Haiku XP-G

[Drywolf]

Love my Haiku!

 

[McGizmo]

nbp,
You needn't apologize for anything. I am sorry you are experiencing these problems and I don't like providing lights that can experience them. As I understand it, the control chip which decides when and if to switch levels has a minimum operating voltage of 1.8V. What I believe happens is that under load and given the resistance in the system, there are times when this control chip does not get the voltage and power necessary to perform its desired function. You miss a shift as it were. Cree does not bin the LED's by Vf but I suspect the batch of warm LED's I am working with have a typical Vf higher than the stock XP-G's I am working with. Back when we were working with single level drive converters it became apparent that 500 mA to the LED was a sweet spot when coupled with a primary CR123 battery. I think that is still the case but I have tried to respond to the demand for higher current and higher flux by driving the LED's at higher levels and certainly the LED's can handle it.

It doesn't matter whether you use momentary or latched switching. Changing levels requires a very short off duration.

I am willing to explore any number of possible solutions to your present problems and of course that includes simply taking the light back and providing a refund.

 

[nbp]

Good evening McG,

I tracked down a couple spare SF primaries I had put in my bugout bag and gave them a try. It is slightly better, but still hit or miss. Certainly not the crisp mode changing that I experienced with the Haiku I borrowed some days ago. Perhaps this particular light is a bit fussier than the other one. :shrug:

That being said, I am happy with the light as a whole, and would very much like to make something work short of just return/refund. If you are agreeable to it, I would be interested in getting the light or just the LE swapped out for the cooler tint xp-g, which it seems is not prone to this malady. Let me know if this is a possibility, or if there is some other solution that is preferable to you. Feel free to email me to hammer out details if you prefer that method of communication.

Thanks again for your help and speedy response, it is much appreciated. :thumbsup:

Nick

 

[fyrstormer]

What is the voltage on the SF primaries you used? If it's below ~2.8V, the batteries will have so little current-generating capacity they won't be able to properly power the light even if the voltage is high enough.

 

[milkyspit]

What is the voltage on the SF primaries you used? If it's below ~2.8V, the batteries will have so little current-generating capacity they won't be able to properly power the light even if the voltage is high enough.

Guys, sorry for butting in but thought it might help in some small way. Testing a 123 cell by voltage is notoriously unreliable. The better test method would be to test its ability to push current for a very brief instant. One can do this in a pinch by setting a multimeter to the 10A (sometimes 20A) scale and shorting the cell FOR A FRACTION OF A SECOND: as quickly as you can note the reading, stop! One would hope to see a reading of 7-10A though the specific value will vary by brand of cell.

A FAR BETTER WAY to do the above test would be to test the cell on a ZTS battery tester. The ZTS testers essentially do what I've described above, but in safer and more sophisticated fashion. I was skeptical about these testers for a long time but after a few years of using them, I'd have to say they do a great job of assessing the state of 123 primary cells. (Note that the bar graph which attempts to estimate percentage life remaining in the cell is probably NOT accurate, but it doesn't matter: you're looking for cells that test at 100%... anything less means the cell is on the way out.)

Sorry for interrupting, Don.

 

[nbp]

I don't have any fancy testers to use on these cells.

But the fact is that I have tried multiple new cells from two reputable Made In USA brands, and they don't do the trick. I don't think it is reasonable to have to sift through a pile of batteries looking for one that works. Besides, even if I did, how much would I be able to use it before the cell was no longer strong enough and the problem resurfaced?

It's an awesome package, but by Don's own statement, it could use some tweaking. I am highly interested in making the Haiku a part of my EDC kit, albeit with the more stable emitter/ driver setup.

 

[Obijuan Kenobe]

Well, just to make sure it is said clearly, this is a statistical anomaly. Don is the man as far as reputation and track record.

The Haiku design is solid. Your particular Haiku is funny.

I'd be willing to bet that when you send it back, Don finds something fundamentally wrong with it...fixes and returns you the Haiku you expected to get and Don thought he originally sent.

obi

 

[fyrstormer]

Guys, sorry for butting in but thought it might help in some small way. Testing a 123 cell by voltage is notoriously unreliable. The better test method would be to test its ability to push current for a very brief instant. One can do this in a pinch by setting a multimeter to the 10A (sometimes 20A) scale and shorting the cell FOR A FRACTION OF A SECOND: as quickly as you can note the reading, stop! One would hope to see a reading of 7-10A though the specific value will vary by brand of cell.

Well, yes. Testing current is important too, if one has a multimeter that can handle multiple amps. I got one that can handle 10 amps and it only cost $10 or so. However, amperage is dependent on voltage and resistance though, so high voltage implies high amperage assuming one isn't attempting to push electricity through a known insulator. That being said, the reason I specified 2.8V is because I've noticed my 123 primaries tend to fall off a cliff in terms of amperage when the voltage is lower than ~2.8V. It's a good guesstimate for when it's time to toss the battery.

...The Haiku design is solid. Your particular Haiku is funny.

I'd be willing to bet that when you send it back, Don finds something fundamentally wrong with it...fixes and returns you the Haiku you expected to get and Don thought he originally sent.

I agree. I sent a Haiku in for service once and it turns out my emitter combined with my current-sensing resistor just happened to add up to abnormally low output. Don replaced the current-sensing resistor and all was well.

 

[nbp]

I genuinely hope that you guys are right and that it is a simple fix. I'll leave that to the master.

You know why this happened to me though, right? Because I started that thread about McG reliability. I was guaranteed to get the one "funny" light in the bunch.

 

[milkyspit]

Well, yes. Testing current is important too, if one has a multimeter that can handle multiple amps. I got one that can handle 10 amps and it only cost $10 or so. However, amperage is dependent on voltage and resistance though, so high voltage implies high amperage assuming one isn't attempting to push electricity through a known insulator. That being said, the reason I specified 2.8V is because I've noticed my 123 primaries tend to fall off a cliff in terms of amperage when the voltage is lower than ~2.8V. It's a good guesstimate for when it's time to toss the battery.

The problem is, multimeters measure voltage across a resistance measured in the megaohms... even the low end multimeters, although it is true that in general, the better the meter, the higher the resistance. No flashlight will EVER present such a high resistance to the cell, not even remotely close.

But your observation on the ~2.8V threshold was an interesting one, so I pulled out a bunch of cells around here, both fresh and depleted, to see if I could use the same shortcut. It certainly would be convenient. I went into the test hoping it would work.

Here's what I found:

* I tested 4x new SureFire 123 cells, all testing 100% on the ZTS, and found voltages of 3.06-3.23V.

* I tested 19x depleted SureFire 123 cells from my discard pile, all testing 0%-10% on the ZTS, and found voltages of 2.83-3.13V. All but two cells were above 2.93V. Half the cells were above above 3V. I took the highest voltage cell from this group and tried to run a Novatac 120 flashlight. It was unable to run at anything higher than moon mode.

That's the problem with relying solely on voltage.

Anyway, I was only trying to help.

 

[fyrstormer]

Interesting. That's more data points than I used, and your ZTS multimeter is better than my $10 one. *shrug* I'm inclined to believe your results over mine since you have better equipment.

 

[McGizmo]

nbp,

I will be happy to send you a replacement LE using the cooler, stock bin of XP-G LED's. You also have the alternative of going with a High CRI LE which uses the Nichia 119 LED. Frankly once I started offering the High CRI version I figured this would be the choice for those where quality of light was a priority over maximum lumens. I personally prefer a cool tint over a warm tint but that is just a preference obviously. When it comes to actual color rendering, the tints, unless way off white are secondary. The 119 LED is driven at a lower current which makes it easier on the converter as well as enhancing the runtime of the light.

I have also used a ZTS testor for years and found it invaluable in trouble shooting and ruling out a battery in a problem light.

I suspect a number of the Haiku owners are using R123 cells and from what I gather, the shifting through levels is basically a non issue for them. It has been my experience with my personal lights that as soon as I see some missed shifts it is an indication that the battery is on the way out and I kind of appreciate this form of indication. However if the light can't make use of a good portion of the battery's energy without issue the situation or system needs to be rectified.

 

[nbp]

Hi Don,

A cool xp-g would be great! I really like the beam profile with this emitter. Thanks again for being so helpful. Despite the tiny glitch, I have nothing but good things to say about working with you.

Please let me know what I should do with the other LE. If you want it back I can send it out when the new one arrives. Also, if there are any tricks to swapping them, that'd be great. It looks simple enough, but I'd hate to goof anything up.

I might grab a couple AW RCR123s on my next Lighthound order too if it seems they work better with the Haiku than primaries. I will take whatever suggestions you can give me.

Thanks again!!

 

[McGizmo]

nbp,
I just put a replacement LE in the mail to you. Please return the original and let me know what I need to send you to cover the return postage. Replacing the LE is pretty straight forward but care must be exercised to keep the O-ring that separates the LED from the rear of the reflector in place. If you hold the head upside down gravity will keep the O-ring in place as you screw out the original LE and then screw in the new one. Once the new LE is firm against the shoulder inside the head and seated, the O-ring is again captive.

 

[nbp]

Don, you ARE the man! Thanks so much. I will get the other parts out to you as soon as the new one is in. Don't worry about postage, it's nothing.

Even with a glitchy mode change, this Haiku is really growing on me. I love how it fits my hand, the ergonomics are just lovely. I can see after only a day why some of you have many McGs! I am already trying to figure out a way to raise funds for a Mule to compliment the Haiku. :shakehead I will have to be very careful how much time I hang around in this subforum I think.

 

[nbp]

Hi Don,

New heart came today, went in easy peasy, thank you so much. You rock! The other one will go out to you on Monday. I'm lovin' this light. I will definitely be saving for more of your creations. XM-L Mule here I come.

 

[ChipZero]

I got my Haiku XP-G R5 today. If i buy a 2 x CR123 Pak could i use it without damage the my Haiku?
Thanks for help!

 

[jumpstat]

I got my Haiku XP-G R5 today. If i buy a 2 x CR123 Pak could i use it without damage the my Haiku?
Thanks for help!

Nope don't think so, unless you use a 17670 cell in the 2x123 pak if you could find one that fits.

 

[ChipZero]

@jumpstat thank you!

 

[jumpstat]

@jumpstat thank you!

Personally, the single cell Haiku with the 185 lumens on high is more than enough. Of course, there are many more other lights offer much higher output. Pop in a R123 cell in the Haiku for guilt free lumens..