Destructive Incan Bulb Tests - Updated 8/27/2010 (Newer Info Added)

[LuxLuthor]

6/28/09 Note: Ictorana's Graphs are now being presented in this other pinned thread.

8/14/09: Shoutout to Drewfus2101 for doing destructive testing of most of the potted Maglite bulbs in charts linked here.

I started testing bulbs after thinking about various ways to get some REAL bulb data, rather than everyone relying on AWR's Hotrater spreadsheet to predict overdriving Incandescent bulbs. I tried to control for as many variables as I can think of, and am showing photos of my setup. I have used two or more new bulbs for all tests listed below done on different nights to verify and correlate results. Complete testing method is listed below thumbnail images.






Setup for potted bulbs:

To eliminate resistance, I'll check it using a Magswitch bulb holder I just setup. Soldered 14AWG to parts as shown in these thumbnails:



Martek 4509Q (Par36) Bulb Tested
---------------------------------------------------------

• 4509Q (13V 100W) **New**

​

---------------------------------------------------------

WA Bulbs Tested
---------------------------------------------------------

• 1111

​

---------------------------------------------------------

• 1274

​

---------------------------------------------------------

• 1326
  

​

---------------------------------------------------------

• 1331

​

---------------------------------------------------------

• 1160
  

​

--------------------------------------------------------

• 1164

​

--------------------------------------------------------

• 1166

​

---------------------------------------------------------

• 1185

​

---------------------------------------------------------
MagCharger Bulb Tested
---------------------------------------------------------

• LR00001 (6V 10W)

​

---------------------------------------------------------
Kaidomain Bulb Tested
---------------------------------------------------------

• KD-773U (6V 13.8W)

​

---------------------------------------------------------
Carley Bulbs Tested
---------------------------------------------------------

• CL-809 (10.5V 12.6W)

​

---------------------------------------------------------​

• CL-1057 (6V 12W)

​

---------------------------------------------------------

• CL-1794 (7V 21W) Custom made for FiveMega
  

​

---------------------------------------------------------

• CL-1794 (7V 21W) Revised Default Lumen Rating
  

​

---------------------------------------------------------
​

• CL-43 (14.8V 41W) Custom made for FiveMega

​

---------------------------------------------------------

• FM-1909 (11.4V 63W) Custom made for FiveMega
  

​

---------------------------------------------------------
Philips Bulbs Tested
---------------------------------------------------------

• HPR71 (6V 10W Potted)

​

---------------------------------------------------------​

• 5761 (6V Philips)

​

---------------------------------------------------------
Westinghouse Bulb Tested
---------------------------------------------------------

• 4743 (12V 20W)

​

---------------------------------------------------------
GE Bulb Tested
---------------------------------------------------------

• 347118 (6V GE)

​

---------------------------------------------------------
AW (Generic Chinese) Bulb Tested
---------------------------------------------------------

• 6V 30W 2000Hr Chinese Generic

​

---------------------------------------------------------
ROP Pelican Big-D 3853 & 3854 Bulbs Tested
---------------------------------------------------------

• 3853-L 7.2V 11W Pelican Big D Low (NiMH)
  

​

---------------------------------------------------------​

• 3853-H 7.2V 24W Pelican Big D High(NiMH)

​

-------------------------------------------------------​

• 3854-L 6V 11W Pelican Big D Low

​

---------------------------------------------------------​

• 3854-H 6V 24W Pelican Big D High

​

---------------------------------------------------------
Hikari Bulbs Tested
---------------------------------------------------------

• JC-5033 (12V 35W)
  

​

---------------------------------------------------------​

• JC-5043 (12V 50W)
  

​

---------------------------------------------------------​

• JC-5051 (12V 75W)
  

​

---------------------------------------------------------​

• JC-5607 (6V 20W)
  

​

---------------------------------------------------------
Osram Bulbs Tested
---------------------------------------------------------

• 64250 (6V 20W)
  

​

---------------------------------------------------------
​

• 64430 (6V 35W Osram Brand)

​

---------------------------------------------------------

• 64430 (Tungsram 56580 version from Svetila)

​

---------------------------------------------------------

• 64430 (Top Bulb Chinese Generic)

​

---------------------------------------------------------

• 64275 (6V 35W)
  

​

---------------------------------------------------------

• 64610 (12V 50W)
  

​

---------------------------------------------------------

• 62138 (12V 100W - USL)

​

---------------------------------------------------------

• 64625 (12V 100W - USL)

​

---------------------------------------------------------

• 64623 (12V 100W - Torch)

​

---------------------------------------------------------

• 64458 (12V 90W)
  

​

---------------------------------------------------------

• 50050 Ministar - 12V 50W (Osram Link)
  

​

---------------------------------------------------------

• 64432 IRC - 12V 35W
  

​

---------------------------------------------------------

• 64440 IRC - 12V 50W (Osram Link)
  

​

---------------------------------------------------------

• 64447 IRC - 12V 65W

​

---------------------------------------------------------

• 64633 (15V 150W)
  

​

---------------------------------------------------------

• 64655 (24V 250W)
  

​

---------------------------------------------------------​

• 64657 (24V 250W)
  

​

---------------------------------------------------------​

• 64656 (24V 275W)

​

--------------------------------------------------------------------

Step One - I measure bulbs at increasing voltage steps, up to the point where they flash. Each voltage setting only stays long enough to get stable readings.

Step Two - I will compare these measurements to the AWR Hotrater Spreadsheet (based upon WA website's default & predicted overdrive values) for correlation with my results among all the tested bulbs. This may lead to some adjustments being made on a revised Hotrater Spreadsheet.

Step Three - I will pick what appears to be the two best predicted voltages that optimze lumens with bulb life in an 8-12 hr predicted range, and run the bulbs until they die. The brighter, lower life choice will be done first.

Test Platform: I mounted a bare KIU to a block of wood, with 4" wires soldered to 20" 14 AWG test leads, plugged into a Mastech HY3020-D Linear current/voltage adjustable power supply. New bulbs are inserted in KIU & cleaned with Isopropyl alcohol.

At the same height as the bulb, a black plastic (1" I.D. x 12" long) tube is placed in a horizontal vise position, aimed by looking from far end so bulb appears centered in tube. Inside of tube was abraded to avoid any spot reflections. At exactly 1 meter from bulb, Meterman LM631 light sensor is inserted in far end of tube that has a thin layer of black foam to seal around sensor and hold it in place.

For Step One, a Fluke 179 using a thermistor probe positioned 4mm from bottom of bulb envelope to get relative bulb temps. A black blanket is positioned behind the bulb, and along one side to eliminate reflected light. The idea behind this test platform is to consider an incan bulb as a "point light source," and take Lux measurements with the tube to get a narrow cone of lumen output from the side of the filament.

There is no way to use an Integrating Sphere on a practical basis for this many bulbs, and I am not using a reflector or measuring out the front of a bulb which may introduce hotspot variations. I am using the Amps displayed on the Mastech during this step which only displays 1 decimal.

This platform will not correlate with "bulb lumens" from a manufacturer, in part because I am only measuring a narrow cone of light, and from a distance of 1 meter. However this step will be useful in providing more realistic comparisons between bulbs as long as the same test platform is used. The 1 meter distance helps eliminate bulb artifacts.

Step Two will hopefully allow being able to revise a spreadsheet so it becomes an accurate tool. Right now, we know there are major errors that magnify as you get farther from the manufacturer default results. We may also be able to compare manufacturer claims against other bulbs to see if their default values give accurate Lux measurements when compared.

Step Three will not require the Lux measuring setup, but I will verify the same Lux reading as I got in Step One when starting. This is basically a run time verification of the projected life. I will only be doing this in steps of 1-3 hours at a time, so I verify the bulb failure time at what looks like the best two voltages. I will also use my Fluke to measure the actual Vbulb and current during this step.

I may also consider a Step Four of mounting Lux sensor on wall, and set it to record "Maximum" reading and shine bulb from a distance using a Maglight with adequate battery source and reflector to get a somewhat objective controlled "Torch" hot spot output between various lights. This would be ideal if I could supply regulated (and verified) Vbulb from best result of Step Three. It would very hard to control all the variations from light to light, but would still be some interesting data.

 

[TorchBoy]

Re: Destructive Incan Bulb Tests

Sounds like very useful tests. In step one are you ramping/stepping up the voltage or turning it on at each level from a cold filament? Both could be useful if you're doing soft starts.

 

[LuxLuthor]

Re: Destructive Incan Bulb Tests

Sounds like very useful tests. In step one are you ramping/stepping up the voltage or turning it on at each level from a cold filament? Both could be useful if you're doing soft starts.

I am just stepping up voltage after I take my readings. I want to control and avoid any possibility of higher spiking voltage on startup (like turning off power supply then back on)...because I'm ultimately trying to assume using this information for soft starting setups. I also would have no idea of the comparison between this power supply starting up and a set of batteries in a non-soft starting setup. The issue you raise gets more into bulb life than it does stable lumen/lux output at a particular voltage....which we assume from the Hotrater.

Gonna do one of my 35W 6V Osrams now just to see how that goes, and an 1166. I'm waiting for one of these bulbs to explode on me.

 

[TorchBoy]

Re: Destructive Incan Bulb Tests

I also would have no idea of the comparison between this power supply starting up and a set of batteries in a non-soft starting setup.

Good point. Right on the limit where you're testing a subtle difference may have a significant effect.

I'm waiting for one of these bulbs to explode on me.

I hope you're wearing safety goggles etc.

 

[DM51]

Re: Destructive Incan Bulb Tests

LL - this is excellent! Extremely valuable work - very useful indeed.

STICKY !!!

The figures for the WA 1185 are very interesting, and as you add others this will become a major reference resource - exactly what is needed.

 

[LuxLuthor]

Re: Destructive Incan Bulb Tests

I added 1166 & the Osram 64430 vs. the TopBulb (Chinese) 64430.

You can start to see how the Hotrater is falling apart on its predictions. I can also see the need for visual reporting of beam color/shape & the Hotspot Lux from a distance I mentioned in Step 4. Right now I am putting these each in a separate Excel sheet, but it's going to be hard to combine all this information in one display.

 

[karlthev]

Re: Destructive Incan Bulb Tests

Lordy Lux, this otta be data you could SELL to the manufacturers!! Thanks!!! Rather incredible work here!!!


Karl

 

[Valolammas]

Re: Destructive Incan Bulb Tests

Excellent! I'm looking forward to your test results.

It just crossed my mind, that do you think the open air environment of the bulb in your setup vs. a small, enclosed space of a flashlight head will have any effect on bulb life? I know heat dissipation isn't much of an issue with incans, but since you are testing them to death, I just thought the filament might be able take a little bit more abuse in open air due to less heat build-up. I don't mean to criticize or anything, I think you are doing a great job! Just a thought.

 

[LuxLuthor]

Re: Destructive Incan Bulb Tests

Thanks Karl. Well this is something I have always wanted to know...and at least there is some objective info now.

Excellent! I'm looking forward to your test results.

It just crossed my mind, that do you think the open air environment of the bulb in your setup vs. a small, enclosed space of a flashlight head will have any effect on bulb life? I know heat dissipation isn't much of an issue with incans, but since you are testing them to death, I just thought the filament might be able take a little bit more abuse in open air due to less heat build-up. I don't mean to criticize or anything, I think you are doing a great job! Just a thought.

Yeah, I think it is only reasonable to suspect that contained Mag Head heat would affect bulb life somewhat. That's why this is all kind of relative to other comparisons I do with the same setup, and why my Step 3 will run two more of each bulbs at what looks to be the "sweet spot" (you can see my note on far right "pending" of the two voltages I'm intending to use)...and starting with the higher voltage.

If my pending voltages don't last as long as they should, or even longer...you can still use that to extrapolate a bit above and lower voltage. It's pretty hard to figure out the life of bulbs in a battery powered setup...where you may not run it for more than 20-45 seconds, or where you are using a direct drive setup with startup spikes, battery drain, etc. I have no practical way to keep track of the total bulb life in a light...unless i only had 1 or 2 that I used all the time...so this will at least give us a "ballpark" real life vs. predicted values.

 

[JimmyM]

Re: Destructive Incan Bulb Tests

The results from the 64458 testing will be really interesting.
The hotrater had the bulb lumens over 11,000 at 20.4V. So the measurements Lux makes will be really eye-opening.
Just be sure to move that black blanket nearest the bulb. It won't lasy long against the likes of the 655s and 458s.

 

[Patriot]

Re: Destructive Incan Bulb Tests

LL - this is excellent! Extremely valuable work - very useful indeed.

STICKY !!!

The figures for the WA 1185 are very interesting, and as you add others this will become a major reference resource - exactly what is needed.

Big time STICKY. This is really great work! Just think how long you guys have been modding and how nice this would have been to know all those years. Thanks to Lux, we'll finally have some very useful info.

Nice work Lux :thumbsup:

 

[mudman cj]

Re: Destructive Incan Bulb Tests

Excellent work Lux! I am especially interested in the discrepancy you are measuring between predicted lumens and measured lux. This measurement may be good enough as it is, but I thought I would propose a couple of alternatives for discussion.

The potential problem with lux measurements IMO is bulb artifacts, and from your post I think you would agree that this needs to be considered. I also agree with you that lux readings within a reflected (or refracted) beam are subject to a lot of variation and would make repeatable results very difficult to obtain. Your current method still seems subject to bulb artifacts though, since the meter long tube is effectively sampling the output at a small solid angle of the total output (aka lux when adjusted to the right units). What if there happens to be a bulb artifact projected right onto the light meter at the end of the tube?

One way to deal with this is to sample at multiple solid angles and average the results, so you would just reorient the bulb or detector tube. This may be borderline practical.

Another method would be to 'sample' a large solid angle with the help of a reflector and measure the light intensity at the end of your meter long tube through a frosted glass lens to diffuse the beam and obtain a more even light intensity reading. To make myself clear, the diffusing lens would be placed at the beginning of the tube and the meter would be at the end just like it is now. In this way, lux variations within the hotspot would be smoothed out.

Thoughts?

 

[LuxLuthor]

Re: Destructive Incan Bulb Tests

Mudman, the main reason of putting the sensor 1 meter away from the bulb, and position a side reading of bulb envelope was to avoid bulb artifacts. The tube is 12 inches and doesn't run the full meter. Just to see if there was a change in average peak lumens related to transverse filament being parallel or perpendicular to sensor, I rotated bulb block 90 degrees and found no difference (still within 1-2 Lux which I consider at least within the margin of error) after lining it up again. I will check that again with some of the "uglier" filament bulbs.

I suspect that most artifacts we see are more related to output coming out the front of the bulb, and also from the reflection magnification picked up by the reflector. If you put most of the bulbs we use in candle mode and look for artifacts on a white wall perpendicular to bulb, and 1 meter away, they do look like a point light source.

However, for Step 4, I think it would be most useful to have something like a sheet of tracing paper between bulb in reflector testing for torch mode...and the sensor to simulate a bounce test, but still reflect the hottest spot projected on the paper/screen.

 

[petrev]

Re: Destructive Incan Bulb Tests

:thumbsup:

 

[JetskiMark]

Re: Destructive Incan Bulb Tests

LL - this is excellent! Extremely valuable work - very useful indeed.

STICKY !!!

The figures for the WA 1185 are very interesting, and as you add others this will become a major reference resource - exactly what is needed.

I agree, outstanding work Lux. Thank you for your time, effort and testing expenses.

Regards,
Mark

 

[jimjones3630]

Re: Destructive Incan Bulb Tests

Well done Lux, very well done. 2nd or 3rd the Sticky:thumbsup:

 

[TorchBoy]

Re: Destructive Incan Bulb Tests

Measured life in hours for the top setting is zero - could be added to the spreadsheets.

 

[Lips]

Re: Destructive Incan Bulb Tests

$20 donation to help cover bulb expense/cost if you'll take it...







This should give us some good voltage set-points for AW's new D regulator, will it not...

 

[LuxLuthor]

Re: Destructive Incan Bulb Tests

Thanks Lip & everyone. Yeah, one of the main reason's I wanted to do this was the find "The Sweet Spot" for regulated drivers. I'm saving my tray of dead bulbs !!! I actually tested a couple of them twice tonight to verify correct numbers to see if there was a signifiicant variation in bulb quality control. I also made sure to make the transverse filaments perpendicular to tube, so it is looking at side of filament.

My double runs of the 1185, 1164, 1166, 1331, 5761, C-43, 64625, 64623, 64458 were almost on the money...like mostly within 2 to 5% which is much better correlation than I hoped for. I redid the Osram 64430 because I realized I forgot to make sure it was centered in the tube...late last night, and results were better on repeat. There's some more tests I put up just now. It is getting pretty interesting comparing some of these now.

 

[plasmaman]

Re: Destructive Incan Bulb Tests

Great work Lux - very useful indeed :thumbsup:

Look forward to seeing the GE34711, Osram 64250 and WA1274 soon!