New highest efficiency step down converter

[Klaus]

Anyone saw this yet - seems Kenji upgraded his SF-2

Kenjis SF2-2 review updated

Klaus

 

[georges80]

Nothing too surprising there - the LT1618 (in all the xxxboy designs) is not a particularly efficient switcher, especially when pushed to the limits (above about 600mA) in a boost circuit. It is even less efficient as a down converter. The one claim to fame for the LT1618 is that it has a current sense scheme built in and that makes it a nice chip for a minimum parts count current regulated design and it is quite flexible in its use as a boost or buck converter.

george.

 

[kj]

Klaus,

Oh, you found my update before I make a post on the review thread. What a fast /ubbthreads/images/graemlins/ooo.gif

George,

I understand electronics experts like you may not be surprised at this /ubbthreads/images/graemlins/wink.gif but, as a user, an efficient small converter is very welcome.

Just FYI, the parts count of the SDC is still small - there are seven parts and the board is 0.55" dia IIRC. Based on my rough measurement, the efficiency is about 90% when driving Luxeon III with 123x2.

(BTW, I'm interested in your converters too, but I don't have good hosts for them /ubbthreads/images/graemlins/smile.gif)

 

[BigHonu]

Does anyone know what the availablity of these converters are like?

thanks

Brian

 

[dat2zip]

Something doesn't add up here. The LT1618 is not that inefficient. Looking at the review page, the LED was changed.

If the previous LED tested with the LT1618 was a L bin for example and this one is a H bin, that could be 1/2W of power difference. That alone would offset any difference in efficiency. I'm not aware of the DB being that inefficient. My testing has shown that even at 1A it's around 80-85% efficient. So, something doesn't add up. By the chart theres more than a 50% difference.

Wayne

 

[Klaus]

Wayne,

I think you are certainly right that the LED is different as detailed on Kenjis page - but the "original" LED was built into the host by ARCmania/Mike Jordan himself and he always had an affiliation for lowest Vf LEDs so I don´t believe that the former one could have been an "L" LED, I would **guestimate** it was an "J" worst case, but probably ARCmania will comment and/or Kenji will post some more graphs where the same LED will get tested with different converters to have an equal playing field.

Klaus

 

[dat2zip]

I totally agree. Like I said, something doesn't add up. I just looked at Roys plot on a DB700 and it was over 2 hours of regulated light. I would expect the DB1000 to do something less, but, not much less. Efficiency difference between the DB1000, DB917 and DB700 are not going to vary dramatically.

No matter, MC converter board is still impressive.

 

[kj]

Wayne is right - the LED was also changed as I wrote in my page. I'm sorry this fact is not clear by the chart alone. I don't know what LED was used before upgrade, but compared with another chart where the K Vf bin was used, I guess the Vf was rather high... it could be "K" or "L". Assuming the Vf was around 4V with the L bin, the power consumption should be about 20% larger because now the Vf is 3.29V under load. Still the new converter seems efficient /ubbthreads/images/graemlins/smile.gif

 

[ARC mania]

Actually the bin number of the Lux III driven by a DB1000 is a TV0J. Light output could have been increased if a reflector similar to the one in the PM6 was used.

Arc mania

 

[kj]

Really? I would estimate DB1000 + TV0J runs in regulation for 60 minutes or so. My chart shows my SF-2 ran in regulation only 40 minutes before upgrade. It's a big difference. I believe my measurement is okay because I actually took the runtime two times and got the consistent results. Maybe that particular TV0J had a very high Vf at 1000mA? I'm not sure...

 

[PeterB]

My experience is, that the downboys drive the LED with much more current than the sense resistor setting would predict. I have seen something like 20% more current.

 

[McGizmo]

If the output current is comparable and the LED's are of similar Vf, this chart is implying a level of efficiency of less than 50% for the DownBoy?!?!?! That doesn't sound right. /ubbthreads/images/graemlins/icon3.gif Other than a sense resistor for 1000 mA being added, was there any other modifications done to the DownBoy?

 

[kj]

I built the Aleph LE with SDC and DB1000 and I took the runtime measurements. I used the same LED this time (i.e., I replaced the LED to measure).

I also measured the voltage and currents under load.

Super Down Converter
====================
Vout = 3.4V, Iout = 1.06A, Wout = 3.6W
Vin = 5.6V, Iin = 0.75A, Win = 4.2W
Efficiency = 3.6W / 4.2W = 85.7%

Downboy 1000
============
Vout = 3.41V, Iout = 1.08A, Wout = 3.68W
Vin = 5.5V, Iin = 0.855A, Win = 4.7W
Efficiency = 3.68W / 4.7W = 78.3%

I measured the currents by inserting a DMM in series, so the results are not so accurate and the actual efficiency is probably a little higher.

 

[McGizmo]

Nice work kj! /ubbthreads/images/graemlins/thumbsup.gif The lux looks pretty high there! Is it being measured at one meter or possibly less? For run time, I realize it is arbitrary as long as the distance is the same for all tests.

I think the SDC is a real winner for 2x123 and Li-Ion with proper Vf selection of LED! I believe the SDC has a Vin limit around 6 volts? There is always a compromise! /ubbthreads/images/graemlins/frown.gif

Although the DB is less efficient, its ability to accept 3x123 and 4x123 as well as 2xLi-Ion will likely keep it as a driver to consider. The SDC is really a well made converter and fits nicely in the LE! /ubbthreads/images/graemlins/thumbsup.gif Nice driver ArcMania! /ubbthreads/images/graemlins/grin.gif

With the LED's getting brighter and converters getting more efficient, we aren't through yet! /ubbthreads/images/graemlins/grinser2.gif

I have a runtime that I would love to post in light of the subject of this thread but it will be up to the designer as to when and if it will be revealed. Actually, since this driver is a buck/boost driver, perhaps it doesn't fit in with the subject or only half fits. /ubbthreads/images/graemlins/icon15.gif I realize that such a statement is obtuse and in fairness, I would add that this driver too has its potential short comings and is not a magic bullet. I am getting close to the line as it is in terms of disclosure here but I don't think the mystery will remain a mystery for long; at least I hope it doesn't. The real reason I even bring this teaser up is that I am becoming aware of a need to get more comprehensive documentation and complete parameters on these various converters. As more of these converters become available, it will become more difficult to navigate in the process of converter selection and the pros and cons will have to be weighed! As a builder, I don't look forward to the posibility of walking an uninformed person through their options! /ubbthreads/images/graemlins/banghead.gif

 

[kj]

I measured at one meter and I agree this lux reading is pretty high... and it seems a Luxeon lottery. I replaced this with another TW0H and I got approx 3500 lux, which is normal I think.

New efficient buck/boost converter? That's cool /ubbthreads/images/graemlins/popcorn.gif/ubbthreads/images/graemlins/popcorn.gif

 

[McGizmo]

kj,
Send me that abnormal LED and I'll clue you in on the new driver! /ubbthreads/images/graemlins/evilgrin07.gif Just kiding. In the run tests I have been doing with the Aleph 1 and drivers set to 1000 mA, I have been seeing lux in the low 3000's. Like you say, there is the lottery and it may be variations in the light meters as well. I don't air cool the flashlights and they do get warm! In some run tests, when I grab the head to feel the temp, the beam will jar and I will get a spike in the graph. /ubbthreads/images/graemlins/blush.gif

I have a 1000 mA driver of undisclosed nature in my prototype Aleph 3 currently and it has registered 6600 lux. Same LE in the Aleph 1 measured in around 3200. Although the spill is real good in the Aleph 3, at these high currents, the spot is so bright that the spill pales in comparison; especially if the target is not to far away. /ubbthreads/images/graemlins/icon15.gif

I would be curious to see what your abnormal LE would lux at coupled with a 38 mm reflector!!!

 

[dat2zip]

It's ok to post the graphs if you want. There is the caveot that the availability or more detail on the converter(s) may be released at a later time and date.

kj,

Great graphs and details. Using the same LED eliminates the LED bin code variable and help normalize the graph in comparing apples to apples. Thanks.

Wayne

 

[McGizmo]

Cool! /ubbthreads/images/graemlins/smile.gif

Here's a graph of NewWiz1000 with TY0J and 2x123 (SF 123's)

What I like about this driver is that the Vin needs to be between 2.75 and 6 approximately but with the buck/boost, no worries of matching Vf, drive current and Vin. /ubbthreads/images/graemlins/thumbsup.gif

Oh yeah,
Here's the same LE in an Aleph 1 head on a Pila 150S:

 

[kj]

Wow! It seems the efficiency is around the mid 90's /ubbthreads/images/graemlins/thumbsup.gif That's pretty nice and I'm very interested!

 

[Icarus]

/ubbthreads/images/graemlins/wow.gif /ubbthreads/images/graemlins/clap.gif When will this new driver be available? /ubbthreads/images/graemlins/smile.gif