I just received my LM4 and tested 3 lights, plus ambient:
First impressions, Opple 4 ambient CCT and DUV taken at very low Lux is unreliable
On the three lights the
DUV is 20 points lower on Opple3 than on Opple4 on two cool lights, and 30 points lower on the warm light,
CCT mostly close on both Opple 3 and Opple 4,
CRI Ra approximately similar on both Opple 3 and Opple 4,
CRI R9 seems to show relatively useful values for comparison purposes… for example, the XM-L2 has much lower R9 than the High CRI lights… although I think it is actually lower than measured. Normally Low CRI lights have Negative R9, considerably lower than the 18.7 the Opple 4 reports.
===
here is a simpler data dump… from a single light w NW XM-L2:
first impression, the
R9 of 78.5 from Opple 4 is dead wrong it should be negative since this is a Low CRI LED…
here is an example of an XM-L2 that maukka measured at R9 -31:
Sad because I was hoping the R9 data would be useful. That was my main reason for buying it. My other reason was to get the precalculated DUV, but it is also wildly inaccurate with my CW TS10… so… foggedaboudit
That’s unfortunate that the R9 and DUV appear to be so inaccurate. Those features were also what I was wanting most, especially the R9 reading. Glad I resisted the urge to buy one.
On the hypothesis that the issue is with CW LEDs, I tested an sw45k:
the numbers seem less wildly different… the R9 of 80.3 is in the balpark for a 9080 LED… When maukka measured an sw45k he got R9 of 97, but it is not the same LED: His DUV value of -0.0055 is much different than the two Opples I tested… though again, not the same LED…
pic is a link to source:
It does seem like the Opple 4 might have more severe issues with cooler lights, and low cri lights…
hard to be sure unless someone does a side by side w a sekonic, for the same LED…
Probably need more data. It’s looking like there are inconsistent results where calibration might not help, unless each device is calibrated. The same correction factors might not be applicable to different units if they’re all different.
The documentation describes different approaches to calibration. Without going into the details: Measuring a prototype and calculating the matrices from it for all chips doesn’t give good results for all possible measurements. The R-values might never be entirely ok.
Based on my observations so far, I suspect that R9 measurements from low CRI lights are not reliable at all. Perhaps the equipment is incapable of displaying negative R9 values, resulting in strange numbers.
However, I am not very concerned about inaccurate R9 measurements for these lights. I already know they have poor CRI. For these lights, I don’t really care whether R9 is high or low. What I am more concerned about is whether R9 measurements for high CRI lights are useful. If they can give me ballpark estimates of the quality of high CRI lights, then the new Opple can serve its purpose.
The Opple is less than one-tenth the price of low-cost calibrated equipment. I think comparing measurements from such cheap equipment is not going to end well. I own both the G2 and G3, and the numbers vary widely. However, they are both quite useful because they give me consistent guidelines across different lights I own. While the fact that they are not calibrated means that the absolute number may not mean much, comparing measurements of different lights certainly gives me useful information.
I have now re-tested several times with nearly all of my lights, with tripod and simultaneous reading in dark room. I will post later more details of my amateur findings, which basically haven’t changed since my very first “oh shit” post.
LM4 = “Biphasic” nature: 1. “Ok” with High CRI, low CCT: important parameters CCT, Ra, Duv’s appear OK. Duv tend to be more positive than my two Opple 3’s which themselves tend to be too negative. A “reasonable” trend is perfectly fine for me in this low cost device. 2. Really bad with low CRI, high CCT >5500 (with some exception)
Wild numbers, will provide details later. No change: crazy R9 100, crazy low Ra, wildly negative Duv, etc.
LM3 = if you see one, buy immediately
Now that I look even more closely: Very impressive how my 2 are tracking each other. The x point actually reads nearly identical - amazing for a $30 device. The y point are different but forming a fixed pattern, one LM3 has larger Y than the other. Y point on the graph affects Duv, so the one with the higher y value reads more positive Duv. But again, it’s a fixed pattern, which makes all the difference.
My 4 hasn’t arrived yet so I haven’t fully formed an opinion. I’ve had the 3 for quite a while. Remember what these were originally made for. That is helping interior lighting specialists and designers. I doubt that opple is testing these things with flashlights at 1 meter. So it’s not going to bother me even a tiny little bit if it is not accurate above 5000k. And 6000k is just plain *;#@√§. I still understand why Duv is important to some people at 5 and 6000k. For me whether it’s a tiny fraction above or below the line once you get below 4000k I don’t think really matters. I still look forward to testing it and maybe my views will change. The last lights that I will test will be the 5 and 6,000 k lights that I never use.
agree, Low CRI (left) gives wrong R9:
In this Low CRI example the DUV difference between the two lights does seem somewhat representative (though duv 0.0144 seems exaggerated). otoh for the High CRI light on the right, the Ra and R9 and DUV seem realistic.
otoh, the Opple 4 might be more useful for comparing High CRI lights:
Opple 4 seems quite accurate for the TS10 CCT, Ra, and R9… imo the DUV reads high… but this is one of the most credible comparison Ive done so far… both lights set to 65 lumens and measured on Opple 4 at similar lux…
(I hope someone will do an Opple 4 comparison to a Sekonic or Xrite)
219b 3000K with IDENTICAL DUV results from both Opple 4 and Opple 3… wow!
light was set to 9 lumens, and just 1.5" away from sensors.
519a 5700K
Opple 4 DUV reads 0.0032 higher than Opple 3. otoh, CCT reads almost identical, while CRI R9 reads about 4 CRI higher on the Opple 3:
light was set to 100 lumens and 5.5" away from sensors.
Preliminary results (may change), posting for the sake of comparison. Lights on tripod in dark room, about 15 ft away from sensor. Remember this is my lights, my LM4 and LM3’s, so YMMV. No big deal, just post your numbers so we could compare and learn something. 3 groups that I could ID:
High CRI, sub-5500K CCT - that tested “normal.” Or relatively so.
All lights with Nichia 5000k or below, all lights with GT-FC40 5500k and below.
The Duv numbers are more positive vs my LM3, but YMMV !!
Low CRI, above-5000K CCT - that tested “normal”
Wurkkos TS32 with SFT40 (why here and not in IF22a? I don’t know)
JKK76
Haikelite HK05
Low CRI, above-5000K CCT - tested “odd”. Not all, but any combination of
R9=100
Ra oddly low or blank
Duv outright negative, or too negative in lights with known green LED
CCT reads approx. 300 high
(Too negative, too high, or too low etc. are vs numbers from other users, reviewers, specs, LM3, etc.)
TS30S 1st Gen w/ SBT90.2
TS30S Pro w/ SBT90.2 (Duv positive, but only 0.0020 for a very green LED)
IF22a w/ SFT40
Nitecore MH12S w/ SST40
Nightwatch Chaos
Mateminco PD90S
Nightwatch Super Valkyrie
Lumintop D3
I noticed the dubious R9 value on measurements others posted, but was waiting for more measurements to be shared before drawing conclusions. The other measurements I looked seemed at least in the very rough ballpark - I think someone shared an R9 around 20 measurement on an SBT-90 (a bit high), and R9 above 80 on a 519A (about right)
But I agree an R9 of 78 from an LED with an Ra of around 60-70 is wildly off.
The difference of a reported Ra of 61.2 from the Opple 4 compared to 71.1 from the Opple 3 also suggests much worse Ra accuracy than I was hoping for, even recognizing the fact that this is a budget instrument. I had the impression that in the few instances when Opple 3 users were able to make comparison to proper spectrometers, the results were usually only about 2-3 points different.
I am hoping this is just a correctable bug with this first batch. It seems like the new sensor with more color measurement points should be capable of improved accuracy over the Opple 3.
On a separate tangent, it’s interesting to me that Opple is continuing to use a discrete channel sensor. I understand these sensors are cheaply available, but it seems to me that a refinement of the classic DIY spectrophotometer design that relies on a diffraction grating and an imaging sensor could be possible at low cost:
I would have thought this technique could be refined enough to make a commercially viable product capable of near continuous spectral measurements using low cost components like the camera modules used in entry level smart phones. I think the main challenge is calibration. Perhaps that’s a much more difficult challenge than I realize?
agree… now that I understand the Opple 4 better, it seems clear that bogus R9 is only an issue with Low CRI LEDs, and the giveaway is that low CRI Ra that produces High CRI R9 is a symptom of Low CRI LEDs… Basically the Opple 4 cannot give accurate R9 for lights with low Ra.
otoh, with high CRI LEDs, the issues are less critical… the Opple 4 reads higher DUV by about 0.0030, than the Opple 3, but the Opple 4 adds R9 data, that I find valuable.
Both the Opple4 and Opple3 give pretty close CCT values, when testing High CRI LEDs at similar lumens output, and similar lux on the two Opple meters. I find that gratifying. See the TS10 example above as a prime example.
so, based on my last few tests, imaged above, I do think the Opple 4 has useful application for comparing High CRI LEDs to each other.
For example, Im looking forward to comparing the R9 of an LH351d (coming in the mail), to a 519a… I want to confirm that the LH351d has lower R9, and only the 0pple4 has R9 sensors. The Opple3 has no R9 data.
This last observation will cause me to return LM4. Please check your finding to see if it is same as mine. To check this you must have both LM3 and LM4 and 2 phones.
I experimented with my S21D 519a. Light on tripod, sensors fixed in position 15 ft away next to each other. No change in position or direction of either light or sensor during test.
x and y of my LM3 goes up and down but in the order of 1-3 points, 10 points AT MOST. For example x=0.3081 going up to 0.3091. I do this a lot and would sit a long time watching because some LED like 519a the Duv goes down a lot with time & heat build up. I am used to how stable the numbers are. Lot of times x stays same, and y drops 1 or 2 points at a time.
x and y of my LM4 goes up and down by 20-30 points and subsequently causing Duv to go up and down quite a bit. Check your LM4 Duv, hopefully with a stable light with stable output like Convoy S21D. FET driver may cause output to change too much and this affects Duv. There is something very wrong and very different with, at least, my LM4.
@jon_slider what’s the best way to ask for refund through Aliexpress? What’s the chance of getting our money back?
Sometimes, even when a refund is okay with the seller the shipping cost to China makes it not worthwhile. At least that was my conclusion on a couple of occasions from the US.
