My QB22a build and my home built driver. more efficient than meanwell?

Hey there folks, couple pics attached of my QB22a build. It's 15 strips all in parallel. Running my own home brew drivers
I've built 4 of these, and just wondering how the driver efficiency compares to that of meanwell drivers which everyone seems to be using.
So I'm running my strips at 11.05A , making 737mA per strip. binning current is 450mA so I'm running at 163% of binning current which if I understand correctly is just fine and dandy as rated current is 200% of binning current. You'll note they are well and truly over engineered as far as thermal management is concerned, each strip dissipates it's heat into a 1200mm length of 25x25x1.5mm aluminium extrusion - so compared to a quantum board it's running lower current per chip (I think, just going off the QB specs) with vastly better heat dissipation, which should equal significantly better photon efficacy (not sure how much better, please chime in if you know). The extrusions have a fan blowing across them and barely get warm to the touch, maybe 30 degrees C.

The driver consists of a 20A constant voltage power supply tuned up to about 55-ish volts which supplies two constant voltage/constant current DC to DC converters to smooth that voltage. those two converters are boost only and are set to constant voltage mode to supply a stable and clean 55 volts to the output stage.
You'll note a ferrite toroid hanging off the side, just to keep the noise down and keep the microprocessor in the output stage from freaking out. The output stage is a programmable CC buck converter good for 60 volts at 20A, so theoretically could run at 1200 watts, but there's no way I'd run it that hot. Never over 600w anyway as the output is only 45-48 volts or so. The key to efficiency here is not to boost voltages too much - DC to DC converters lose a lot of efficiency the greater the increase in voltage across them, so 53-55 volts in and 55-56 volts out and we're golden. You'll note they are hung up - that's to keep the weight off the lighting rack and for thermal management. all the heatsinks on the driver arrangement are fan forced.

The driver shows 501 watts, that's reading though the output and I'm showing 520 watts on the 220V AC side, so about 96% driver efficiency if my crappy math skills are holding up. Sound right? The driver efficiency drops off dramatically if I turn the output up to 600 watts, down to about 80%, but since I don't really need the extra light I keep them running low.
So if anyone's running similar power levels through a meanwell driver do you know the driver efficiency? I'd love to know how mine stacks up.

Also, if anyone knows, just how in the fuck do you tell which flux bin the chips in a QB22a are in? seriously this has baked my noodle something fierce. All I can tell from the data sheets is that the bin data is on the bottom left hand side of the information sticker on the PCB, immediately left of the QR code. All it says is S04 which doesn't match up with any of the LM301b bins at all. Confused.
Considering there is something like a 10% difference between the lowest SL bin output and the highest SM bin output you'd be quite reasonable to want to know which bin the chips on your expensive strip came out of.

Some things to note:

Photo quality is a bit poor, sorry about that. the room is packed so full of gear it's hard to get in there and get decent photos and my phone camera is a POS. but you get the idea.

The quality of my wiring work is poor. I'm aware of that. Hard to believe I've been wiring stuff my whole life for a job looking at that, but it's basically a prototype. If it works out I'll do it proper but as long as it doesn't catch fire in the mean time there's not much point being meticulous with it.

Please let me know your thoughts, all criticism is welcome and appreciated.

Kind regards
TGS

I find 96% very hard to believe and suspect the accuracy of the DC power meter, some are known to be poor. I would double check it with a DMM. What was used to measure the AC power? I would expect both the DC supply and the boost converter are under 96%, and the losses stack up. What's 90% of 90%? Why add the boost converter, rather than just running off the CV supply? Why two circuit boards, they both appear to be DC-DC converters? I'm kinda curious what ya have going there.

I have some very similar boost converters, but don't recall the efficiency, I think it was low 90's. I have a couple EB strips running off a cheapo CV supply ATM, maybe I'll connect a boost converter and take some measurements.

1212ham said:

I find 96% very hard to believe and suspect the accuracy of the DC power meter, some are known to be poor. I would double check it with a DMM. What was used to measure the AC power? I would expect both the DC supply and the boost converter are under 96%, and the losses stack up. What's 90% of 90%? Why add the boost converter, rather than just running off the CV supply? Why two circuit boards, they both appear to be DC-DC converters? I'm kinda curious what ya have going there.

Click to expand...

Hey thanks for your comment, I was wondering if someone who knows DC to DC converters was going to call me out on the efficiency. It seemed unlikely to me also, and the thing is the AC measurement was just taken from an inexpensive plug in power monitor which is dubious in it's self. To add to that the harmonics coming off the output from that switched mode AC/DC CV supply are horrendous, so I'm guessing (without an oscilloscope) that there is a fair bit going back into the 220VAC supply too which probably isn't doing that power monitor any favors.
I do have a very accurate Fluke DMM which is actually certified as calibrated, so I'll put that on the AC side and see how much current it's actually pulling and get an accurate voltage measurement too.
I'm confident the DC voltage, power and current measurements shown on the output are accurate, I've tested that programmable buck converter and comments from other people who have used that device seem to suggest it's measurements are extremely accurate. It's actually a lab power supply so you'd expect accuracy from it.

The reason I used three stages is because of those harmonics, The AC/DC CV supply wrecks havoc on the microprocessor in that final stage without some kind of filter on it and when I built that thing I didn't have any ferrite toroids. I've got some on order and when they show up I'll take that second stage out of the circuit, hang a bunch of toroids on the output of that AC/DC CV supply and see if the microprocessor holds up. If it does I can use those second stage CC/CV units for another project (they were surplus anyway).
All this sounds like reasonable justification to have a few beers, get on Aliexpress and use the whole situation as an excuse to drop a bunch of money I don't have on the oscilloscope I've been dreaming about. Hell, since I'm putting myself in debt why not get a PPFD meter at the same time

Scopes are cheap.
I have one of these, it went from crap to impressive after installing the BenF firmware v3.64.
https://www.ebay.com/itm/DSO201-Pocket-Handheld-Digital-Storage-Oscilloscope-DS0201-Open-Box/303119776431?hash=item469358baaf:g:ACMAAOSwkCZcqgpI&LH_BIN=1

I'd buy a newer version, there are several of this brand. https://www.ebay.com/itm/DSO201-Pocket-Handheld-Digital-Storage-Oscilloscope-DS0201-Open-Box/303119776431?hash=item469358baaf:g:ACMAAOSwkCZcqgpI&LH_BIN=1.
https://www.ebay.com/itm/DS212-Smart-LCD-Digital-Oscilloscope-USB-Interface-1MHz-10MSa-s-Coupling-AC-DC/352126333400?epid=10004573686&hash=item51fc5d71d8:g:JFEAAOSwW6RcnGpH&LH_BIN=1

As for the efficiency, your not going to get 96% going through three circuits, no way. It's been said that the Mean Well HLG's aren't even that efficient. Why not just run off the cv supply and avoid the losses in the additional circuits? I'm not giving you a hard time, but that's just a cheap no name supply from china, the word "lab" in the description means nothing. Maybe it's defective or damaged?

Just get a lux meter and you can buy a lot more beer. It's easy to convert lux to par with some accuracy.
https://www.ebay.com/itm/TA8131-Dig...323769?hash=item4689020d39:g:WTAAAOSwKb1bNuZ1
https://www.rollitup.org/t/slow-yellowish-growth-under-vero-29-3000k-3500k.962065/page-3
https://www.rollitup.org/t/how-many-umols-for-fllowering.972399/

S04 isnt lm301b I dont think they are the same binning numbers are they? That's lm561c s04 binning. I dont think the binning for lm301b is the same. Lm561c has s04, s5, s6. Not sure s7 ever came out. Or s6 are the highest. And you can get lm301b+ and the cri will make a difference. The 4000k 90 cri lm561c matches the 3500k 80cri lm561c on the par curve charts.

1212ham said:

As for the efficiency, your not going to get 96% going through three circuits, no way. It's been said that the Mean Well HLG's aren't even that efficient. Why not just run off the cv supply and avoid the losses in the additional circuits? I'm not giving you a hard time, but that's just a cheap no name supply from china, the word "lab" in the description means nothing. Maybe it's defective or damaged?

Click to expand...

Yep, turns out I was dreaming about that 96% figure.
Good news is, the lab power supply display is pretty accurate. Bad news is the el-cheapo plug in type power meter is a bust if there's a reactive load or anything that generates harmonics on it. works fine when you plug the kettle into it. I know exactly what my morning coffee cost me.
So the Fluke says 3.951 amps and 227.4 volts on the AC side, and the output is 45.81 volts drawing 11.03 amps. making an eye watering 55.2% driver efficiency. Power meter goes in the trash.
I do have one of these lighting rigs set up without the intermediate boost converter, I get a few minutes run time out of it before the harmonics off the CV power supply makes the microprocessor have a fit (the display blanks out, but it still runs) so I'll check that one just to see how much more efficient it is.

When you say "run off the CV supply" are you suggesting connecting the LEDs directly to the CV supply? I don't think I'd be comfortable doing that. Maybe if I were only running them at binning current, but a small variation in voltage has the potential to get these things very hot without some kind of CC control.

whytewidow said:

S04 isnt lm301b I dont think they are the same binning numbers are they? That's lm561c s04 binning. I dont think the binning for lm301b is the same. Lm561c has s04, s5, s6. Not sure s7 ever came out. Or s6 are the highest. And you can get lm301b+ and the cri will make a difference. The 4000k 90 cri lm561c matches the 3500k 80cri lm561c on the par curve charts.

Click to expand...

Yeah it's weird man, according to the samsung data sheet for the QB22a they only use lm301b diodes, but that binning code doesn't seem to appear anywhere in the lm301b data sheet. wtf.

TheGreatSouthern said:

Yeah it's weird man, according to the samsung data sheet for the QB22a they only use lm301b diodes, but that binning code doesn't seem to appear anywhere in the lm301b data sheet. wtf.

Click to expand...

I think theres only two lm301b diodes. Lm301b and lm301b+

The s04 refers to length and number of diodes I believe. Theres s02, s04, then it goes to L04, L06, and L09. All my strips with lm301b are lm301b+ which are good for up to 220lm/w the lm301b is 202lm/w

My H inFlux_L06 strips are lm301b+ it said it on the info sticker on the static bag. They are single diode row of 112 diodes I believe. 560mm length. The L09s are double row.

Although the datasheet shows as an example S01 binning. But then shows ZZ~00 so you might be right. But my H inFlux_L06 dont show an actual binning like that. They said LM3o1b+ binning. And if that's the case then the LO4 is the binning.