The Sofirn HS21 headlamp has spot, flood, and red LEDs, selectable by a rotary interface. It uses an 18650, rechargeable by USB-C! Check out the infrared sensor!
Click the image (or this link) to read the full review! Thank you!
2 Thanks
Your reviews set the Gold Standard for the most complete information available. Thank you for all the time you put into your reviews. Truly excellent info.
I really appreciate the outstanding range of tests you report in your reviews. Including Color Spectrum with CRI and Tint DUV, Runtime, Regulation, Sustainable Output, LVP and PWM testing.
Great to see the HS21 has Regulated Output and NoPWM… Im a fan of the Red and High CRI White options, and I like the UI.
I really like the HS21, it is an excellent headlamp with both Red and White LEDs.
2 Thanks
Thank you very much for that comment!
Great review as usual! I am really surprised that flood was only 111 compared to the advertised 150, that’s a huge difference!
1 Thank
I’d argue that this is within the uncertainty of the measurement (e.g. compare the runtime plots on turbo with those by koef3). Also, I think perceptually the difference (if there is one) would be hardly detectable.
For a halfway decent calibrated setup the difference between 150 and 111 lumens is way more than it should be with measurement errors. I can make some measurements of mid of my sample. Keep in mind, that different values could be also explained by some variation in LED performance (flux bin), this can lead to differences of ± 7 percent.
3 Thanks
I looked at the constant portion of both runtime graphs on turbo. If I read it right, the zeroair’s readings appear to be systematically lower by some 6%, 22%, and 3% for Spot, Flood, and Hybrid beams respectively. Flood difference (if I read it right) seems a bit high but I’d argue that the other two are as expected given the likely difference in the measuring riggs, the sample sizes of one, and the intrinsic difficulty of integrating the beams to measure lumens.
Does Zeroair uses an integrating sphere?
Only with spheres (in proper geometry) there is no measurement error due to the different beams. Could imagine that tubes or boxes have some sort of error while measuring spot and flood beam patterns. I discovered some errors (at around ± 30 % sometimes) while using a box for light flux measurements, this is why I switched to a sphere some time ago.
I got 161 lm for Mid, with almost full battery. It seems realistic. For a oldschool Thrunite Ti3 / TiS I got 127 lm which is very close to what was measured back in the days (around 120-130 lm).
Ok guys those are good points, but his/her setup is the same for all his tests I imagine, so why the measurements for the other lights are a lot closer to the advertised values? Look at his test for Silkhunt for example.
Seems like zeroair reads consistently about 20-30% lower on Flood.
Thanks @koef3 , that’s interesting. Would be useful to hear zeroair as well. So a difference in 20-30 is barely perceptible. What is the minimum difference that is perceptible. My current light in medium mode (HD15r) is rated at 120 lumen, so if I get a light that outputs 150 won’t feel much of a difference?
There is no real difference for the naked eye. It can be measured of course, but for real life use it is not relevant.
1 Thank
Interesting, so what’s the real difference between models? It seems the range of lumens across modes is quite similar across most models I have consulted? Is it the color of the light, battery run time, etc?
Yes, CCT and beam characteristics in particular are decisive for the perceived brightness (or the amount of light that comes out). It’s all about brightness on the illuminated surface. For the same luminous flux, a lamp with a narrower beam (more throw) appears brighter than one with an extremely wide beam, such as a Mule or the HS21 on flood setting.
1 Thank
That’s interesting. Do we know the CRI of the HS21?
Is there a technical reason why a flood beam would be lower when measured in a non-sphere integrating arrangement, but the spot beam would show less of a bias? I intuitively thought that it would be more likely the other way around, but I know next to nothing about it.
Let me try to explain:
We have a tube, bent into a U-shape if you like. Light comes in on one side and out on the other.
If a flood lamp is used, there is already strong scattering near the entrance hole; the light is well mixed and hits the sensor at the other end of the tube.
With a thrower (or a more spot-heavy lamp), however, a large proportion of the light only hits the bend in the tube. A large proportion of the light is therefore only reflected later in the tube and enters the last part of the tube at almost full power. Less light is lost in the course of the tube, and it is possible that reflected light - depending on the geometry installed - then hits the sensor directly, which distorts the measured values. An integrating sphere prevents this, as the reflections are always at the same distance from the light source due to the spherical shape. That is the trick.
This would explain why he measured consistently less in flood. You would actually expect this flux with flood light with a tube. I could imagine that it was never calibrated with a flood lamp. It is also possible (if no Maukka lamp or similar is used) that the calibration light source suddenly had deviations in luminous flux due to some properties such as the battery or driver. This is not necessarily noticeable without a thorough plausibility check and a check with a second known light source.
But in any case, it is extremely difficult to reliably calculate these deviations due to the beam angle and the beam in the final result. Every lamp behaves differently; every optic produces different results with different LEDs. How can a reviewer with such a tube calculate these complex relationships? An integrating sphere would be particularly useful for such reviewers (where testing lamps is more important than measuring them). And to be honest, I think it is extremely important to improve the measuring accuracy of these sites, especially for reasons of visibility and range. Nobody needs ± 3 percent accuracy, but it should be ± 10 percent. Such reviews can be decisive for purchases and, in the most extreme cases, can make the difference between success and failure.
1 Thank
FWIW and I don’t know if it’s applicable to the discussion, but in my Convoy S21D, using lumen tube the output is higher when I use a floody optic vs a throwy one. I measured 1900 lm when using 60 Bead TIR (more floody), and 1700 lm when using 10 Clear TIR.
1 Thank
Interesting. Perhaps there are losses in the optics, maybe in combination with the difference of the beam in the tube. 1700 to 1900 lm is just over ten percent. However, normally the more floodier optics are more inefficient, at least this is the case with those from Carclo.
Sorry change of subject. How wide is the flood, have you been able to measure it? And compared to say Wizard C2 pro?
1 Thank