Review: EagleTac D25A Ti 219 (mid-2014 botch, er, batch)

They said I’d need to return it via the store where I bought it… which I was hoping to avoid, since GoingGear has been great and I prefer not to burden them with someone else’s problem.

I find that Zebralight’s moon modes measure below spec even on their own scale, but I still find the output to be at a useful level. The highest moon mode on my Zebralights is actually what I’d consider the ideal level of output, even if it’s only 0.1 lm instead of 0.3 lm. I can’t complain about the runtime, either… one of mine gets used like 6-8 hours per night as a night light and only needs a recharge once every month or two.

The human eye has a much wider dynamic range than most cameras, so I usually take three exposures to show what the human eye can see in just one viewing.

As for the blue ring, I’ve found it distracting. While using the light indoors, I can see the blue ring as far as 5 meters away. This is generally the case when I’m looking down a long room; the walls next to me have a bright blue crescent on them.

The blue ring goes away if I add diffuser film though, which I intend to do once I have other issues sorted out. It seems to improve the tint a little too. I’d really like to see the beam without the lens though.

This is a light-orange-peel reflector. I don’t see the x-shaped beam pattern on any of my smooth-reflector lights, or on any of my other orange-peel lights. Actually, I don’t see it on any of my other lights regardless of the type. It’s not a big deal though; just a tiny quirk which makes the D25A unique. Its light cone has the narrowest point about 8cm in front of the lens, instead of at the lens itself like most lights.

This D25A has full-on PWM, not high-frequency noise or oscillations.

My ZL SC52 has high-frequency oscillations on its L1 mode, and my JB RRT01 has noise/oscillations on several of its levels. I can see it, but only when I’m specifically looking for it with my usual PWM test. Those two oscillate quickly back and forth between two nearby brightness levels in order to get an average output somewhere in-between. The D25A isn’t like that though. It strobes completely on and off in a manner which looks identical to the common, cheap nanjg driver at 4.5 kHz… and I see it even when I’m not looking for it.

FWIW, the reason I see the 4.5 kHz PWM is because my eyes are weird and don’t do “saccadic masking” very well. This video explains more, if you’re curious. Basically, the strobing catches my attention every time I move my eyes while using a PWM-based light to look at stuff. Instead of the usual blur I’d see while moving my eyes, I see a bunch of discrete frames, especially on reflections, and it’s distracting. Also annoying when I encounter PWM-based LED tail lights on the road.

My D25A definitely has a Nichia 219B in it. It looks different than my 219A lights, and identical to my 219B lights.

All my Nichia 219 lights are spec’d at 4500K 92CRI, but the 219B lights appear to be actually around 5000K. I looked for others with 219B emitters, and they mentioned having similar results; the 219B we’ve all gotten has been significantly cooler in tint. So, I figured the D25A would be around 5000K too, especially since that’s what EagleTac said in their spec. But it’s the warmest of all my 219s, roughly the same color temperature as my ZL H52Fw (spec’d at 4400K), warmer than my 219A lights (about 4500K), and cooler than my ZL H51w (spec’d at 4200K). I don’t have a lot of lights in that range to compare against though, because I prefer slighly cooler tints.

I don’t have a way to measure CRI, but visually the D25A 219 doesn’t make colors “pop” anywhere near as well as my other Nichia 219 lights. It looks roughly on par with the ZL H52Fw, except that it’s more yellow and the H52Fw is more pink. I prefer the pink hue since I don’t see red quite as well as green or blue and the tint corrects for the bias in my eyes.

That is actually quite possible. If I recall correctly, the N219B has a bin available with significantly lower Vf, and if they used that bin it might look quite a bit brighter at the same voltage. But others reported really bright moon/low modes with Cree emitters, so I kind of doubt it. Apparently the anomaly isn’t this unit… probably the entire mid-2014 batch.

How about 2.9 lm for 96 hours? That exists… it’s not quite as bright or quite as long, but it’s available today.

It’s gratifying to see that not everyone buys the PWM-is-just-circuit-noise story. If it was just noise, then my camera couldn’t see it.

Here’s a video I did for the D25A mini showing PWM, which on their web site they say is constant-current. I believe what happened is that at some point EagleTac stopped using constant current circuits to get better tint, which their customers were/are complaining about. And they neglected to update all of their spec sheets—best of both worlds amirite?

It’s really easy to see PWM. Just turn the light on and wave a thin, stiff white sheet through the beam really quickly. For this purpose, I’ve been using a pupil distance measurement ruler from Zenni Optical. It seems almost ideal — thinner and stiffer than a business card, long and narrow, and white.

Here’s what it looks like when checking PWM on my BST-wide on “low”. Its PWM runs at 188 Hz (measured via sound tuner plus the audible whine it makes sometimes), at a duty cycle of 11% (measured via pixel measurements of that photo, and verified against official spec):

The D25A looks similar except that the frames are thinner and much closer together. It’s difficult to get good pictures of it, but I could probably do so if necessary. The D25A’s frames are only about 1mm to 2mm apart, which is consistent with all my nanjg-based lights. The nanjg PWM runs at 4.5 kHz, so I’m assuming the D25A does too.

Copying quotes down is too difficult, so…

I know everyone has different moonlight mode preferences, but if ZL had been remotely accurate, I wouldn’t have wasted my time with the SC52. My H51w (0.18) was nice, and I was looking for brighter, but then rec’d something spec’d twice as bright (0.34) yet with less than half the output - go figure. It also doesn’t make any sense to me to have three ML options, but all of them below 0.1 lms.

We can agree to disagree on the blue and X, just offering another opinion - I can’t see either in normal use.

I’m the same with PWM, it catches my attention in reflections when moving my eyes quickly (the way I prefer testing for it) and I also find it distracting. I ran the time exposure sweep picture I mentioned before and now agree it does look like full on PWM, L > R, SC52, Quark AA, D25A, MDC AA, ~3 lms:

I thought it might have been some sort of oscillation since this is THE fastest PWM I’ve ever seen - I can barely detect it, with the reflection or pencil fanning method, and I’m sweeping those lights in the photo as fast as I can. I don’t know the Hz, but the MDC and 47s Mini are said to be HF PWM but can I easily detect those. Also THIS is an old photo of my ’12 D25A vs. a Quark on ML where the light goes “pseudo-PWM” on low voltage batts (cellphone detection method included)… it never really is fully “off” so is not truly PWM… the Hz is very slow though and quite annoying, and another reason I think the D25A has poor regulation on ML.

Toykeeper,
It was not my intention to be confrontational, but I felt compelled to question your review based upon your calibration to ZL lumens (sorry, but what I consider to be one of the most exaggerating and inaccurate starting points) and comparing that to what I found to be one of the most accurate (well, at least my samples). But you know, and have well stated, all the key points (ie, ZL scale 40-50% more liberal than ET scale, L2a ~0.1, ML below spec even by their own scale) and appear to have well accounted for them. So between that, the “B” emitter switch, and PWM points (news to me), I wave the white flag and say “Uncle!” I think you have done an excellent job on the review, and look forward to your future reviews.

My apologies for any troubles I may have caused.

Racer,
I used to use the cellphone camera method to detect PWM but have picked up some false readings, and in both directions. For example, HERE is an old pic of my MDC AA on the same mode. And I also don’t believe in the current regulated “tint shift” thing - THIS pic shows all lights at ~100 lumens on top and moonlight on bottom - can you tell which lights are current regulated and which are PWM?

That’s a really interesting picture. First time I’ve seen the SC52’s L1 oscillations captured on film. I haven’t tried to get an image of it because it seemed like it’d be very difficult. You must be very quick! :slight_smile:

I’m also going to have a hard time capturing the Qlite driver’s PWM in an image, since it runs at 19 kHz instead of the usual 4.5 kHz. It’s fast enough that I don’t notice it during use, and can barely even see it with the card-waving method.

However, I just today discovered that I can build a spectrometer from inexpensive parts, and I’m really excited about it. I’ve been wanting to map the spectral output of lights for a long time, and sometime soon maybe I can. :slight_smile: (especially interested in comparing a N219 against my BST-wide and the sun… but it’ll also be neat to see how the D25A 219 compares against a N219 without the AR coating)

FWIW, I discovered two things today:

First, on a light with a similar blue halo (Olight S10, 2012 edition), removing the lens gets rid of the halo and makes the beam like 500K cooler and much whiter in color. This mostly confirms that I should be able to fix the tint and CRI issues by swapping the D25A’s lens for a plain piece of glass. That is, assuming I can get it apart.

Second, I got a slightly older aluminum D25A today (it was on a really good sale), and it has some major differences. Its moon mode is only a fraction as bright as my titanium model’s moon mode. Also, the lens is actually user-replaceable like the specs say. The front bezel comes off easily to expose the lens and reflector and LED. The head and body threads also swapped gender, and the head is narrower than on the titanium model. It seems the titanium version has a completely different head design than the aluminum model.

The aluminum model still has PWM on some levels though, and its moon mode has an initial pre-flash before it steps down to the correct level. But overall it’s much closer to spec.

I updated the OP with “Regular D25A Measurements”. I measured another D25A, a cool white XM-L2 model that I think is from 2013. It’s much closer to spec.

Tint: Not sure, blue-ish white with a primary blue halo around the outside of the spill. So, definitely cool white.

Lumen specs are listed as OTF lm / emitter lm, since EagleTac lists both on their site.

D25A-XML2 low / “moon-mode enabled” group:

  • Low (spec=0.5lm/1lm): 0.43 lm, current-controlled
  • Med (spec=9lm/11lm): 12.1 lm, with PWM
  • High (spec=85lm/121lm): 121.9 lm, current-controlled

D25A-XML2 high / “moon-mode disabled” group:

  • Low (spec=6lm): 5.86 lm, with PWM
  • Med (spec=20lm): 23.62 lm, with PWM
  • High (spec=85lm/121lm): 121.9 lm, current-controlled

Turbo:

  • Turbo (spec=141lm/200lm): 201.4 lm, current-controlled

This was, mostly, pretty close to the “emitter lumens” values specified on EagleTac’s site. Their “OTF lumens” specs seem low though, compared to the ZL scale my light box is calibrated to.

I was actually a bit surprised at how closely the measured output resembled the specs. Sure, it mostly matches the “emitter lumens” instead of “OTF lumens”, but at least it matches some part of the spec. And apparently ET’s “emitter lumens” are about the same as ZL’s lumen specs. Just FWIW. I mean, the numbers match up really closely, if you ever need to convert from one to the other.

I’d love a D25A Ti-219 with the lumen levels of the XM-L2 aluminum model (maybe a bit lower on high/turbo though), the titanium body, and a plain piece of glass instead of the AR-coated lens. Sigh.

Hello, Toykeeper… glad you found a decent sample to test. Your results appear consistent with my tests between the two lights and also with your statement in your OP that ZL and ET differ by ~40-50% in lumen scale. As you know I consider ET’s specs to be more ANSI accurate for reasons I mention in post 21 above, and because the ET/47s scale matches most of my light collection. The D25A has been one of my preferred lightbox calibration lights.

- I think with moonlight mode off, Low and Med spec should be 3 lms and 18 lms since the ET footnote say “6x and 2x, the L and M with MM-on.”

- Curious what voltage cell you used to test the ML mode since I find the relatively poorly regulated ML to be quite volatile depending on V. ie, I get half the ML output on <1.3v NiMh as >1.6v Alks or L91s.

- Might be interesting to run a side-by-side output/runtime test of the D25A 85 lm/2.5 hr mode vs SC52 116/3 hr mode.

- Interestingly, I have not read of any other owners of an N219 D25As having massive out of spec low modes… on the other forum that is. Perhaps you really have received a bad sample?

Thanks for the update!

ToyKeeper, you said in the the review that your 219B in D25A ti is a little bit warmer or yellowish than your other 219Bs. I just read from the datasheet (if I understand it right) that they have cool and warm tints too. I just bought a 219B from IOS and it has been described as “D220” in the specs. It does have a little bit of yellow but I don’t mind that because I’m used to warmer tint now. If you can see from the table lum. flux of D220 is min = 220 and max =240. So, I guess the higher “ranks” (like D260, D280 and D280) will have cooler tint almost to white and that’s why the lum. flux is higher too.

The specs say: “With moon-mode setting disabled, lowest output will increase by about 6 times, to 4 lumen, and the second lowest output will double, to around 20 lumen”.

Even though the multiplication doesn’t work out very well, I used the numbers from ET’s specs.

Fresh-off-the-charger Eneloop. I don’t have alkalines on lithium primaries to use for testing.

It might be, but I’m not doing runtime tests. I enjoy selfbuilt’s runtime graphs, but unfortunately he didn’t test SC52 vs D25A… he only tested SC52 vs the 2xAA version (D25A2).

Something is clearly wrong about the ET specs. It’s a 219B but they claim it’s from the “B11” bin which doesn’t exist. That’s not even the correct naming scheme. It should say D220, which is the brightest bin available for the 92-CRI emitter they claim to be using. Then again, it should probably also say it’s a 4500K tint instead of 5000K tint, but this particular bin of 219Bs seems to be noticeably colder than the equivalent bin of 219As, so I can understand claiming it as 5000K instead of 4500K.

In any case, the yellowish tint isn’t normal. Anything with 90+ CRI at 4500K should look white, not yellow. I have ten lights with Nichia 219s in them, and the EagleTac looks rather different than any of the others. I have three lights with too-strong AR lens coating, and I can remove the lens from the other two. All three make the beam warmer and color-tinted, and all three look quite a bit better when I take the lens off.

So, I’m pretty sure the tint issue is the lens, not the emitter.

It occurred to me just now that I can actually test it; no need to guess. I took the removable lens from my regular D25A and held it over another Nichia 219B light, with a third 219 shining next to it for reference. When I shine the white 219B through the D25A lens, the beam turns warmer and yellow and colors no longer look nice in it.

I have seen similar results with another lights’s AR lens.
I had a Nichia 119 that looked slightly yellow with lens and less so with lens removed.
Also removing reflector got rid of all yellow.
I’ve noticed the coatings on the lens as well as reflectors can change tints.

The “around 20” statement seemed to me to be the averaging of 18 for the XML and 22 for XPG. OK, got the 4 lms, but then why do you use 6?

No big deal - it’s your call of course, but those two modes are the only ones that stand out as inconsistent to the rest of your measurements & lumen scale differences.

’cause I R dumb and sometimes I need things pointed out more than once before I “get it”. D’oh. Fixing the OP now.

I’m glad you identified the problem. I just found the correct datasheet for high CRI 219B (min 85). Here. You’re right D220 is the brightest bin available for high CRI. Do you know why I see some yellow with my 219B D220 from IOS? The lens is without coating.

It’s hard to say. Individual emitters vary quite a bit, and the bins Nichia uses aren’t as narrow as the bins Cree uses.

I’ve seen at least 300K of color temperature variation between different 219B emitters in the same bin, maybe more like 400K. And next to a 6500K cool white emitter, if your eyes are calibrated so that 6500K looks white, anything in the 4500K to 5000K range will appear warm and sort of yellow. OTOH, if your eyes see 4500K as white, the 6500K tint will look rather blue.

Next to other emitters of a similar color temperature though, a high-CRI light should look pretty white. Do you have any other 219s or other high-CRI neutral/daylight lights to compare against?

A smooth reflector basically mirrors the emitter from the center outwards, so you are seeing the corners of the chip pointing towards the center of the beam, creating a dark X. It all averages itself out at distance, since ideally, your reflector will image to a point infinitely far. In my opinion, I almost think that seeing a dark X up close is a better sign than seeing a fully filled circle up close.

Perhaps the lens isn’t coated for a broad range of wavelengths, but maybe just coated for wavelengths centering around blue? AR is for increasing transmittance, and most typical high output LEDs are on the cool side, so if you had to pick one coating, you’d pick one for a blue wavelength.

Also, these may be truely current regulated, but thought I might posit the possibility that even though they are a smaller chip, perhaps they are more efficient at lower currents than the driver (or flashlight maker) is expecting, and perhaps this could somehow result in more light out for the same amount of power to the LED.

It certainly looks pretty white next to my T6-4C and T5-5B (of course it is much warmer) while the only other light with 219B I have right now is the L10. The L10 certainly a little colder (whiter) than my 219B from IOS. I guess not by much, maybe just my eyes. It could be the driver or reflector that caused that yellowish tint.

I don’t think a flat AR coated window will do much splitting of colors. A large focusing lens will. The way the colors get separated in reflector lights is that the light emitted sideways that hits the reflector has had a longer path through the phosphor, so it has less of the original royal blue or whatever left in it than is in the light emitted forward that forms the spill. That can be an advantage with a typical bluish LED because the spot is warmer and the spill is often too dim for the tint to matter much. But the bluish spill blinds other more than a warmer spill would.
On the other hand, I don’t see much advantage to anti-reflection coated lenses in flashlights, unless you want a very clean ringless beam.

I verified it on another D25A, and on an Olight S10. Stock AR lens equals nasty tint with a blue halo, no lens equals beautiful beam, plain piece of glass also equals beautiful beam. It’s not the reflector. I think this is also why so many ZebraLights have green tints, but I can’t remove the lens to test those.