4Sevens Quark Round-up Review: Q123, QAA, Q123-2, QAA-2 - RUNTIMES, COMPARISONS, etc

[GarageBoy]

I've been waiting for this. I REALLY wish it was more efficient than the Fenix, but pretty darn close

 

[selfbuilt]

I think that the LEDs with extremely high Vf are quite rare so with a dozen lights measured, the minimum could be not far from the population average (and it was the measurements intention, as I understood). One of the reasons that I think my results should be close to accurate is a comparison that I made to another light. As was expected, the Quarks had similar efficacy, except the points where the driver either hits the sweet spot (18lm mode) and the efficacy substantially rises or the points where the driver becomes inefficient and the efficacy substantially drops (moon mode and RCR for 123-2).

I thought that I made it clear that these are your samples by calling them "Selfbuilt's [battery type]" 🙂 I made these tables to verify a few things and the results were interesting enough that I thought I would share them. They show how extremely different the circuit efficiency can be in different modes and for different chemistries. This has a direct practical meaning for flashlight purchase and usage decisions.

Since the tables are predominantly based on your work I didn't want to take the results out of your thread and post them somewhere else - that's why I've put them here.

I'm sympathetic to your goals, and think there is value is your approach. The problem is we don't know how variable the Quarks are in their output (only David could estimate), but we should expect runtimes to be quite variable (based on all the combined factors that influence runtime, including even small changes in Vf). I really think you would need a much larger sampling of runtimes from different specimens to have full confidence in the final computed efficiency estimates.

That being said, I think the broad strokes description you gave above is quite likely true (i.e. 18 lm sweet spot, decreased efficiency at extremes). I agree it is very useful for people to understand these differences when making their choices. I am just concerned that the actual computed efficacy numbers may be magnifying so many sources of uncertainty as to be unwarranted to anything above 1-2 significant figures. I would prefer to let people draw their own conclusions from the individual output/runtime curves (where it's clear that it's just n=1 sample per condition).

FYI, I do appreciate you posting them first in this thread so we can discuss it 🙂. Please feel free to use my results (ideally in combination with other posted runtime data) as you continue to refine your methods and analysis. I would just prefer not to be directly cited as the data source in the title, as others may incorrectly infer that I performed or validated the calculation. I think something like "Wapkil's Quark efficacy computation based on CPF posted runtimes" would be a better heading than "Selbuilt's xxxx" . 😉

Why does the AA perform 1.5x as well on a 14500 as the CR123A does on a 16340, when they have the same electronics and the batteries have the same capacity?

In my testing of various AW RCR and 14500s over the years, I find 14500 typically has 25-30% greater capacity even though they are rated the same. This is in testing the cells with the same head in lights have multiple battery tubes.

But of course, that is just an average. I've seen a fair amount of variability within each type. If you are familiar with standard deviation, I can tell you the typical SD is about 15% of the mean value for the 6-8 cells of each type I have tested. I try to use batteries as close to the mean as possible, but it's not exact.

So when you factor in the variance introduced by testing different sample Quarks in the above test (noticing also that the Q123 is producing slightly higher output than the QAA), the ~50% increase with 14500 is within expected variability. 🙂

 

[wapkil]

I'm sympathetic to your goals, and think there is value is your approach. The problem is we don't know how variable the Quarks are in their output (only David could estimate), but we should expect runtimes to be quite variable (based on all the combined factors that influence runtime, including even small changes in Vf). I really think you would need a much larger sampling of runtimes from different specimens to have full confidence in the final computed efficiency estimates.

That being said, I think the broad strokes description you gave above is quite likely true (i.e. 18 lm sweet spot, decreased efficiency at extremes). I agree it is very useful for people to understand these differences when making their choices. I am just concerned that the actual computed efficacy numbers may be magnifying so many sources of uncertainty as to be unwarranted to anything above 1-2 significant figures. I would prefer to let people draw their own conclusions from the individual output/runtime curves (where it's clear that it's just n=1 sample per condition).

FYI, I do appreciate you posting them first in this thread so we can discuss it 🙂. Please feel free to use my results (ideally in combination with other posted runtime data) as you continue to refine your methods and analysis. I would just prefer not to be directly cited as the data source in the title, as others may incorrectly infer that I performed or validated the calculation. I think something like "Wapkil's Quark efficacy computation based on CPF posted runtimes" would be a better heading than "Selbuilt's xxxx" . 😉

I must say that I find your response disturbing. It is a rare occasion on a discussion forum when I have to agree with almost everything that my interlocutor writes 🙂

The tables were only another form of gathering and presenting data. Obviously, based on results of a single measurement they cannot have a defined statistical significance. My approach, I think, is similar to yours when you present detailed results of a single experiment, without determining how representative it is. I believe that for most of the reviewers here this is the only practically possible way. It is not scientifically sound but nevertheless quite informative when complemented with attempts to remove errors caused by faulty or radically underperforming samples.

The efficacy results depend on the battery capacity, the LED Vf and the driver efficiency, each being an independent variable. I thought about a way of introducing samples variability into the tables but I don't know how variable the components really are so I didn't know how to do it. Dependant on many factors, the efficacy cannot be used to precisely determine what to expect from a light but with more that sevenfold difference between the lowest and the highest it clearly shows some patterns. I hope that with an explanation how it should be interpreted, it may be valuable to some people.

I was focused on making it clear that these tables are not a result of my own research but merely a compilation of other people's work. You are probably right that the effect may have suggested that also the responsibility for the results lies with others. Of course this was not my intention.

Like you, I also expected the runtimes to be quite variable. Comparing the specification, your runtimes and csshih's test for QAA and Q123 I was surprised to see that they turned out to be so close to each other (QAA spec: 72min, QAA csshih's: 73min, QAA Selfbuilt's: 72min; Q123 spec: 48min, Q123 csshih's: 43 min, Q123 Selfbuilt's: 41 min).

The way the data was presented was a result of my laziness and the lack of time. I just took them as they were in the spreadsheet set to 2 decimal places cell format, without trying to make them more readable. When I find time, I will re-introduce the tables with the changes you suggested and an explanation how the results can be interpreted.

 

[Blacksidesniper]

Re: 4Sevens Quark Round-up Review: Q123, QAA, Q123-22, QAA-2 - RUNTIMES, COMPARISONS,

WOW!


It's good to get confirmation from the best reviewer around! :twothumbs

This best reviewer indeed! 😀

 

[selfbuilt]

I've done a little more testing on the 4Sevens CR123A batteries, and the results are interesting. I've posted this analysis in its own mini-review in the batteries sub-forum:

CR123A Comparison Review: 4Sevens, Titanium Innovations, Tenergy, Surefire, Duracell

Enjoy! :wave:

I must say that I find your response disturbing. It is a rare occasion on a discussion forum when I have to agree with almost everything that my interlocutor writes 🙂

Glad we are on the same page 🙂 - I look forward to seeing your revised tables. It is interesting that other reviewers are getting such consistent numbers ...

 

[Black Rose]

The Q123 differs in that the clip is permanently embedded inside the head (shown above). You can request a clip-less version when you check-out on the 4Sevens site. Personally, I find the clip gets in the way when changing batteries on the Q123, but YMMV. :shrug:

I had the opportunity to meet with Eric from 4Sevens.ca Friday evening and demo these lights. Very well designed and built lights.

I have to agree with you regarding the clip on the Q123 - it definitely gets in the way when changing the battery.

If it wasn't for the fact that the Q123 has such solid threads, I think it would be easy to cross thread the tailcap.

 

[HoopleHead]

Thanks for a stellar review, as usual. Much appreciated.

 

[Crenshaw]

im glad i pre-ordered a warm tactical now...😀

Crenshaw

 

[buraianto]

Am I reading the numbers wrong, or are the numbers in the key on the graph "2xAA Lo-Med Comparison: Sanyo Eneloop" not matching up with the graph? For example, Quark AA-2 (R2) Hi is labeled as 4hr 53 minutes, which is 293 minutes, but the graph looks like it's around 600 minutes.

 

[selfbuilt]

Am I reading the numbers wrong, or are the numbers in the key on the graph "2xAA Lo-Med Comparison: Sanyo Eneloop" not matching up with the graph? For example, Quark AA-2 (R2) Hi is labeled as 4hr 53 minutes, which is 293 minutes, but the graph looks like it's around 600 minutes.

Sorry, just fixed that. Excel sometimes deletes my x-axis scale reference when I'm customizing the graphs for individual lights, reverting to the sampling frequency. You will see the correct axis labels now if you hit your browser re-load button.

 

[buraianto]

Thanks for the update. Thank you for such a great amount of work you gave to us, Selfbuilt.

 

[selfbuilt]

FYI, I've just updated my CR123A battery shoot-out thread with "Med-Hi" runtimes on the EX10.

As you'll see, the 4Sevens cells hold up as well in that light as they do in the Quarks. Not sure why capacity is bit lower on my Olight T10 and LiteFlux LF3XT ...

 

[Gliderguy]

To revisit a subtopic introduced earlier in this thread, about the lack of lockout, I have two ideas:

Manufacture a compatible tailcap made of a non-conductive shell. The lockring and "guts" of the tailcap, of course, would be conductive so that fully tightened you would have a connection possible. I am looking at the thickness of the existing tailcap and wonder if any of the new polymers would hold up with that fine a thread without having to be really clunky to work?

And on review, an even better idea:

I guess a total internal redesign of the tailcap could allow the shell to still be aluminum if the lockring was MOSTLY non conductive except for a specifically designed contact for the battery tube that was isolated from the shell (just like the head works when you loosen/tighen the bezel. Actually, it could probably be a variation of the same circuit board blank even.)

 

[chadvone]

another great review, Ordered one. has anyone asked if they would lego with the fenix's ?

 

[Burgess]

Certainly won't mate with Fenix lights.


Didn't you read the posts about "Square Threads" ?


😎

_

 

[chadvone]

Thanks Burgess, I am sure I read that post.

 

[PlayboyJoeShmoe]

Gee thanks self built.

I've been saying I don't need a Quark JUST to get a moon mode.

Then I see the pics! Looks ever so much nicer and grippier than my beloved Fenix P2 or L1.

Problem is I can not buy one just as I can not buy an EZAA.

What I really mean to say is YOU DA MAN!

 

[selfbuilt]

Then I see the pics! Looks ever so much nicer and grippier than my beloved Fenix P2 or L1.

Sorry if the review is making you feel like you have to upgrade. But on the plus side, the performance of your P2D-Q5 would be pretty close on standard batts (except for moonlight mode, of course).

FYI, I have just updated the review with a couple of traces from a second QAA that I purchased for a gift.

(hit your browser re-load if you don't see the new QAA #2 traces in light purple).

As expected, there is a bit of variability in max output and runtime, but this is well within normal tolerances for all the components.

 

[recDNA]

Thanks for the great review. It's nice to actually SEE the numbers. Now I KNOW I want the 123X2!

 

[Ctrain]

Sorry if I've missed it somewhere... Surfing the net on a blackberry isn't the easiest 🙂 just wondering what the UI is like on the tactical versions? I like UI's where you can hide the strobe ond sos... Just curious!