Replacement of carbide lamps by LED lamps

[gillestugan]

That is, if the outside surface is frosted, might some fraction of the light reflected back from the frosting into the lens be internally reflected off the inside of the lens and have a second chance at making it out through the front?

That sounds likely.

I have been looking at the Ledil wide optics. They have this kind of tiny bubble texture on the top that works almost as good as a frosted diffuser in reducing glare, but with higher efficiency. I have ordered a pair for some testing. Will also test them on the MC-E as the larger die area tend to make the beam even wider. Using a MC-E is of course more expensive, but you can gain a lot in efficency from driving each die with 1/4 of the current.
I have also ordered a carclo 10403 which I will try to frost slightly.

Here is an overlay comparision of some wide lensens. Of course it doesn't tell the whole story, but it can give a hint. I would have liked to include the carclo 10170 reflector, but couldn't find a graph for it.

 

[uk_caver]

Is having a wider spread than a 'naked' LED (like the Carclo 10403) actually useful underground?

Typically, when moving roughly horizontally, a caver is looking along a passage, or across a floor.
Objects lit up by the centre of a flood beam are typically relatively far away - many times further away than the floor by the caver's feet, and possibly much further away than nearby walls.
Also, despite the fact that eyes can move, typically a caver will tend to turn their head if there's anyuthing of particular interest.
Peripheral vision, being more biased towards non-colour receptors can probably get away with less light than central vision needs.

Personally, I find a more centre-biased flood to be a more useful beamshape - having a little more throw, it has less need for supplementation with a spot beam, while still giving the benefits of a smooth flood.

I can see that there's a use for flat-flood optics like the Carclo 10403 for building lighting, giving a ceiling light with less of a hotspot underneath than a naked LED, but that's for applications where the LED is lighting a flat perpendicular wall, not lighting the inside of a tube or angled shallowly across a floor.

 

[Tobias Bossert]

I agree with with uk_caver: there should be as much light going to the center of the viewing area. Since no visible edge is comfortable, the spill should reach outside of the viewing area.
I like the pattern of XR-E very much, it is much more adopted to our needs than that of P4 - but I don't like its artefacts!

I have made some test in my cellar with some light distributions. First we start lightning a plane wall. The light is oriented rectangular to the wall giving a circular 'spot' on the wall. The following diagramm shows some pattern. The X-axis is the ratio of the distance from the spot center and the distance of the light source from the wall.
So '80%' means e.g. for a light that is 1m apart from the wall, that this point was measured 80cm away from the center of the spot.

'reflector' means Carclo wide angle reflector 10170. SSC P4 as well as Cree XR-E were measured as bare emitter, with the reflector pressed directly against the starboard and with the reflector clicked into an appropriate holder.
As you can see, the reflector does a very good job when used with P4 and its holder. Very good for architectural lightnings - not for caving, because we don't have a plane wall in front of us most time.
The reflector works well for us with XR-E increasing the flux inside of our viewing area. I took the combination without holder as a startpoint for my 'diffuse-light'. There is a sharp edge with this reflector - not good for using round the clock. But that's no problem, because we need a diffuser anyhow to reduce glaring.

The pattern given in the first diagram has no sense for walking in the cave. Therfore I calculated the spatial pattern out of it.

As a next step I used different diffusers with the combination XR-E and reflector direct. Until now I have made tests with plane diffusers only. They reduce glaring strongly, they do not reduce the 'beam' on-axis so strong and they nicely smoth out the edge. But with increasing off-axis angle the loss increases dramatically. Therefore I decided to study domed diffusers next. I'll come up with those results...

Regards
Tobias

 

[gillestugan]

Beautiful!
Makes me want to delete my posted overlay as your post makes it look like child-play (which it really is in comparison).
May I ask how you were able to create those diagrams?

I too don't think Carclo 10403 or a similar optics will be of any use. Im mostly curious. Was able to get free samples, so why not try them. They will probably be less than half as bright in the centre as a bare XP-E, and even less when diffused.

A flat-flood glare just as bad to the sides as to the front, so it is better to have more centre-biased for less problems with the glare. That is without a diffuser of course.

 

[Tobias Bossert]

Revised diagrams in post #103 now

In post #103 I explained the different scaling of X-axes of the diagrams but not the scaling of Y-axis and how I measured it.

I compared a SSC P4 USXOI with a Cree XR-E Q5 WC at 100mA. So the overall luminous flux may be different too, probably the XR-E delivers about 10% more than P4. Manufacturer specs state 91..118lm@350mA for P4 U but 107..114lm@350mA for XR-E Q5. So the ratio XR-to-P4 could have taken any value between 0.9 and 1.25 (luminous lottery). I'm not able to measure real overall luminous flux in my cellar.

I put the light source 50cm apart from a plane wall. I measured every 10cm from center of beam along the wall covering -60cm to +60cm. I allowed Excel to spline the curves. I did not normalize the values, it's just the reading of my lux-meter.

I now have changed the vertical scaling of the diagrams in post #103:
All values are now normalized to "P4 center-value = 1,0"

I hope this clarifies the diagrams.

Regards
Tobias

 

[Tobias Bossert]

Revised and translated comparison now!

In post #84 I gave a link to Bucherls comparison of cave headlamps.
This was German.
Now we prepared an English version:

http://www.hirlatz.at/lampenvergleich_en.html

The explanations are extended now and a switchable comparison mode was added in 'detail 1'.

The original link to German page

http://www.hirlatz.at/lampenvergleich.html

still works and has the same extended explanations too.

Tobias

 

[Tobias Bossert]

Testing diffusers

Post #103 dealed with various combinations of LED-types and wide angle reflectors. I came to the result, that the combination XR-E and Carclo 10170 without holder (bond to the board directly) give the best distribution to start with tests of diffusers.

The diagrams in this post here are normalized to the brightness of the bare emitter in the center (black curve). As in post #103 there are two different diagrams.

The first one shows the normalized brightness of all the eight combinations tested on a plane wall. The light source was 50cm spaced from the wall. The distance along the wall is given as percentage of this spacing; 0% means on axis and 100% means 50cm away from the center (this equals 90 deg full width angle).

The second diagram shows the according angular pattern.

Both diagrams show the pattern of the bare XR-E without any optics (black curve as a reference). The pattern of the bare XR-E is very suitable for illumination of the whole viewing area - but a bare LED glares extremely.

For each of these eight curves i took a photo with identical camera setting. The camera was 125cm apart from the wall with wide angle optic. The bearer will never encounter such a situation, because the patterns cover a very wide angle. You can see the linear distance along the wall at the folding rule (hard to read at 800x800) and the full witdth angles on the yellow labels (30deg, 60deg and 90deg).

This photo is for bare emitter and belongs to the black curve 'bare emitter':

The 2nd curve (light blue) is the combination XR-E and Carclo 10170 without holder (reflector bond to the board directly) similar to that one described in post #103 but with a plane clear uncoated glass lens in front of it.

This photo is for XR-E with Carclo 1070 and glass lens and belongs to the light blue curve 'clear glass':

The flatness on a plane wall is nice, but not suitable for walking in a cave. The edge is much too sharp and artefacts are visible. And the glaring is the same as with bare emitter.

Domed diffusers

I tested a Khatod PLJT 35/02, which is nearely clear. I cut it down to a spherical cap with 6,5mm hight. the remaining diameter is about 24mm. This cap neither reduces the glaring nor improves the light distribution dramatically. I was astonished, that this dome provides some collimating effect and concentrates the brightness to the center!

This photo is for XR-E with Carclo 10170 and clear PLJT and belongs to the reddish long-dashed curve 'clear PLJT':

I made many experiments with PLJT 35/02 and vapor of different bonding agents (also cyanocrylat), but it was not possible to get a defined and homogeneous mat finish. I also tied to sand the internal surface - not good. The best result was achieved with a cleaner polish as used in the household. With this domed diffuser no glaring is left, the light source is about 300mm².

This photo is for XR-E with Carclo 1070 and clear PLJT and belongs to the reddish short-dashed curve 'sanded PLJT':

Plane diffusers

Next I tested three commercial and one home made plane diffusers.

First I tested XR-E with Carclo 1070 and L2optics OPTX-1-DIF8, which is a clip-on device for their 26mm optics. I bond it with the polished side onto a clear glass lense with epoxy to achieve a stable front lens. So the light enters at the rough side, quite the other way round than intended use. The glaring is reduced but still annoys, the light source appears 10mm² subjectively.

The photo belongs to the greenish long-dashed curve 'OPTX-1-DIF8':

Next I tested XR-E with Carclo 1070 and Ledil Ledilstar P SUB, which is a clip-on device for their 35mm optics. I bond it onto a clear glass lens as explained above. The glaring is reduced very well (30mm² subjectively).

The photo belongs to the greenish short-dashed curve 'Ledilstar P SUB':

Additionally I tested XR-E with Carclo 1070 and a diffuser not looking frosted but structured: OPTX-1-016S from L2optics. The glaring is reduced dramatically (200mm² subjectively) but the loss - especially at higher angles - is dramatical too! This belongs to the alternatingly dashed greenish curve 'OPTX-1-016S':

At the end I tested a selfmade diffuser: XR-E with Carclo 1070 and a plane glass lens etched at the internal side facing the LED. The glaring is nearely absent (200mm² subjectively) and the loss is acceptable. This belongs to the not-dashed blue curve 'etched glass':

Conclusions

• A domed diffuser is much more difficult to produce and brings no real benefit as I thought before :-o
• Some 'diffusers' concentrate light to the center too!
• All diffusers I tested povide increasing losses at high off-axis angles.
• It is possible to compensate the losses of a diffuser in the range of full beam angle up to 80 deg by using a wide angle reflector: This costs higher losses outside this range.
• The angular pattern of a bare XR-E can be 'reconstructed' in the range -40 to +40 deg at a comparable level. But outside this range the 'reconstructed' pattern drops below that one of the bare emitter naturally. The etched plane glass is a cheap part for that.

I decided to use the combination XR-E with Carclo 1070 and a plane glass lens in front of it, etched at the internal side facing the LED. This will give a nice 'diffuse-light' when tilt down about 30 deg.

 

[Tobias Bossert]

For those who don't like to scroll so much, here is a picture containing all eight photos.

 

[gillestugan]

Thanks a lot. Those are great pictures and there is a lot of work you have done, and done very well. A bit disappointing the domed diffuser didn't work as well as hoped, but its good to know a cheap alternative works well.
Losses from the diffuser seems to be a little more than 20%. Maybe this can be improved with a holographic diffusing film for example the LSD.
What are the numbers on the scales i the pictures? I will try to make a similar picture and plot for the Ledil SSS-W I am considering.
The Boom-W for SSC P7 also looks quite good with a beam pattern similar to a bare XR-E. But its probably better to use separate optics for the hall light.

Has anyone found a good cable gland? It would have been nice to have a good waterproof gland, but they all seem to be too bulky. Smallest I've found requires a 12mm hole for a 6mm cable.
I've previously used rubber grommets but they don't feel secure and have no strain relief.

 

[likeguymontag]

I've previously used rubber grommets but they don't feel secure and have no strain relief.

For homemade strain relief, I've sometimes used several layers of heat-shrink. Heat shrink also changes the diameter of the cable in small increments, so you can experiment and get an extremely tight fit through your grommet. You can tightly attach a zip-tie around the cable on the inside face, and be fairly sure it won't pull through the grommet.

The Underwriter's knot is designed to address pull through, but probably wouldn't work well with the kind of cable I expect you're using.

 

[Tobias Bossert]

Losses from the diffuser seems to be a little more than 20%.

Yes, you are right, 20% at least! Imagine that the wide angle reflector catches all light outside +/- 45 deg, that's a lot. Most of the light catched by the reflector is necessary to compensate the diffuser losses inside this angle, since bare emitter and emitter-reflector-diffuser combination just achieve comparable brightness inside of +/- 45 deg.

Maybe this can be improved with a holographic diffusing film for example the LSD.

This looks very interesting. Pleasy give some links where such a holographic diffusing film is discussed and where to order some samples.

What are the numbers on the scales i the pictures?

There are two scales in the pictures.

The folding rule is scaled in cm and is centred with its reading "100" in the center of the beam. E.g., if you read "140" this means, this point at the wall is 40 cm out of the center of the beam; the same is true for a reading "60". Since the source was separated by 50 cm from the wall, readings of "50" and "150" are those points at the wall, were the distance at the wall from the center of the spot equals the separation between source and wall. In the first diagram these points are scaled as "-100%" or "100%".

The yellow paper labels above the folding rule mark the edges of beams with 30 deg, 60 deg and 90 deg full width. The yellow labels "90°" correspond to the reading at the edges of a beam with full width of 90 deg and thus both are 45 deg out of axis. The X-axis in the second diagram is scaled in "°" (=deg) off axis, that means, the yellow labels "90°" correspond to readings on the scale of the second diagram of "-45°" and "45°.

Readings "50" and "100" on the folding rule equal scale points "-100%" and "100%" in the first diagram and correspond to "90°" on the yellow labels, which are equal to the scale points "-45°" and "45°" in the second diagram.

Has anyone found a good cable gland?

I'm not experienced in building water- and bombproof housings, this is a task for manufacturers commercially producing headlamps. My mission is to improve the light distribution of headlamps only.

 

[Tobias Bossert]

Now the aspect of glare

In post #107 I posted the test results with respect to the achieved light distribution. That is what you will see when you bear the helmetlamp. This post now deals with what your friends will see, when you look to them - quite different point of view!

In post #107 I also gave a rough subjective estimation of glaring. This was done with the light source 1m apart from my eyes and the eyes beeing in the center of the beam. The XR-E Q5 WC was driven with 100mA.

In this post I'll try now to document the amount of glaring with photos. That's a difficult task, because the dynamic range of photos is limited.

The result leads me to assume that the subjective sensation of glare is not "linear". Anyway, the subjective glare and the "measured" brightness and extension of the light source don't comply directly.

I mounted the LED with its axis tilt downwards by 30 deg and shot the photos from horizontally in front of it. This simulates the situation when you talk to another person face by face. The camera settings for the photos in post #107 were f=1:8, t=1s @ ISO100. The current to the LED was 100mA and the spacing between LED and white wall was 50cm. For these tests here, the settings were f=1:22, t=1/1000s @ 100ISO.

Even with the camera turned down in this manner it was necessary to reduce the LED-current to 10mA. Nevertheless the pictures are overdriven at the most bright parts. On the other hand, the dark parts are drowned out.

Here is a picture with the photos of all 8 combination tested. The layout of the photo corresponds to that one in post #108:

As I stated at the beginning, the subjective sensation of glaring doesn't compare directly to the photos - but you can see how the diffusers work.

The photo "clear-glass" is a good excample to demonstrate dynamic range of photo and eye: I had no problem to see the reflector in total, althoug the die glared me. on the photo you can see the lower part of the reflector only, that is the area wich is active in the upper beam direction.

 

[gillestugan]

I was not able to make the measurements on the Ledil SSS-W with the photo resistor and ohm meter as I hoped. It was too inaccurate, maybe I will try again later in combination with a diffuser over the resistor.
Post some beamshots instead. Emitter is a XR-E 50cm from wall.

The Ledil optics have very high efficiency and doesn't glare much at all, but Im not sure I want to trade 20 degrees more narrow optics for 15% higher output.

I decided to use the combination XR-E with Carclo 1070 and a plane glass lens in front of it, etched at the internal side facing the LED.

I get significantly higher losses from my etched glass when I place the etched side on the inside (towards the emitter) than on the outside.
The LSD seems (too) expensive, $80 for a 10x10cm sheet. Samples can be requested here.

 

[Tobias Bossert]

The Ledil optics have very high efficiency and doesn't glare much at all, but Im not sure I want to trade 20 degrees more narrow optics for 15% higher output.

Definitly I would not accept further 20 deg reduction of beam angle for 'diffus-light' even when this would result in about 15% increased efficiency.
The Scurion uses room light with about 160 deg - too wide for the person bearing the helmet lamp. I reduced it to about 90 deg, tht's the absolut minimum, which could be accepted only when the edge is very smooth.

I get significantly higher losses from my etched glass when I place the etched side on the inside (towards the emitter) than on the outside.

The lamp will be used in caves and there is a lot of loam. With the rough surface out you may get problems with loam, since it is difficult to clean it only by hand. Therefore I excluded 'rough side out' up to now.

Visually I didn't experienced this heavy difference. But I will make some relative measurements and post the results soon...

What will be tested? Not all possible combinations, since some exclude from other reasons.

With etched, sanded or sandblasted glass lense I can turn them to matt side out without any problem.

Cheep plastic diffusers like OPTX or LEDILSTAR are thin and not solid enough for a cave lamp. So I need to cover them by an additional glass lens (or a thick polycarbonate lens). The problem with this is, that I get 4 surfaces with backscatter instead of 2.

I avoided 2 of these backscattering surfaces by bonding the polished surface of the plastic diffuser directly onto the inner side of the glass lens. This works fine, because the refractive index of epoxy is similar to those of glass, polycarbonate and acryl.

It is not possible to bond the rough side of the diffuser to the glass lens, because the epoxy will fill it up and the diffusion disappears.

I'm interested how big the difference is objectively?

The LSD seems (too) expensive, $80 for a 10x10cm sheet. Samples can be requested here.

Yes, much too expensive for you and me. And at the end - if I can find a manufacturer using our findings and ideas - it will not be used because of price. So I must accept the losses of 20 to 25%.

 

[gillestugan]

Beautiful drawings, but I feel a bit guilty. I didn't mean to give you more work. Just was a little bit surprised when I got more than 5% lower output when I turned the frosted side in (measured with a photoresistor). 5% is hard to notice, but it means you can lower the power consumption of the leds by at least the same %.
The frosted glass I use is a quite heavy frosted photo filter.

The first idea is often the best, and this search for lenses and diffusers are a good example. but.. Im still hoping for another solution as I would like a more simple method. It is a lot of work to get a frosted lens in the right dimension, then mount it to the reflector and then make it securely mounted to the headlamp at an angle.

 

[Tobias Bossert]

Here is a comparison of rough surface inside and outside


I tested the thre "winners" of the last test:

• etched plain glass lens
• OPTX-1-016S
• LEDILSTAR P SUB

In all diagrams there is a grey curve for bare emitter XR-E, which was measured with a plane glass lense (uncoated clear glass), since in caves we can't use the bare emitter without protection. This glass lens takes some % of light (4 to 6%).

All tests were carried out with the combination XR-E / Carclo 10170 without holder and in front of this either the etched glass or a combination of commercial diffuser and clear glass lens as explained below.

The etched glass is compared with etched surface inside (facing the LED) and with surface looking outside:

This is the result with the etched glass lens:

As you can see, no one orientation is really better than the other, but the effects are different: With the "usual" orientation "etching outside" the brightness in the center is about 12% higher, but with "etching inside" the beam gets considerably wider.

With both commercial diffusers the effect is comparable. They were tested in combination with a plane glass lens:

This is the result for OPTIX-1-016S, which is not frosted but has a rectangular texture:

This is the result for LEDILSTAR P SUB, which is rough-frosted:

The subjectiv glare is uninfluenced from the orientation of the diffuser.

At the end I can verify the general rule to use "diffuse side out". But especially if you want to get a very wide smooth pattern, the opposit works better.

 

[Tobias Bossert]

Using 1 or 4 dies of MC-E with a Carclo 10195

As you remember, my "testdevice_2" had four independant LEDs. I announced trying to reduce the device to three LEDs without loss of functionality.
I was not sure on this, because I didn't know the result of useng one die of MC-E only in a TIR. Now I'm sure, the approach of a 3-LED-Lamp will work!
I made some photos from that. The optic was 50cm apart from the wall. the yellow labels mark a full beam width of 30deg (+/-15deg). As usual, those photos don't show what you will imagin subjectivly, because the eye has a quite different dynamic as a photo.

XR-E Q5 WC @100mA with Carclo 10195 (20mm medium frosted) and Holder 10205

MC-E K WC all four dies @25mA each with Carclo 10195 and Holder 10512

As you can see, the beam is broader with MC-E than with XR-E (may be full beam witdh is about 10deg wider) and the pattern tends slightly to a rhombus. The LED was mounted diagonal so that one die is on top, one on bottom and two side by side horizontally. But there is no artefact and no donut - esthonishing since Carclo 1095 was developed originally for Rebel, quite the opposit, a small emitter!

MC-E K WC only the uppermost die @100mA with Carclo 10195 and Holder 10512

As you can see, the center of the beam is declined downwards by about 10deg now ans seems to be slightly higher than broad. But it is still smooth without edges and rings.
This could be used to light-up a diffuse light smoothly without anoying edges.

In the following picture I mounted both photos of MC-E beams together. On the left side the uppermost die is driven allown and on the right side al four are used.

Conclusion for my project

1. I can use the combination of Cree MC-E / Carclo 10195/10512 to integrate the strong 'hall-light' (2.8A) with the week 'light-up' (30mA) without the need for a driver-dynamic of 100:1.

• For my 'work-light' I want a superposition of 'light-up' and 'diffuse-light' diverging by about 30deg. Since the tested combination with only one die is displaced by about 10°, the champfer in the casing for the diffuse-light can be reduced to about 20deg.

Thus the 'testdevice_3' should look like this...

...in theory, but I'm not professional manufacturer and my hand made result will look more improvised!

 

[gillestugan]

Here comes my test of diffusers. (at last)
The meter was connected to the led with a string som distance to led was constant. I also measured light level every 5 degrees, so I ended up with 200 measurements... PM me if you want the numbers.

The diffusers were placed tested 7mm in front of the bare emitter. When using the carclo 10170 80 degree reflector it was placed direclty in front of the reflector.

The lenses and diffusers used were:

KD multicoated 1,5mmX50mm lens
KD Sodaglass 1mmX26mm lens
LSD 5 degrees (light shaping diffuser) from POC
LSD 30 degrees (light shaping diffuser) from POC
LSD 80 degrees (light shaping diffuser) from POC
Fastcar 5L diffuser (unknown manufacturer)
Fastcar 10L diffuser (unknown manufacturer)
Frosted Glass (Rowi international photo filter)

XR-E emitter, Linear scale

Carclo 10170 on XR-E, Linear scale

XR-E emitter, logarithmic scale

Carclo 10170 on XR-E, logarithmic scale

Now, it is in the area around 40 degrees it gets interesting.
Notice for example how the Fastcar 5L has the same brightness as the frosted glass at angles between 40 and 50 degrees, while at the same time having 15% higher brightness in lower angles. Not bad at all.

I also have taken some photos for evaluating anti-glare performance. Unfortunately I was not satisfied with the fastcar diffusers and it seems frosted glass is still a good solution as the LSD:s are horribly expensive.

And Tobias, good it worked fine with the single-die carclo 10195 setup. Thanks for testing.
I tried it with the 10194 and it looked really bad. Going to use that setup too. Looks good and makes for less components than driving all the dies at low level.


Edit:
Some more information:
I placed rough side towards the emitter on all diffusers.

The fastcar 10L was not as effective as frosted glass for reducing glare. Better than LSD5, but the "grain" is quite large. I would estimate the frosted glass is 50% better. It is still much better than a bare emitter/reflector, so if someone wants to trade some glare for better output it may be a good alternative.
It is 0,3mm thick, and quite though, it would probably hold for a while if placed in front of a 20mm reflector. Not bombproof, but i think it would work.


I took new close-up pictures to show grain size.

 

[Tobias Bossert]

Hi Anders,
very good work and very informative posting!

Could you add the secondary grid to the logarithmic diagrams (not only 1, 10, 100 but also 2, 3, 4, 5, 6, 7, 8 and 9) please, this would be better to compare bare emitter to emitter&carclo. Just replace the uploaded diagramms...

I like your results of 'Carclo - Fastcar 10L' since this combination seem to reduce glare and has losses lower than the frosted glass you tested. What is the subjective effect of glare suppression of this combination, does it match to frosted glass?

Which orientation of diffuser did you use during test, rough side in or out?

Your findings are corresponding with my results: A diffuser reducing glare sufficiently will reduce the 'gain' of the Carclo reflector 10170 to about zero, that means you just can achieve the pattern of the bare emitter within a beam of +/-40° and pay for glare suppression with losses outside this angle.

 

[gillestugan]

Thanks. I have added the grid to the diagrams, good suggestion, I was a bit tired yesterday when I posted them and didn't think about it.
Also added some more information at the end.

I still think a 20 degree LSD would have been great, but the cost is way to high to justify.

I am still working on the single pole rotary switch circuit based on the AX2002 drivers. Have some problems with diodes and will post eagle schematics when solved.