Flashlight Host
I used this
2 1/4 Mag D from FM which was a perfect length for 2 x Emoli cells. Here you see with rulers that battery tube length is the same as with an AWR Hotrater used on left.
I measured the distance from back of switch to lip edge of Mag (including back threads, but no tailcap) in mm, to give a good idea of how much battery room you have and how it
compares with other switches we have been using:
- AWR Hotrater - 159mm (typical of KIU type setup, retaining spring clip removed)
---- All others below have retaining spring clip inserted in Mag and switch pushed up against it from bottom ----
- AlanB - 160mm (From center contact screw 156mm)
- AW - Multi-level Switch - 164mm (From center contact post 162mm)
Battery SourceI used the Emoli 26700 cells from the Milwaukee V28 Tool pack as shown below. Two of them fit perfectly in the FM light by using a simple 1" (Internal Diameter) PVC pipe that was "clamshelled" with a hacksaw
(otherwise batteries are extremely tight and have no heat expansion/dissapation capability). Cardboard around battery must be removed, and PVC halves can be taped or not.
The setup Alan provided was for a 2S cell number, and these batteries have normal Lithium 3.7V rating, charged up to 4.2V. They generally store about 2600-2700mAh of output capacity. They perform well at high current demands which I needed for the bulbs I tested.
AW has his own version of these cells that are 20mm shorter which can also be used with similar characteristics.
It is important to avoid over-discharging these "safe Lithium Manganese" chemistry cells below 2.8V per cell, so Alan set the driver to shut off when the total voltage of the two cells drained to 6.0V
Bulbs UsedPhilips 5761 was primarily used. We did not want to start out with the higher watt/amp bulbs in testing this prototype. However, the 5761 was chosen because as the link shows, this bulb has a great lumen output, but a very narrow range & a requirement for a more robust current output than most Welch Allyn bulbs. It uses approximately 5.5 Amps at 7.3V
In the past,
many attempts to maximize this bulb without flashing it failed without using the AWR regulation, or various Rube Goldberg configurations with NTC's, resistors, diodes, etc.
Despite the higher quality appearance of the Osram bulb holder, it was not as effective at holding the G-4 size bulbs such as the 5761, so a slight "kink" was made by placing bipin pins on a wood 2x4 (glass off the edge), and lightly tapping down on a screwdriver blade poised over pins. This avoids damage to the glass envelope.
Other bulbs that could tolerate a peak 7.3V range were also tested, and all worked as expected.
These can be reviewed in this destructive test thread if desired:
Osram 64250, 64275, 64430 (& Tungsram 56580 version)
Welch Allyn 1111, 1160
Function & PerformanceGeneral ObservationsI used the settings Alan described in this post. He programmed this prototype before sending. I did not do any programming changes, and Alan is still tweaking and adjusting the current version of the software.
Most of my testing used the Philips 5761 bulb, consisting of 3 complete charge cycles of the Emoli cells, and numerous variations of buttons to test all programming. Run times with various output power settings were well over 30 minutes which correlates with a 5.5A draw on 2.6Ah battery source, considering my variable settings.
I took a 45 min dogwalk outside in a light mist/rain which exposed the light to humidity, but it did not get dripping wet. No problems were noted, except two alarmed skunks digging for grubs who did not appear to enjoy the lumens. :devil:
Two more complete cycles were done to test long run times on max (7.3V) with the Osram 64430 bulb which draws an even higher 6.5 Amps. Head was removed, and light used in "candlestick" mode. It got hot on upper outside, apparently mostly from bulb radiation, approximately 110-115°F (by Fluke 179 & Sentry ST-650 Laser Thermometer), but not too hot to hold. Each charge with this bulb lasted approximately 21-22 minutes on full power, and also correlated with battery capacity.
Neither of of these last two runs shutoff at 10 mins as Alan indicates he programmed. There was no noticeable drop off in intensity, but I did not keep a light meter to verify. I did use a faint object shadow that I could only see on Max, and it was always visible on these long Max runs.
First round, I manually turned it off after about 14 minutes as it got quite warm/hot in upper switch area on outside body. I let it cool about 5 mins, and again ran it on Max for another 6 minutes before it auto reduced to Med, then shut off after about a minute.
Exhausted battery voltage was always lower in the front battery in all instances. This next to last run had 3.01V & 3.41V in Front/Rear cells respectively. After 5 mins they recovered to 3.36V & 3.72V
The last of these cycles with 64430 started with 8.26Vbat (cells closely matched), and ran non-stop unattended on maximum for exactly 19::55 before switching to medium output. It then shut itself off at 20::48 sec. xxx minutes. Immediate removal of this different set of cells showed 3.08V & 3.32V recovering to 3.38V & 3.68V after 6 mins.
Button FunctioningI verified that the feel and reliable "clickability" under the rubber boot worked better than I expected. There was no problem doing double, triple, or 5x clicks to activate functions, and without false clicks. The button has a very nice and responsive feel to it. I pressed lightly, and firmly to test for possible button damage, and have subjected it to at least 1,500 clicks!! it appears to be working just as well as upon arrival. I like the idea however, that the button assembly could be replaced in case it is eventually damaged.
1) Memory stored light setting only happens when ramping up/down function is used. Once Alan told me that is the ONLY way to store a custom setting followed by off, then single click on to use stored setting, it worked 100% of the time.
2) Upon insertion of batteries, from off single click gives medium setting (Vmax + Vmin)/2. So medium voltage setting is (7.3 + 2)/2 or 4.65V
3) From off, Single click held down always gives low (2V) setting. This is a pretty sickly orange, but may be useful reading in a tent. Most incan bulbs just look pretty bad when set so low, and in this case I would likely bump it up a volt. Every bulb is different.
4) From off, once memory setting is made, single click always gives memory setting.
5) From on, single click always turns light off.
6) From on, single click held down, either starts ramp up or ramp down. Releasing yields stored memory setting for next single click on after turned off. Each Subsequent single click held down from continued on ramps in alternating up/down directions, and entire range from 2V to 7.3V takes 6 seconds.
7)(This is an edit change) With freshly charged batteries, from Off, double click always gives medium (4.65V); From On, double click always gives Max ....still troubleshooting. This is the behavior with the 6.5Amp Osram 64430 bulb. If I switch to 5761 5.5A bulb, then it gives Max from double click on or off.
Also am now noticiing with 5761, if I ramp up to Max, click off to store, the Max reading is stored unless I let it sit off, unattended for about 60-90 sec, then memory of Max fails as single click puts it to Med. However, if then followed within a few seconds by off & repeat single click, it now goes to Max, and keeps going to Max on single clicks, followed in close succession.
Now this is appearing to be an intermittent memory storage failure if left sitting off for 60-90 seconds. Repeat attempts to reproduce failure of memory to keep Max after sitting 60-90 sec as same batteries are run down do not happen...in other words after initial fresh battery install runs down, the Max memory stored reading works.
If I go back to 6.5Amp 64430 bulb with same batteries (disconnect/reconnect tailcap), now even this bulb from off goes to Max on double click, and memory stored Max stays working after 2+ mins in off.
This appears to only happen with freshly charged batteries
8) From On or Off, triple click always gives medium
9) From Off, 5x clicks gives medium flash and locks light off. Light appears to be dead.
10) Deactivate locking with another 5x sequence (again medium flash verifies change). Very handy to keep high power fire-starting torch lights disabled when sitting around 'flashlight noobs.'
Voltage/Amp measurementsI'm not sure how useful this will be, as my Fluke 179 does not do True RMS
1) Sticking probes in bipin holes gives 2.686V no matter which single/double/triple click setting.
2) With 5761 bulb in place, I carefully touched probes to sides of legs and got readings of:
Low - 1.024V
Med - 2.678V
Max - 6.921V
3) With 5761 bulb in place, I switched to Amp mode, removed tailcap, pressed Neg into bottom of battery, and Positive lead touching bare side & got readings of:
Low - 2.02A
Med - 3.08A
Max - 5.48A
Summary & Suggestions1) Overall, this driver KICKS ***! Especially remembering that it can be programmed to do all these functions at different voltages, and much more.
2) I like the feel of this button, and the method of the memory position storage. The functions of all the clicking more than meets my needs. I suppose some may want SOS or rapid blinking since so many LED's have it, but I didn't miss it. The lockout setting is brilliant and gives an added measure of safety.
3) I could not test the 60C (140F) temp shutoff, and I could not think of a viable/safe way to hook up parallel bulbs to increase current.
4) Not sure why the 10 min unattended shutoff didn't work, but I think that is the only issue I would consider a software bug. Maybe it thinks it is in the Pacific Time Zone.
5) The battery shutoff protection is just wonderful to have with these cells. I'm not sure why the back cell always drained lower than front, maybe has something to do with front rounded screw giving less contact area, or electrons flowing from positive end to negative? I might even consider setting to shutoff a bit higher, or your plan to change to shutoff at half of power might have a little better buffer. In all cases, the lowest any of my cells were showing opening and testing immediately after shutoff was 2.95V
6) Rounded screw contact on bottom of switch would take less battery space and give better contact if it was flat, rather than a round head screw. This will also improve current flow by giving more contact surface.
7) Thumbs down on the Osram bipin holder if you plan to use this for G4 pin size bulbs. It would be fine for GY6,35 (typical Osram 100W bulb). I'm thinking this wire is 20AWG, not 18. If so, that is not ideal for higher watt bulbs, despite the rating stamped on bottom.
8) The aluminum sled and open design appears to work very well in terms of fitting all the lights I have. It easily fit in all my mags. Not too tight, not too loose in terms of width. Being able to keep in the retaining spring gives extra protection from batteries pushing switch assembly forward which is always a concern if KIU assembly (i.e. in AWR Hotdriver) has the setscrew loosen. I always drilled a slight hole for setscrew to give it better anchoring, but no need with spring clip useage.
9) Warning should be given to people about static electricity, and grounding themself before handling switch or programming components. I think as long as people are careful, inserting the switch/programming stick onto the pins should hold up. It is a little hard to hold onto and make sure button switch is lined up with pins once assembly is inside light and set screw tightened. People may need to use tweezers and small light, but not a big issue. Switch removal with wire loop was clever and reliable. The rubber boot fit over switch in every light I tried it in.
10) The 200ms soft starting is wonderful. Love it like this.
11) I didn't yet check the voltage drive as you described in PM under direct drive scenario, but think I covered all the other items. Let me know if that is important, and I can still try it later.
12) It seems like there should be an even "Baker's Dozen" but I can't think of anything else except that I should now get to bed. I didn't proof this for typos. Thanks again!
