LED driver design question

OldGuyInOregon · May 15, 2015

Help! I am designing a new LED lamp. I want 2x Vero 29's for the COB's. The last time I did this, I got the Meanwell drivers and they are way freaking expensive. I want to power my COB's using a scrap laptop ACAD and a simple converter circuit. Getting rid of the Meanwell driver more than pays for the second COB and second heatsink.

Is there any reason not to realize a boost converter from some discrete components and hang an output fet off the back of an LED heatsink? What is a integrated LED driver really buying me for all the money? Please comment! Thanks!!!

Dion · May 15, 2015

hi nice to see you again

u know i like cheap chines drivers- no reason you can find one for 10 bucks on aliexpress

OldGuyInOregon · May 16, 2015

Dion said:
hi nice to see you again u know i like cheap chines drivers- no reason you can find one for 10 bucks on aliexpress

Hey Dion! I have been here, I mostly read and think... Nice work on the FAQ!

Hmm. If I can find the right part, I suppose I don't care where it comes from, but I have been using Mouser and Digikey because the folks I know that do this stuff professionally use them.

The part needs to allow Vin 19VDC, Vforward 40V, Iforward 2.2 A (from one or two channels).

I will have to see if the integrated driver is better, I get something like this

Dion · May 16, 2015

http://www.aliexpress.com/store/group/100-160W-CE-PF0-95/346689_253141008.html

http://www.aliexpress.com/store/group/100-160W-CE-PF0-95/346689_253141008.html

just use a driver- im self tought and know nothing about lecky other than using it to power leds- if i understood u wanna add a resistor and step up booster to run of a 19v laptop power suply?

so smth like this https://www.fasttech.com/products/1612/10004219/1320700-12-24v-20w-high-power-dc-led-driver only with a current of 2.4A?

OldGuyInOregon · May 16, 2015

Dion said:
just use a driver- im self tought and know nothing about lecky other than using it to power leds- if i understood u wanna add a resistor and step up booster to run of a 19v laptop power suply? so smth like this ... only with a current of 2.4A?

Your input is awesome. Perhaps others will be interested, it seems like LED drivers are probably the least understood part of LED lamp design.

Yes, I want to use a laptop power supply, I call these acad (AC ADapter), the standard ones are about 19v @ 4.5A, and I want to attach about an 80 watt load. I get them at goodwill $3 a piece. See the attached bill of materials (BOM). On the BOM, I show $18 for a driver, but I do not know how to do this yet with a single IC or a single circuit board. Anyway I end up with two vero 29's running at maybe 38v x 1.1A on big heatsinks.

On the last link, yes, the circuit parameters would be OK if adjusted. The problem I have is that there is not enough data to know if it would be a good solution. The picture shows me a circuit board and two caps and a big ass coil. I can guess it is a transformer. If the circuit implements a step up transformer, then my questions are:
1) efficiency - I think transformers have lower than 80% efficiency, I think you'd like over 85%,
2) thermal - how would one thermally manage this thing,
3) does it really protect the LED - I don't know, back later with a comparison of "nFET boost converter" vs. "step-up transformer" for LED driver apps.

When I did my last design four months ago, no matter the type of driver circuit, you start running into problems when you drive much more than 1A because few available LED's greater than 1A = few available drivers = few choices on other circuit parameters. Also, low quantity of parts means big price increases for incremental increases in driver output current. Example: if a part outputs 650mA for $6 (like that third link), you'd hope a similar 2A part would cost $18 (3x as much), but my guess is it will be more. I ended up with that $50 Meanwell part because I wanted 40v @ 2A. It was the cheapest thing that does this I found, and at $50, it is too expensive.

Zen of LED circuit design

Knowing what is required is most of knowing

Dion · May 16, 2015

yep thats why i buy chinese drivers lol

up to 40v caan be had for around $10 with about 2.4A being the highest current ive seen

but for paralel wiring/ multiple cobs it gest too expensive for this guy

hence i build mine with single driver per cob although i have toyed with the idea of using V boosters and a phone charger

i sir do not undertsand electricity enough to play with it- when i fiddle i stay with dc power only

OldGuyInOregon · May 20, 2015

Recommended by another person... http://www.alibaba.com/product-detail/100w-led-driver-2400ma_737618664.html

I said "It won't work with 19vdc, those drivers are 120vac". He said "Yeah, get one of those, and if you want to, you can put your acad in the same room. They use less power if you don't plug them in".

Then I found this... http://www.ledsupply.com/led-drivers/buckblock-2100ma-constant-current-led-driver-with-dimming

I think this is it.

Dion · May 20, 2015

yeah exactly what u r looking for
i think rapid led has some buck boosters too but ultimately you pay just as much/more than using an ac to dc driver
glad u found what u needed though

good luck

OldGuyInOregon · May 20, 2015

Now I am thinking I was making the problem too hard... If I use a diode that has Vforward of about 18VDC. Then I do not need a boost converter.

BXRA-56C4500, has Vforward of 17.9VDC @ 1750mA. So three of them in parallel gets Vforward of 17.9VDC @ 1.5A in each of the three diodes.

How does this look?

OldGuyInOregon · May 20, 2015

Here is the details on the estimated bill-of-materials

OldGuyInOregon · May 20, 2015

I think this gets 0.86 X 0.94 about 80%, not great, but it is cheap

Dion · May 21, 2015

is it really worth all teh hassle though chief?

80%? is that power factor/efficiency?

the drivers usually give 95% efficiency

OldGuyInOregon · May 21, 2015

Dion said:
is it really worth all teh hassle though chief? 80%? is that power factor/efficiency? the drivers usually give 95% efficiency

I am just designing. I will not know if the lamp is worth building until the design is finished. If it is not, I will try again. I learn much from the designs whether I build them or not, and I think that is all that matters to me.

Efficiency, the amount of supply power that actually makes it to the diode. I am getting 93% with the 42V Meanwell part. That is why I did not spend too much time on looking at that transformer circuit board, because the theoretical efficiency is not great. To get a better efficiency, I just need a more efficient 19VDC source, because the balance of the circuit should be OK. As long as I think the driver efficiency is over 90% I will not worry about the acad efficiency at the moment.

Hang with me brother!

OldGuyInOregon · May 21, 2015

OK, 1) I do not need the zener diode, and 2) if a diode fails, then the other two fry because too much current. I either need to add another diode (so one failing leaves three and they can still take the current), or make the current-sensing branch R2 a separate branch for each diode. Hmm.

Dion · May 21, 2015

OldGuyInOregon said:
I am just designing. I will not know if the lamp is worth building until the design is finished. If it is not, I will try again. I learn much from the designs whether I build them or not, and I think that is all that matters to me.

Efficiency, the amount of supply power that actually makes it to the diode. I am getting 93% with the 42V Meanwell part. That is why I did not spend too much time on looking at that transformer circuit board, because the theoretical efficiency is not great. To get a better efficiency, I just need a more efficient 19VDC source, because the balance of the circuit should be OK. As long as I think the driver efficiency is over 90% I will not worry about the acad efficiency at the moment.

Hang with me brother!

thanks for your explanation i love to learn about this stuff and obviously you know a lot- i have often thought bout overcoming the cost of drivers using v boosts but still seemed to difficult for the $ saved

OldGuyInOregon · May 21, 2015

Dion said:
thanks for your explanation i love to learn about this stuff and obviously you know a lot- i have often thought bout overcoming the cost of drivers using v boosts but still seemed to difficult for the $ saved

Thanks Dion! I agree that reducing the cost of the driver and power supply is the only way forward. I want the power supply and driver to be about 10% of the cost of the lamp, not 50% or more, that just seems wasteful. That is why using the acad is required. They are $3 and I can easily get one. Also, 90 watts is the perfect size for a lamp. The reliability, thermal, safety, damp-resistance of these things - perfect. Right now the only downside is 85% efficiency or so, but I will revisit that issue before I buy anything.

OldGuyInOregon · May 21, 2015

So the 2nd attempt failed. I removed the zener and added a diode to the circuit to address the diode failure issue so now 4 in parallel. The new circuit and the original circuit I posted also have a problem.

The analysis goes Vin = 19V, Iin = 4.5A, Vforward = 17.9v, V at R2 = Vin - Vforward = 1.1v, Q1bias = > 3v.

This circuit will not work because there is insufficient voltage across R2 to bias Q1. If Q1 never turns on, then there is no current through the NFET, and the diodes never turn on.

What this is telling me is that when I went looking for diodes, I tried to crowd Vin too close to Vforward, and there needs to be a 3v or 4v difference. I think I am going to decree that Vforward has to be less than 15VDC, and look for another diode.

Back in a bit.

Dion · May 21, 2015

OldGuyInOregon said:
Thanks Dion! I agree that reducing the cost of the driver and power supply is the only way forward. I want the power supply and driver to be about 10% of the cost of the lamp, not 50% or more, that just seems wasteful. That is why using the acad is required. They are $3 and I can easily get one. Also, 90 watts is the perfect size for a lamp. The reliability, thermal, safety, damp-resistance of these things - perfect. Right now the only downside is 85% efficiency or so, but I will revisit that issue before I buy anything.

yup thats why i use $8 chinese drivers

OldGuyInOregon · May 22, 2015

For sure a Chinese driver would be less work, because I am still digging. If there is something commercially available that is better than what I am able to design, I will buy. If I don't do a high level design, I will never know.

I am at 94.2% efficiency on the current version of the driver. I have a few ideas about the acad, so I am not worried about it for now.

I think my circuit works, I just needed a nap and different transistors than I originally picked. That is good, because the diodes available with Vforward less than 14VDC suck.

OldGuyInOregon · May 22, 2015

I did not know any better, but I think if you want parallel diodes, you need more than two branches.

LED driver design question

Member

Active member

Member

Attachments

Active member

Member

Attachments

Active member

Member

Active member

Member

Attachments

Member

Attachments

Member

Active member

Member

Member

Active member

Member

Member

Active member

Member

Member

Attachments