Haikelight MT03 II modding (driver)

I didn’t notice any PWM on the lower modes (tested with rain drops), so it probably reduces the duty cycle like a proper buck driver should.

With the coming XHP70.2 it will have a little more overhead Voltage, which is nice for when the batteries run low.

It has a timer step down for the higher modes.
But to be honest, i haven’t used the light that much before i decided to have a look inside.
That’s when i saw the tiny coils and the thin shelf, and i decided to mod it.

Anyway, i will replace the coils first and see where we go from there.

No, straight PWM'd FET, nothing fancy - cheap, simple. Basically what we've been doing for other single XHP70 lights. I have an L6 with a FET+1 driver working well. Might give the FET+single 7135 a try. On my 16X XHP50 light, the single 7135 is not behaving well - probably too much amps combined with voltage from a 2S3P driving 16 XHP50's.

@ EasyB

I never said that the lower mode is the cause for the saturation of the core ! I said that IF , in lower modes, you have a significant increase of current ,not only in led , but in the whole circuit , this is a sign that the inductor is not able to store , and give back the amount of energy required . Maybe I was not so clear in explanations, could be , English is not my native language...

First of all, " a sense resistor" is NOT the same thing as " a limit resistor" , and the difference between this two is about HOW they act in a driver.

A sense R , is "reading" the current , and the voltage across it , is sent to the Switching Regulator Controller , as a feed back , in order to hold the current constant. If the led is heating,it is gone "ask" for more current , but that means that the voltage in the feed back loop is going up , so the controller is gone reduce the duty-cycle to the gate of the FET , regardless of what duty-cycle is getting from generator ,to the point where the system is in good ballance. So , it is able to change , within some limits the duty-cycle. This is one of the limits I was talking about.

If , due to some reason , the feed back is exceeding the upper limit of voltage ( for ex. more than 0.25 V , depending on the type of the Controller ) , the internal architecture of it , is gone "latch", sending to the gate of FET a continuous pulse , as a 100% duty-cycle . In that moment the FET's gate is driven with a continuous voltage , and not pulses of variable duration.This is translated in loosing the modes.If the cause that generated the latch is not removed , you will have only the highest mode. That's why you can use a FET like a ON-OFF switch by connecting the gate to +V or -V , depending if it is a N or P channel FET.

Not all the drivers has these Switching Regulator Controller ( SRC) that could be designed for N channel or P channel FET transistors. Also , there are cheap SCRs and not so cheap ones , the last ones are designed with internal circuits , in order to prevent the latch . A more complex SCR requires more external components , so the cost of production will increase. So , the manufacturer are designing a driver as a final product , and not a one that could be improved by the customer...

About "limit Resistor" :

Some of drivers are just simple , Pulse Generator + FET with no feed back , and a Limit Resistor . The idea is that , at 100% duty-cycle , the sum of FET Rds on , plus the limit R, to limit the current to a safe value for the load :

V led= V supply - V r limit.

This type of drivers will not loose the UI, no matter how hard you drive them. Also you will see that the constant current is not so "constant".You can include here the , so much controversial DDdriver...

I'll continue tomorrow ..., damn , is already tomorrow , I need rest... See you...

I have been looking for toroid cores, but it’s probably too big.
So i will use the square ones that i have.

Adrian, can you tell me what replacement diodes i should buy? (not stacking)
And what replacement FETs should i buy?

Sorry for the delay Jerommel , the winter came back here and I had to clean my yard...Now, I'm back in business ...

I respect your decision , the new toroid should do , just keep an eye on the heating...

About the diodes , I see that you do not want to stack its. Good call , I'm not doing like this either , but in some cases I had to do that ( no space at all ).

The Schottky diode : SK 86 ( Diotec Semiconductor ) should do the trick . I'm using this diode for few years now , and never failed on me.Is a little bigger than ss54.

The next diode , from the link :PDS1040-13 Datasheet(PDF) - Diodes Incorporated , ( just click on the image on the upper , left corner ) , 10 Amps , are the best I ever used ,and most of its were recovered from power supplies ! Even Fenix are using these for their TK 75 ! This one is easy to mount , regardless the distance between the board traces ,cause of the katode that is used as a heat sink , and the height of it , less than 1 mm ! I'll choose this one...

On the FET side , here the possibilities are almost endless...Is all about if me , or somebody else, ever used some in their modding attempts. I hope that at least few more guys to chime in , with their experience...COME ON BLF MEMBERS , THIS THREAD WILL BE , LATER , YOUR GUIDE IN MODDING THE HAIKELITE HT03II !!!... What I used and had good results are the following :

The old 436 :aod436_Rev4_rohs.xls - aod436.pdf , this one is a "mule" , reliable and with good parameters.The order of the pins are the same like the ones from your board :gate , drain , source , from left to right .Is a little bigger , but can be soldered in the space.

TPCA8028-H Toshiba Semiconductor, TPCA8028-H Datasheet , smaller , odd terminals (as a IC) , reversed order :123 pins=source , 4 pin=gate , 5678 pins=drain

Excellent Rds on , great power dissipation with no heat sink ( 1.6 - 2.8 W) . Used it in a buck driver LX-9028, 12.6V ,3X XM-L2 serie , current 4.5 Amps , working as a dream...

PH2925U N-channel TrenchMOS ultra low level FET - PH2925U.pdf , suitable for drivers with 2S batteries , not for single cell drivers (the threshold gate voltage a little high ) , very reliable , good dissipation capabilities .Reversed pin order , but not a problem , just flip it.

These are the ones that I used in more than one tests. For sure there are a lot more , but I do not know , or test ...

I do have a strong suggestion , based on my trials and errors ( A LOT OF ERRORS... MORE THAN SUCCESS !!!) . Do all the test on ONLY one channel , and after you got the best result ,operate the other 2 . The resistor mod should be the last one , because , most of the time , due to the spreading of components parameters ,you will have differences in led's current , that can be equalized thru resistor mod.

Good luck , just wait to hear from you...

Hey! sorry to derail this topic somewhat but you seem to know what i am trying to find a answer to.

What value on the inductor is the one to match if i am trying to find a new better one?

I am modding a [H2-C driver](http://www.kaidomain.com/p/S026504.H2-C-22mm-1_5A-1-or-2-cell-5-Mode-Boost-Driver-Circuit-Board-for-Cree-XHP35-XHP50-12V-(-1-pc) it has a 2.2 uH inductance inductor now. Will any 2.2uH inductor work with this driver?

Trying to find something here that will handle more amps then the original one.

:smiley:

@ Tjhosan

In theory any inductors with the stated value will work ! Even with an other value , up to 2-3 times bigger.., BUT , ( the omnipresent BUT...) , there are some things to think about...

The value of a inductance is given by the inductance factor (different for each material used for the core) , number of turns , size of the wire , size of core , the way the turns are made (parallel , crossed ...) and Quality factor ( given by the reactive power and active power in the inductance).., too much theory , right ?

Lets go to the practical side !

The size of an inductor is the one that tells you how much current is able to deliver .In a word , bigger size , better behavior . A bigger size is due to the increased diameter of the wire used that has to fit inside the core ! If you measure 2 inductors of same value but different sizes you'll find that the bigger one has a lower resistance ( in ohms) ! And here we are going to find an other parameter of a inductor : lose resistance , that should be as small as possible in order to minimalize the loses. The size of the core is important , because this one is the one that "stores" the energy . It is the same thing like an transformator from 220 VAC to , lets say 12 VAC , where the size of the core is telling you how much power can you use.

In the last decade , the passive component manufacturers started to make inductors not by wire but by pressing under controlled temperature of mixed metalic powders. Good for miniaturised electronics but not so good for power electronics !

Avoid these in high power drivers !!!! As I said before , even is not very easy , use DIY inductors , wired with a wire that can sustain the current without significant heating , they are easy to solder in any position , can be mechanical sustained with silicon compound and easy to change during the tests...

Ok i think i get it :slight_smile:

Thank you!

You're welcome , anytime...

Regarding making your own inductor:
Here is a calculator:

The “52” core material you mentioned has a relative permeability of about 250. With a reasonably sized torroid, it only takes 3 turns to get an inductance of 10 micro Henrys. I guess one could use large diameter wire and reduce the inductor resistance to negligible levels (less than 5 mOhms). How critical is the exact inductor value?

Power loss and heat generation due to the flyback diode forward voltage is significant. I read about “smart diodes” (made by Texas instruments for example) that have some circuitry that switches a FET on and off, I think. They can have effective forward voltages of 0.05V, compared to 0.4V of a typical schottky diode. Are these ever used in voltage converters to reduce losses?

I wouldn’t even know where to find toroid cores with specific core materials…

No updates on this from me yet, i’m busy with other things at the moment.

Hi , EasyB !

Good to see that this Jerommel thread triggered so many questions and opinions !

I 'll try to answer to your questions briefly , please forgive me , and do not forget that this thread was intended to help Jerommel to decide himself , what path to chose for modding the HaikeLite , but as ussually, all the derail from the subject are in the best interest of everybody...

How critical is the exact inductor value ? In Math , if 2+2 is not equal with 4 (!) , we have a BIG problem ! In drivers we do not work with fixed , critical frequencies , like in NASA communications.., for years I tried to get the exact value from calculations , until , one day , I changed an inductor in a chopper , with one with the value 5 times bigger than the original ( was a mistake that I discovered too late , after I started the chopper ).., and guess what ? everything worked perfect !!!

Since then , (doing a lot of tests on DC-DC converters), I understood that a critical value is extremely relative , and only values of teens times bigger or lower will cause a malfunction . Oh , man , I said I'll be brief , and look what happen ! Hm , I 'm getting old and talking too much...sorry !

So , the value is not critical , but the quality and size is !

About the "smart diode".., this is a beautiful dream that will NEVER be true ! i mean that this kind of diode was not , is not , and will be not used in converters.

In fact is not a real diode and can be used only with clear DC voltage . Was created especially for photovoltaic panels. Is working as a micro fast switch , with fixed frequency !!! Try to imagine a convertor within a convertor , syncronized for EACH driver built !... A nightmare ! A p-n junction is a diode, since the beginning of electronics , can be improved a lot , ten years ago Schottky was a revolution , and I'm sure that will see better things in classic electronic...

Stop , I think I bored you enough !... Cheers !

Nobody is pushing you to do something ! Let the things to settle in your mind , and when they will be ready you will feel the need to do them.., at least I'm like this !

When the time came, you'll know...

Mean time a link for you :180009678 Vacuumschmelze Ferriet ringkernen / ferriet ringen | Mouser Nederland .

You can ask them about , an information cost no money ...

Okay, i found Schottky diodes 80SQ045 Vf<0.44 V, little bit better than Diotec SK84 Vf <0.55 V (sold in same local webshop)
Both are 8 Amperes, i think it’s enough.

How about the 3R030 FET? Yes, from the BLF driver :smiley: I have 3 of them available.
I think it’s this one: http://assets.nexperia.com/documents/data-sheet/PSMN3R0-30YL.pdf but i’m not sure.
It says it’s 0.003 Ohm.

Those toroid cores are expensive! :open_mouth:

(on another note, i just bought a slab of pure copper 60 x 100 x 4 mm on ebay, i could make a shelf from it, but need to find ‘someone somewhere’ to cut out the 1.5 mm aluminium shelf with a lathe, so i can replace it with 4mm copper.)

Oh , those FETs are a good find.., I should remember about them , but , as I said, I recommended only what I tested ! Excellent , first time I'll have the opportunity , I'll try them.

Ya , you're right the toroids , at about 5 euros are expensive , I didn't look at the prices...

On the other side , the copper could be a good solution , and this mean a disc , PERFECTLY press fit , with the head heated to around 100 degrees (or above) , to be sure that , when the head is going to heat , you'll have the best heat transfer.

A big mass of material doesn't mean a better dissipation , but the SURFACE does ! So . you can calculate easy ,if the original shelf is able to transfer the heat outside the head , where the real cool is done ! I'll give you an example and you can do this with the real measures of your light head.

Lets say : D (diameter of the shelf )= 40 mm. The surface (S) is:Pi (3.14) X square D , all divided by 4. ( 3.14 X 1600)/4=1256 square mm= 12.56 square cm.

This is the minimum surface that your shelf should be able to "touch" with your light head. Translated in plane geometry , you will have a rectangle with the big side equal with the length of circle diameter of the shelf , and the smaller side equal with the thickness of the shelf .

L(length) of circle = Pi X D = 3.14 X 40 = 125.6 mm

Thickness of shelf(l) , lets presume =2 mm.

From here we calculate the surface of the rectangle . S =L X l = 125.6 X 2 = 251.2 square mm !!! So , this is the surface of efficient transfer to the head. Compared with the 1256 sq. mm of the shelf we see that a shelf of 2 mm is too thin , and the shelf will NOT be able to transfer the heat , efficiently...

We 'll need a minimum of 5 mm thickness for the shelf (1256 / 251.2 = 5 mm)

Yeah, i have a couple of BLF drivers that i don’t use, so i can use the FETs from them and later on put a bunch of 7135 in stead maybe.
I personally don’t like DD drivers, but a couple of Astrolux lights came with those.

Some are cheap, some are very expensive. I guess the expensive ones are better = needs less windings. but when i look for toroid cores on AliExpress or Ebay, i really don’t know what i should buy…

The plan is to remove the shelf and put a 60mm copper disc in there, which will have a lot of surface contact with the remaining outer rim.
Press fit would be great, but i can screw it on the outer rim too.
(hope you understand what i mean).
But that way i don’t feel the need to incorporate the reflector in the thermal path, so i can later replce the reflector with TIR optics if i want to.
TIR optics will be better for the XHP70.2 which will probably have more tint shift than the XHP70.
We’ll see. :slight_smile:

When I said press fit, I didn't think to a literally pressing the copper with a press machine ( this is hard to do and , also, could damage the shelf and ruin the head !!!)

I was thinking of that 60 mm copper disc , with a real diameter of 60.010 mm , and the shelf machined , with the diameter of 60.000mm.The aluminium head heated somewhere between 100-200 celsius , in a electric owen , ( not to much , in order not to change the colour of anodisation ) , and the copper disc , refrigerated in the freezer for few hours ( -18...-24 cesius) . In that moment , the contracted copper disc will slip easily in the dilated machined aluminium shelf .At the room temperature you will get an excellent contact between copper and aluminium . I've done this, couple of times and IMHO , is the best way to do a professional job . Do not forget that never you will have the temperature of the copper to minus and the head at positive temp. , in normal use of a flash light !!!... If you can find a lathe, be sure that the man that is working on it will know what to do , if you'll explain him what you are going to do.

I know its a little different application but freezinf metal does work. On cars if I have a ball joint that’s being a real major pain in the butt, I’ll get it nice and well frozen and it will give enough clearence and expand right back afterwards. Works every time it contracts just enough.