Super light ?

[curious]

Could you post some pics of your beam pattern at night for us.

Ok, it was quite cold outside so the composition is not perfect. I made three shots for comparison setting camera in manual mode - same exposure and same distance from the camera and light to the garage door.

1. Niterider halogen 20w+12w both lamps on (my main light on upright ebike):

2. Trailtech HID MR11 flood bulb (-FX) 13W. This will go on my recumbent.

3. Niterider Ultrafazer Max 0.5W single LED, my backup light/blinker.

As you can see the 32W halogen light is pretty good with very wide coverage, but it eats 32W. TT HID 13W has smaller (but still good) spot size and slightly higher brightness in the center. And the Ultrafazer Max while quite bright and efficient for 2xAA LED light is well just a backup light.

 

[lifepo4ever]

my p7 MTE look like the same beam of the first pic and its cost me 35$ and its very brigth for riding on my ebike , i use it a medium on the road and if put it on
high the car driver hate me to much , the only thing i am not happy with , there no focus ajustments , i saw one on ebay with the speed focus and the name of the ligth is lenser with a p7 emitter but when you look a video on youtube of the ligth its look brighter then my 600 lumen MTE and the lenser is rated 200 lumens
http://cgi.ebay.com/Led-Lenser-P7-C...9062081?pt=US_Flashlights&hash=item230ba29e41

http://www.youtube.com/watch?v=OHLDR6HAR38&NR=1

 

[curious]

BTW, I am pretty sure the sharp shadow on the upper-left is from my hand holding HID light (while making picture with the other hand). HID is not mounted, I used a long extender cable with cigarette plug adapter inserted into auxiliary 12V port on my ebike. The beam extension is pretty symmetrical in real life.

Also, I turned down exposure quite a bit not to saturate the center spot for high power lights. In reality even the tiny LED light generates enough diffused light for a dedicated paved bike trail at night where you are not blinded by incoming traffic lights.

Camera white balance is auto-only, so it does not show the color temperature differences properly. Halogen is very nice, followed by HID (more or less cool white), followed by LED with even colder spot and warmer (greenish) halo. In the past I've experimented with Luxeon LEDs rated for warmer light outputs (3500-4000K) they are more pleasing yet slighly less efficient and still have a colored halo.

 

[lifepo4ever]

i like halogen lite because for me its more a natural color but consumme lot power then my led flashligth

 

[texaspyro]

I got a box of big, bad boy LEDs from China in today (2-3 weeks earlier than expected). A quick test of the 30 watt units on a 5S A123 pack shows no problem driving them directly without any active circuitry. I actually had to drive them at 6S to get 1650 ma into them (very poor clip leads/meter shunt). They are rated at 2100 mA. 4S drew 65 mA, 5S drew 650 mA. This should be fun... now where did I put that seeing eye dog? Maybe I can borrow one from Amberwolf?

I also got one of those packaged 20W/1500 lumen floodlights that run off of 12VAC/VDC. Looks like it would make a nice headlight. Gotta see what happens when the dark rises...

 

[texaspyro]

I tested the packaged up 20W floodlight. Draws 1.65A from a 4S A123 pack at 13.15V. It has a 140 degree kill zone. Lights up a 20x50 foot wall pretty well from around 10 feet away. Also nothing can hide in a 20,000 sq foot back yard. Pointing it down the road, it looks like it would be a pretty good substitute for your car headlights. It has no forced air cooling and is sealed/watertight/(supposedly) IP65 rated. Could not feel any heat rise after 15 minutes.

I did another test on the 4S pack on the 30 watt LED without the crappy clip leads (but with crappy meter shunt). LED current went from up 50% from 65 mA to 95 mA. My booster cells to try 5S and 6S currents needed to be charged.

I bought some 60 degree optics (reflector and lens) to go with those raw LEDs. Should make for a pretty good flashlight.

 

[texaspyro]

Well, the controller in the 20W floodlight went to rest with its ancestors in the not-so-honorable Chinese elctrocrap pile in the sky. Turned it on, heard a click, light came on dim and flickery and drawing around 30 milliamps.

I charged up my booster cell and hooked up a 5S A123 pack directly to the 30 watt led (rated 16-18V, 2100 mA). Through the meter shunt and wires it delivered 16V at the LED terminals, 1700 mA. Freakin' bright!

With a direct connection to the LED (no meter, shorter wires) it puts out 16,000 lux at 1 foot without any optics. 60,000 lux with the 60 degree lens. Projecting on a sunlit ceiling from 9 feet away, it oushines the sun.

I now need to mount it on the P4 heatsink/fan so that it can run for more than a few seconds at a time.

 

[mushymelon]

Now thats more like it. I assumed that pos ac unit wasn't pushing that emitter up to its potential.

 

[texaspyro]

And remember, those are the 30 watt / 2300 lumen numbers. They make these things up to 300 watts/20,000 lumens (probably more like 15,000). Not sure how you get the heat out of those though. But, running the 30 watt device naked and unsunk seems to produce less heat than I thought it would.

The 100 watt/8000 lumen devices look rather interesting. Should be able to run one directly off a DeWalt pack, plus you could use the speed control input as a dimmer.

 

[texaspyro]

I mounted the 30 watt LED on a Pentium 4 heatsink/fan and ran it for an hour. Zero temp rise. The fan may not be needed. I have it on a crappy dimmer that is dropping the max voltage down to around 16V (from 17.5V) The LED is rated at 16-18V, so I am at the low end if its peak luminosity (the doctors say vision should return in a few days). It puts out around 45000 lux at 1 foot (instead of the 60,000 I had without the dimmer). I need to try Dale's dimmer... it is a better design.

You do get cleaner light without the lens (but it is spread over 140+ degrees). The light with the lens has some variations due to focusing the individual LEDs in the array. Also produces a yellow halo around the edge. I wonder what it would look like with the 5 degree lens (I am using 60 degrees now).

 

[texaspyro]

More playing... I added a switch to the fan. It does look like you should use a fan on these. My heatsink/fan is an Intel P4 cooler. The heatsink measures 3.3 x 2.7" with 1.5" fins. It gets rather toasty without air. It does not need the full fan power. Just a little bit of air is enough to keep it at ambient. The smallest 24V fan you can find running off the 16V should be more than sufficent. I have a couple of crapped out Harbor Freight hand lanterns that I think I'll build it up in.

I did some playing with the 50 watt 22-25V, 2400 mA, 3800 lumen LED. Runs just fine off a 7S A123 pack. No dimmer/regulator used.

Hopefully the 100 watt unit will get here shortly. I still have some functioning retina to fry...

 

[texaspyro]

I also recently got in a 900 lumen flashlight (ebay item 190480928586) Works very well. Lights up the back yard like nobody's business. Runs off a single 3000 mAh 18650 Li ion cell. Has a 5-mode controller built in.

The same LED/reflector is available in a bike light. http://cgi.ebay.com/P7-900-Lm-LED-bike-light-bicycle-headlamp-Flashlight-/180601887638?pt=LH_DefaultDomain_0&hash=item2a0cb61b96#ht_2469wt_941 If it is anything like the flashlight, should make a killer bike light.

 

[texaspyro]

I've been playing with using this dimmer on the big honkin' LEDs: http://www.dalewheat.com/ It's really a neat little device, works very well, and is cheaper than dirt. Pick up a few. Give them away to friends and strangers. Dale can probably use the bucks!

In standby it draws about a milliamp... a bit high to use without a power switch. Also the FET limits you to under 20V. So, I rewrote the firmware (it uses an Atmel ATtiny13A) and changed the voltage regulator and FET. Now it draws less than 15 uA in standby and can handle 40V. That means you can leave it connected and turned OFF for over 10 years and still not drain an A123 cell. With a little more mods (other than the simple part number changes that I did) you could do 60V and cut the power down another 5 uA or so.

I also tweaked the firmware to support a temperature controlled fan on the heatsink... that does require some hacking on the board to implement.

I haves used 5S A123 packs for the 20/30/50W LED, 7S for some 50W, and 10S for the 100 watt. The A123 cells are ideal for driving the LEDs directly without a constant current regulator (which would be nice to have, but not really essential) Their voltage matches the LED requirements very well and they have a very flat discharge curve so the light output does not sag too much. Now I need to build a micro-power BMS (or at least LVC) and charging system for the batteries...

 

[texaspyro]

I hooked up my 50 watt LED to a 7S A123 pack... real freakin' bright. Way more so than the 30 watt one. So much more so, that I went back and checked the 30 watt unit running on 5S. The device was only drawing 1400 mA at its rated voltage. It should have been doing 2100 mA. I checked the LED Vf voltage and found it was quite a bit higher than advertised. No-name Chinese device mis-spec'd... who'd a thunk it possible...

I found a US maker of a 5000 lumen device for a reasonable price. They do have extensive specifications and proper lumen testing, etc. The voltage specs look to be a perfect match for direct driving from an 8S A123 pack.

I added some support to my LED controller for reading the battery voltage and doing LVC protection of the battery and HVC protection of the LED. I can also now boost the PWM duty cycle as the battery voltage sags to help keep the light output constant (this unit is still basically a direct drive/unregulated current LED controller)

 

[texaspyro]

I got a 50V/60A logic level fet installed in the dimmer. Hooked up my 100 watt LED. Unfortunately with the sub-packs that I have right now I can only configure 9S and 11S packs... it needs a 10S pack. At 9S it draws 22W/750 mA. At 11S, I can pump 5A/180 watts into it Time to wire up a 2S booster pack and see what this puppy can burn... as soon as my vision returns.

 

[amberwolf]

I found a US maker of a 5000 lumen device for a reasonable price. They do have extensive specifications and proper lumen testing, etc. The voltage specs look to be a perfect match for direct driving from an 8S A123 pack.

What's your idea of "reasonable price"? And do they sell in hobbyist-sized lots?

 

[texaspyro]

What's your idea of "reasonable price"? And do they sell in hobbyist-sized lots?

About 67 bucks in onesies from Digikey. They have reflectors for beam shaping for around 5 bucks. Search for "Bridgelux" My Chinese 50W (rated at 3800 lumens... who knows what it actually is) was around 40 bucks. The Bridgelux is spec'd at 5000 lumens, and looks like it can do more with a little overdriving.

I've been doing some more testing on directly driving the LED arrays from the battery. It's definitely not the best way to do it since LED current/light output is HIGHLY dependent upon applied voltage, but it is very simple and efficient. My application is basically a big, honkin' flashlight and simple and efficient is good. Plus it has ultra-low standby power.

With an A123 pack, the LED current (and light output) can vary 3:1 from a basically fully charged pack with the voltage matched to the LED such that, freshly charged, the LED current is at its max specified current to when the pack voltage drops around to around 3V per cell (where the pack is around 90% discharged). In the sweet spot of the A123 discharge (3.3V down to 3.2V) curve it looks like the LED output will vary around 25%

I'm looking at what it would take to do a current or voltage regulator for the device... but am probably going to just boost the PWM ratio as the battery voltage sags... which doesn't help if you have the thing already cranked up to full on.

 

[texaspyro]

I added the temp sensor and fan PWM to the LED controller and ran the 50 watt led at full power for an hour. It is on a Pentium P4 aluminum heatsink. Works very well. With a very low speed (24V, 70 mA fan) the heat sink gets pleasantly warm with a 30F rise above ambient.

Right now the sensor is being heated by the light beam. I need to whip up some thermal epoxy and mount it to the heatsink.

 

[texaspyro]

I went to the local ceramics supply store (http://www.trinityceramic.com, really nice people) and bought a pound of alumina(um) oxide for 5 bucks and whipped up a couple gram batch of thermal epoxy (mix with West Systems epoxy to form a paste for about 100 times less cost than the commercial stuff) and attached the the temp sensor to the heat sink (actually to the LED mounting plate). Remember that for a TO-92 LM34 sensor that the heat path to the die is via the leads... leave them as long as possible.

Aluminum oxide is not nearly as good a heat conductor as silver or metallic aluminum, but is not conductive. Plus mixing powdered metals with epoxy makes a rather nice rocket motor igniter... not something that one wants to do around batteries than can supply over 200 amps short circuit current.

The fan is a real weak 80mm unit, draws 70 mA at full power. Room temp is 60F, controller is now programmed to the turn fan on at 25% PWM at 80F (I give it a short burst of full power to make sure it starts spinning), linearly increase PWM to 100% at 144F. Works damn well. Fan turns off at 70F. Fan never got above 25% or so. I like it... Real unit temps will be set quite a bit higher.

Now I need to work out the best way to monitor the battery voltage, get some Bridgelux LEDs and reflectors, and package it up into the Flashlight From Hell. I'm thinking of making the reflectors removable so I can get a narrow beam thrower and a wide beam flood. Hmm, how about on a turret?

 

[texaspyro]

I looked into what it would take to add the battery voltage monitor circuit. The microcontroller ADC inputs are spec'd to have a +/- 1 microamp input leakage current. This would overwhelm the 4 microamp current that the 10 megohm voltage divider that I want to use to measure the battery voltage without adding much standby current. It would take an op-amp buffer to boost the divider current enough to feed the ADC input.

I measured the input leakage current on the microcontroller ADC inputs. The actual measured current was under 2 nanoamps! Looks like there will be no need for a op-amp buffer on the battery voltage divider. Even going to a 1 megohm/40 uA voltage divider, the battery standby life would be over 5 years (instead of the 15+ years).

 

[texaspyro]

This microprocessor in this LED controller is always on, always connected to the battery, and always monitoring a pot, the battery, a pushbutton, and a temperature sensor. When the LED is off, the thing draws less than 20 microamps. At that rate it takes over 10 years to deplete a single 2300 mAh A123 cell.

The same basic circuit would make a nice 1 microprocessor per paralleled cell group BMS. The LED PWM output could drive a MOSFET/bleed resistor for variable cell balance currents. The button input pin/fan pwm output pins would be a daisy chained optically isolated data link.

The micros would do LVC/HVC monitoring and cell balancing (either resistor bleed or switched capacitor). You could configure the thresholds, etc for any cell chemistry.

 

[mushymelon]

Man you have to build me one of these.

Add to cart 1 portable Sun

 

[texaspyro]

I hooked up a voltage divider that draws 5 microamps. Seems to work very well. It shutdowns on an undervolt or overvolt condition in 0.8 seconds and restarts about 12.5 seconds after the battery voltage returns to normal.

I am now playing with the code that boosts the PWM duty cycle as the battery voltage drops to maintain a more constant light output...

 

[texaspyro]

I have all the voltage monitoring code up and running. The unit wont start if the input voltage is too high (that might overload the LED), and it shuts down when it gets too low to protect the battery pack. It takes 15 seconds to recover from a voltage fault before it tries again.

I tested the thing with a ultra-precision lab supply and it is amazingly good. The low battery cutoff kicks in each time within a couple of millivolts of the desired point... Ideally one would like a cell level BMS, but my cutoff thresholds are high enough (2.8V per cell) that the pack would have to really messed up for it to matter.

The A123 cells seem to be very well suited to direct driving the LED. A 7S1P pack runs a 50 watt Chinese LED for over an hour. The LED current/intensity is within +/- 10% for the first 45 minutes.

I tried the PWM duty cycle booster that increases the PWM as the battery goes down. It does work, but I'm not sure that I like it. It limits the low end setting of the light once that battery starts running down. I'll probably drop that idea (or add some more logic so that it does not affect the intensity if it is set low). At minimum brightness setting, the thing puts out enough light to easily read by, but only draws 10 mA... you could run it 24 hours a day for 10 days.


I'm also thinking of spinning the circuit into an intelligent solar powered LED landscape light running off a single A123 cell, driving a 1 watt led. Charges the cell from a 2 watt panel using a maximum power point tracking algorithm. Does what it takes to give you light for the night with the intensity scaled to when it is most likely to be needed.

 

[texaspyro]

Hint... when measuring the voltage on a 7S A123 pack, make sure your meter probes are not plugged into the 20 amp jack... otherwise POW... A123 spits like 200 amps into meter! Meter stops metering. Oooh, that smell... don't 'ya love that smell... Open meter. Find and repair vaporized etch fat etch. Luckily the meter fuse was not damaged Don't 'ya just love Chinese engineering.