprise with the other chip.
Now... I turned it off.. and then on again. This time, it only ran at 2.3A... and again off/on.. 2.0A... again... 1.7A... for some reason, every time I powered this LED/thin HS, the current through it dropped. But the voltage was about the same 35-36V.
<a href="http://s1062.photobucket.com/user/karimwassef/media/0110407D-BAFC-423C-9977-F233DCAAB81D_zpshls7blic.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/0110407D-BAFC-423C-9977-F233DCAAB81D_zpshls7blic.jpg" border="0" alt=" photo 0110407D-BAFC-423C-9977-F233DCAAB81D_zpshls7blic.jpg"/></a>
Then I just let it run.. as it heated, the current went up.. 1.7A, 1.9A, 2.1A ... then it powered off. I assume the converter detected runaway and turned itself off.
I let it cool down a little and turned it back on and it was back at 2.6A.
I'm assuming the undersized heatsink was really not doing the job and allowed the chip to run away.
Interesting that the voltage drop was so steady compared to the linear response with the blue chip on the larger HS.
I have a much bigger HS/fan and I'll try that on a third blue chip.
I don't get the UV chips for a couple of weeks. If cooling will allow me to run the chips in "over-power" mode (like over-clock for CPUs), that would be a remarkable cost benefit.
Last edited:
karimwassef
Active member
Here are the three heatsinks I'm experimenting with
<a href="http://s1062.photobucket.com/user/karimwassef/media/D8704F9A-2A0A-461C-ADF8-D039103B7E04_zpscwi5rgf1.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/D8704F9A-2A0A-461C-ADF8-D039103B7E04_zpscwi5rgf1.jpg" border="0" alt=" photo D8704F9A-2A0A-461C-ADF8-D039103B7E04_zpscwi5rgf1.jpg"/></a>
Since cooling can increase power throughput, I might have to accelerate the liquid cooling project.
<a href="http://s1062.photobucket.com/user/karimwassef/media/D8704F9A-2A0A-461C-ADF8-D039103B7E04_zpscwi5rgf1.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/D8704F9A-2A0A-461C-ADF8-D039103B7E04_zpscwi5rgf1.jpg" border="0" alt=" photo D8704F9A-2A0A-461C-ADF8-D039103B7E04_zpscwi5rgf1.jpg"/></a>
Since cooling can increase power throughput, I might have to accelerate the liquid cooling project.
karimwassef
Active member
So with the freedom of a $5 chip, I decided to experiment.
I removed the series resistor and fuse and ran directly off my 48V source.
I used the original fan and heatsink assembly.
I started at 43V and slowly increased, measuring current. These plots are V & P vs. I instead of traditional diode curve because I wanted to show both on the same curve with the linear and quadratic fits for reference.
I ran three iterations... increasing voltage (and corresponding current) until the current suddenly drops down indicating a failure of a cell in the chip (with an increased resistance and corresponding voltage drop).
<a href="http://s1062.photobucket.com/user/karimwassef/media/karimwassef001/0_zpsttugqseg.png.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/karimwassef001/0_zpsttugqseg.png" border="0" alt=" photo 0_zpsttugqseg.png"/></a>
It doesn't look like the power is driving the failure (with sufficient cooling). It's now limited by maximum voltage (driving a local thermal hotspot to failure). That seems to have been ~49V on the first iteration and ~48.5V on the second.
I removed the series resistor and fuse and ran directly off my 48V source.
I used the original fan and heatsink assembly.
I started at 43V and slowly increased, measuring current. These plots are V & P vs. I instead of traditional diode curve because I wanted to show both on the same curve with the linear and quadratic fits for reference.
I ran three iterations... increasing voltage (and corresponding current) until the current suddenly drops down indicating a failure of a cell in the chip (with an increased resistance and corresponding voltage drop).
<a href="http://s1062.photobucket.com/user/karimwassef/media/karimwassef001/0_zpsttugqseg.png.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/karimwassef001/0_zpsttugqseg.png" border="0" alt=" photo 0_zpsttugqseg.png"/></a>
It doesn't look like the power is driving the failure (with sufficient cooling). It's now limited by maximum voltage (driving a local thermal hotspot to failure). That seems to have been ~49V on the first iteration and ~48.5V on the second.
karimwassef
Active member
So the chip was ~180W capable (with sufficient cooling). After degradation, it's now ~140W capable.
There's also a likelihood that the output of my voltage source wasn't stiff enough so small jumps could have driven current spikes aggravating the failure.
Powering up and down, I don't see any points not coming on though. I had expected to see a point or a line missing after thermal failure. Haven't figured that out yet.
There's also a likelihood that the output of my voltage source wasn't stiff enough so small jumps could have driven current spikes aggravating the failure.
Powering up and down, I don't see any points not coming on though. I had expected to see a point or a line missing after thermal failure. Haven't figured that out yet.
karimwassef
Active member
Another interesting consideration:
If I had just assumed the part was as spec'd and set the output voltage to 34V, it would have only pulled 0.7A and generated ~25W.
If I has set my current source to 3A, it would have run to ~45V and generated ~135W.
One of the complexities here may be that using these chips require individual re-characterization to determine where they can run. That's ok for a DIY, but probably worthless for general industrial use.
If I had just assumed the part was as spec'd and set the output voltage to 34V, it would have only pulled 0.7A and generated ~25W.
If I has set my current source to 3A, it would have run to ~45V and generated ~135W.
One of the complexities here may be that using these chips require individual re-characterization to determine where they can run. That's ok for a DIY, but probably worthless for general industrial use.
karimwassef
Active member
What do you think? Reasonable conclusions?
Similar threads
