Some questions about the Shark driver

[Greta]

Continued and copied from HERE.

Walt17506-13-2007 08:43 AM
Re: Some questions about the Shark driver

Amonra, did you ever find out what happened to your Shark? I'm currently having the same problem. Vf in is = to Vf out. No boosting is going on.

When I first turned it on it would be bright for a split second then go to a much dimmer lever. I only turned it on a few times, and never had it on for more then a few seconds. Yet as far as I can tell, the Shark is fried since it is no longer boosting.

I'm trying to drive 3 Seouls from a 4 AA nimh pack. The leds are 3.25v each, so I would need 9.75v total. The 4.8v is almost exactly the normal operating voltage. (9.75/2= 4.875) It is also above the minimum voltage. (9.75/3=3.25) Did I pull too many amps and fry the board? When I tried to figure out the amps I got a headache and stopped. Too much math at one time.

waTom06-14-2007 09:06 AMRe: Some questions about the Shark driver

This might be a dumb question but my lack of knowledge makes me asking:

I want to run 4 Crees from 3 18650 AWs and I do need to jumper J1, right?
It says that this will set the output to ~22V but I just need ~14,8V?



dat2zip06-14-2007 10:03 AMRe: Some questions about the Shark driver

Quote:
Originally Posted by waTom (Post 2033300)
This might be a dumb question but my lack of knowledge makes me asking:

I want to run 4 Crees from 3 18650 AWs and I do need to jumper J1, right?
It says that this will set the output to ~22V but I just need ~14,8V?



Setting the voltage limit has nothing to do with current regulation. The shark voltage limit is for safety only and to protect the shark from killing itself. It only applies if you do not connect anything as a load to the shark in which case it will limit the output voltage to a safe level. In normal operations the Shark will regulate the current to the LED based on the sense resistor and where the trim pot is set.

FOR LED applications ALWAYS jumper J1.

If the Shark is not regulating check the thread in this forum regarding the Shark advisory and verify you have a good Shark converter board.

Wayne

dat2zip06-14-2007 10:22 AMRe: Some questions about the Shark driver

Quote:
Originally Posted by xiorcal (Post 1999368)
Hi Wayne,

I'm also interested to know what happens under 2.7V. AW's Li-ion cells' low voltage cutoff can be duped by low discharge currents. Is this 2.7V a point where the Shark will lowV-cutoff when running?

Thanks

The 2.7V was extracted from the datasheet. What does this mean?

This means that typically the converter will stop working around this point. Where? who knows. it could be 2.5V on one unit, 2.3V on another unit or they all stop right at 2.7V. Again, this only means that the converter at 2.7V will work correctly and below this number who knows. The datasheet number only guarantees the converter will work at 2.7V and above.

What does it do below the cutoff of 2.7V. Agan, this is intrisically determined by the semiconductor design and how it stops working. Typically converters will continue to run at a reduced level once they fall out of regulation. How much or how long is totally determined by the design of the part of which none of us actually have information to.

I also want to mention that the Remora hold time is setup for battery voltages of 6V or higher. For a single rechargeable battery application one needs to increase the uP supply capacitor to make the hold time longer and make the Remora UI to work below 6V.

Wayne

waTom06-15-2007 03:18 AMRe: Some questions about the Shark driver

Thanks for the "enlightenment" Wayne

Nereus06-29-2007 12:30 AMRe: Some questions about the Shark driver

There is lot of discussion in this thread about how to replace the Shark trimpot with external control. I created external control circuit that replaces trimpot and makes Shark a three mode converter. You can find a detailed description with schematics in this thread.

-N

270winchester07-10-2007 04:43 AMRe: Some questions about the Shark driver

so question:

I'm extremely new to the shark game. Say I just want a single level bright night driving 4 Seoul USWOH emitters with an 8 AA to 2D adapter using 8 1.2v NIMH AAs, what would the default drive current be? I have no itention of messing with the trimpot or the external controls

Nick

Oznog08-13-2007 10:48 PMRe: Some questions about the Shark driver

I would need to power this off a car's electrical system.

While that's up to 14.6V when running continuously, voltage spikes are likely to occur at times. The starter can probably make some impressive ones.

Is 20V an absolute end-all max, or can it tolerate brief spikes well beyond this? What IC is that? I could check the spec sheet.

So is the heat generation really that high? I will need to boost from 10.5v-14.6V to 21.6V @ 700mA of LED load. There's not a lot of opportunity here for heatsinking though. What's the heat source to worry about? The IC or the inductor?

[acourvil]

The only caveot is to ensure you have a boost configuration under all conditions. Letting Vin rise higher than Vout will put the converter in DD mode and will most likely Fry the converter board, LED and possibly the battery. That's not a good situation.

Is this true for the whole range, or just at the high end?

For example, if I use 4 C Li batteries to run 5 Cree Q5s in series with the shark and an external pot, the max in is 16.8, which tanslates to about 3.3 per emitter. It seems like at the low end, if it drops into direct drive it would simply act as a floor on how low the emitters can be driven.

The language above is pretty clear about "all conditions" but I'm not sure I understand why there would be a problem at the lower end.

[dat2zip]

Is this true for the whole range, or just at the high end?

For example, if I use 4 C Li batteries to run 5 Cree Q5s in series with the shark and an external pot, the max in is 16.8, which tanslates to about 3.3 per emitter. It seems like at the low end, if it drops into direct drive it would simply act as a floor on how low the emitters can be driven.

The language above is pretty clear about "all conditions" but I'm not sure I understand why there would be a problem at the lower end.

Yes, you are correct. Under fresh battery conditions and if the wind is blowing the right direction it is more than safe to bypass the converter in DD mode where only the output series diode is in the LED path between the battery and LEDs. As long as this condition does not draw excessive currents it's OK. Under this condition your low could be the DD level and not the level set by the pot or Remora.

The Makita 7 up mod that we mocked up had this problem. I ended up putting a couple of high power rectifier diodes in series with the battery to knock it down a couple of volts. While it wastes power I gained back low and I considered this a reasonable compromise as I rarely use high and of course the Makita LXT battery is rechargeable so it's no biggie to pop it into the charger or swap batteries.

OK, the wind direction has nothing to do with this.

Wayne

[acourvil]

O.k., thanks. I'm kind of new to this, so it's good to know I wasn't missing something.

Now if I can just figure out a way to make my fingers smaller and my eyesight better for soldering on those itty bitty boards . . . .

[dat2zip]

If you can afford a stereo microscope they are worth every penny.

Here's one on ebay.

Wayne

[kenster]

Wayne, I was expecting a much higher price when I clicked on the EBay link. That Trinocular wouldn`t strain my wallet too bad and since my eyeballs act like they are getting old or something and my cheap magnifying lamp isn`t the visual aid that it used to be ...... sWell...... I believe I will grab one of these Tri-gadgets. It`s cool it has a built in stereo so I can ... "Rock Out"! Don`t worry, I hear things all the time so I`ll never know the difference anyway. Hehe!


Thanks for helping me spend more money on tools! Uhhhhh, I think???

Ken


*** EDIT ***

Is the Sharks voltage boost you mentioned still in the mix?

[dat2zip]

You only need a stereo or binocular version. That was the cheapest link I found using 20 seconds or so to take a peek. I'm sure you might find cheaper ones.

Be aware of fixed bases like the one in the link. The working vertical distance is limited by the bottom base which is attached to the whole assembly.

If you want to look at things at different vertical heights the boom stereo microscope is more versatile and more expensive of course.

That's what I have on my bench setup is a boom microscope. Since the scope head can be moved out over free space you can raise/lower the head vertically to focus on larger assemblies and still put the focal point on target. It's also the greatest tool to remove splinters. I have yet to figure out how to get my butt under the microscope while still looking through the eyepieces.

Wayne

[dat2zip]

Here's a picture of my work area after just cleaning it up. It's never this uncluttered!


Wayne

[dat2zip]

Wayne, I was expecting a much higher price when I clicked on the EBay link. That Trinocular wouldn`t strain my wallet too bad and since my eyeballs act like they are getting old or something and my cheap magnifying lamp isn`t the visual aid that it used to be ...... sWell...... I believe I will grab one of these Tri-gadgets. It`s cool it has a built in stereo so I can ... "Rock Out"! Don`t worry, I hear things all the time so I`ll never know the difference anyway. Hehe!


Thanks for helping me spend more money on tools! Uhhhhh, I think???

Ken


*** EDIT ***

Is the Sharks voltage boost you mentioned still in the mix?

??? Voltage boost? If you are referring to raising the Voltage limit higher that is done. The remaining few Sharks of yesteryears batch is down to a few and we are just dipping into the new boards. I just fired one up out of the new batch to confirm they are working correct. I measured open circuit no load of 26V. There is a resistor removed and J1 is no longer needed. It comes pre-configured for 26V and the 11V option is now 26V by default obsoleting the need to jumper J1 and change the feedback voltage divider.

The output capacitor was changed from a 4.7uF 25V to a 1uF 50V rating.

Wayne

[kenster]

I have yet to figure out how to get my butt under the microscope while still looking through the eyepieces.

Wayne

Well, what can I say but.......... I`m just glad you don`t have a Trinocular and a camera in case you to achieve and decided to post some pics of your.......................... I`m outa here!

[kenster]

Folks, I just realized the new 26volt version of the Shark is now available at ths Shoppe! Thanks Wayne and I will be placing an order soon!!!!!!!!

[devil77]

Hello,

i want to build an new bike light with 4 Seoul P4.
Can drive the shark with remora ui board the follow parts?

4 x seoul p4 (this one: http://www.led-tech.de/de/High-Power...980_55_78.html )

2 cell li-ion battery Vmax=8,4 volt, Vmin=6 - 7Volt.

is this possible?

[dat2zip]

devil77,

That configuration should have no issues.

Wayne

[merlocka]

Hello,

I'm thinking about a 5 Q5 Cree setup.

I would plan to drive at ~600-700mA, for a Vf of ~3.4v (according to http://www.candlepowerforums.com/vb/...&postcount=141)

5x3.5=17

17/2 = 8.5
17/3 = 5.66

I'm thinking this close enough for 2 AW's "C" Lion's?

I calculate about 2.2A input current (sound right?), think this would be ok?

Has anyone tried this setup? I just wanted to ask before I start ordering stuff :-)

[LED Zeppelin]

merlocka,
I used a Shark and external pot to drive 5 Q4 Crees, and at full current of 930 mA the Shark was taxed and would step down in brightness after several minutes. On 3 Li-ion cells I had no such problem.

If you plan to drive at a lower current it may work out fine, but a rule of thumb is to try and keep Vin above 1/2 the Vf of the LED string.

[dat2zip]

I have some ideas on two heat sinks. One for the diode and one for the bottom of the board. I am in the process of mocking these up and will share the results in bit.

Here are the two drawings in case you'd like to make them on your own.

The diode heat sink sits on top of the diode and bends over to touch the GND vias next to the diode. The underside of the heat sink is greased first afixed to the diode and the end of the bend is soldered to the via array to wick away the heat to the bottom of the board.



The bottom side heat sink is a U shape copper sheet that is soldered to both via arrays and provides a flat surface that can then be thermally epoxied to a heat sink.


Here are the two parts I made.


Mounted on the board.




The diode heat sink works well. The diode doesn't go ballistic as it would without the heat sink. The IC is still huffing and puffing. The temperature gradiant from the bottom of the board to the heat sink was still very poor. I need to remove the bottom heat sink and check to verify I filled all the vias under the IC and then re-assemble and test again.

Under the current configuration I can only pull from the power supply 2.5A or so before the IC gets unhappy. That's may be hitting the internal thermal limit still.

Wayne

[dat2zip]

I removed the board, cleaned off the epoxy and removed the backside heatsink and I was correct that I did not fill the holes on either lot of vias. The air in the holes make great insulators.

I tried real hard to solder the IC vias and even leaving the iron on for a long time I still believe I have trapped air in the vias. I can't be certain of this, but, the temperature measurements still show the top of the IC getting up to 90C and the large heat sink is just getting luke warm.

I had picked up from the surplus store some real nice processor heat sinks with thin thin fins. They are like 4" X 4" X 0.25". I use these to mount test LEDs on for bench testing.

I was able to cut them on the bandsaw and made a little square one that I glued on top of the IC. I saw a drop of ~20C from 90C to 70C after installing the topside heat sink. I was only able to lower the supply until the supply current reached 3.1A. After that I think the IC was hitting the switch internal current limit and it started behaving strangely lowering the output current from 1A to ~700mA or so.

If I get a chance I will do some more experiments, but, I think that I have learned that filling the vias holes under the IC seems rather straightforward turns out to be rather complex and if not done correctly will create air pockets and render heat transfer from getting to the bottomside.

Wayne

[dat2zip]

I ruined two Sharks in the quest of filling the vias with solder. I think I finally on my third attempt was able to fill the vias with solder. They actually don't want to wick in solder or flow in for some reason.

I had to use 2331 water soluble flux to get the joints and vias to clean real well and the flux then with gravity tended to flow in as the area heated up. It took a good heavy duty soldering iron, proper flux and solder.

You can get the flux in a pen KE1808 from Digikey or elsewhere.

Just blobbing on solder definitely doesn't do anything to fill the vias. It was very difficult.

Once I got the vias to fill I soldered on a thin copper strip bent into a U soldering the middle of the U to the PCB.

If the top of the IC reflects the temperature of the power pad on the bottom I measured about 11C difference between the top of the IC and the joint of the copper thin sheet and the PCB. Since I had the Shark driving 6 Rebels at 850mA and I'm guessing the Vf of the Rebels are around 3.4V would make the stack of LEDs dissipating 16.8W. That computes to less than 1C/W drop across the vias (11C/16.8W = 65C/W).

That's pretty good and about as good as it could be. Hard to get any better than that.

Once I measured that I ran the board and it took 15-20 seconds before the IC shut down. Not bad for just a little piece of copper on the backside and with 10V in it was drawing around 2A from the supply.

I took a flat edge pliers and clamped it to one of the U sides and re-ran the test. This time it did not shut down.

To keep the IC as happy as possible I would fill the vias and add the copper heat sink to the bottom to help spread the heat and provide a reliable path to your primary heat sink (flashlight body). I would also glue on the top of the IC as much mass and surface area possible to remove additional heat from the top side. The two combined methods will ensure that the IC will not shut down.

I would not suggest moving the LED drive current higher than 1A for most configurations or limit the power to 15W or so.

Wayne

[buschpounder]

Wayne, I recently purchased a shark. I was reading on the sandwich shoppe site that j1 is left blank. Does that mean that it no longer needs jumpered? Also, How do you tell if you have a bad shark when you test itwith the 3 volts? The volts on my shark seemed to jump around and the shark got hot and seemed to shut down. It did not produce a steady volt. Thanks for your time.

[hank]

Ya know, if you have a good longterm relationship with your dentist, sometimes you can get them to X-ray something -- dunno who else might offer that, but it's one way of finding out where the voids are.

Now if you had a way to work in a bell jar after pumping the air out, so the returning air pressure would force the epoxy into any voids ...

[Tidra]

Hi,

I would like to use Shark to drive 4 x SSC P4 @ 800 – 1000mA. My SSC LEDs are rated 3-3.25V and because of safety reasons I would like to wire two by two in series and then in parallel, that mean that I would need 1,6 – 2A at the output. Of course I would use some resistors to provide a different stage of brightness.

What kind of input voltage would be most proper for that?

Thank you
Iztok

[LED Zeppelin]

Wayne, I recently purchased a shark. I was reading on the sandwich shoppe site that j1 is left blank. Does that mean that it no longer needs jumpered? Also, How do you tell if you have a bad shark when you test itwith the 3 volts? The volts on my shark seemed to jump around and the shark got hot and seemed to shut down. It did not produce a steady volt. Thanks for your time.

The newest batch of Sharks are now default @26V, J1 is not there nor is it needed at all.

There is also a new ordering option to remove the on-board pot, great for variable brightness mods.

Hi,

I would like to use Shark to drive 4 x SSC P4 @ 800 – 1000mA. My SSC LEDs are rated 3-3.25V and because of safety reasons I would like to wire two by two in series and then in parallel, that mean that I would need 1,6 – 2A at the output. Of course I would use some resistors to provide a different stage of brightness.

What kind of input voltage would be most proper for that?

Thank you
Iztok

Tidra, I would wire the LEDs all in series giving you a string with a Vf of 12-13V. Then the Vin would be best from 6-10V, the higher being easier on the Sharks.

Wiring in series will equalize the current through all the LEDs.

[rolling]

Hi. I have a 22k pot laying auround here, what effect would it have when i use the 22k instead of a 20k?

And would that be sufficient heatsinking for a setup of 4 x Q5 and 9x Eneloop? It is a Aleph E-Screw, all glued together with AA epoxy.

[dat2zip]

22k is fine.

Your heat sinking looks great.

Wayne

[rolling]

Thank you. I assembled it today and at first it worked fine. But after i turned the pod couple of times i heard a sound like a spark. Now it flickers when i turn the pod. What would be a good choise for a pot? Or did I damage my shark?

regards
rollin

p.s.: i just looked and j1 isnt jumpered. do i have to? *edit* I now have jumpered j1 and it behaves the same

[dat2zip]

Thank you. I assembled it today and at first it worked fine. But after i turned the pod couple of times i heard a sound like a spark. Now it flickers when i turn the pod. What would be a good choise for a pot? Or did I damage my shark?

regards
rollin

p.s.: i just looked and j1 isnt jumpered. do i have to? *edit* I now have jumpered j1 and it behaves the same

The latest batch is configured different and J1 doesn't do anything on the new batch.

As for the Shark going poof. If it overheated it could have killed itself. The Shark needs to be heatsinked all the time even during initial testing.

I don't think the pot damaged the Shark. I would suspect the shark got too hot.

Wayne

[dat2zip]

I'm starting to get feedback that the 26V configuration may not be happy with no load. Please be careful and minimize the amount of time power is applied with no load.

I've got the copper heat sink parts back and I think I will kit the two copper parts along with a custom finned heat sink that can be glued onto the IC.

I'll try to take a picture of the kit and post it here so you can see what will be included in the kit.

Filling the holes under the IC with solder can be tricky if not done correctly.

Wayne

[xiorcal]

I would like to run 2 parallel strings of 4-series Edison-Opto KLC8 (Cree chip) LEDs from the Shark. I've done the below calculations, can you please correct me on anything I have wrong, and answer the extra questions at the bottom?

J1 Jumpered for LEDs
External 20K pot
Shark will provide 980mA equally divided by the 2 strings:
490mA per string max
50mA per string min

LED Characteristics:
2.875V at 50mA/string
3.5V at 500mA/string

String Characteristics:
MinV10 (To be sure not fry shark from DD, used 2.5V-per-LED figure)
MaxV14.8

Shark powering characteristics:
Max Vin 10V (More comfortably 9V)
Min Vin 5V (Nominally >7.5V)

At 5Vin boosting to 14.8Vout ≈ 80%eff ≈ max3.7Iin (Too much?)
At 6Vin would be ≈ max3.08Iin (Acceptable?)

At 9Vin ≈ max2.05Iin - min1.38Iin (This is the preferable V/Iin, right?)


Application questions;

Q1.If running from a regulated power supply, is it preferable to have the shark performing the least voltage boost possible, or does it really not matter if the input is 1/3:1V, 1/2:1V or .9:1V as long as the input is constant?
(Q2.Is any Vin configuration better from the perspective of heat generation?)

At max throughput, the Shark would be running .98A * 3.7V*4 = 14.5W ≈ 2.9W heat.
From previous posts I gather that requires heat-sinking.
Q5.How much heat-sinking would be required for the ability to constant-run without damaging the converter? The application is preferably capable of up to 12 hours. Space is not of much concern, however a passive sink is necessary. Would sinking to the inside of a die-cast Aluminium enclosure suffice?

Thanks
-Xiorcal

[DonShock]

Thank you. I assembled it today and at first it worked fine. ........ Now it flickers when i turn the pod. ....... Or did I damage my shark? ......

I had a similar thing happen with one of my Sharks. I had built a setup and it had been running fine with an external pot. Then I decided to add a quick disconnect between the body mounted pot and the heatsink mounted Shark. With the light off and the batteries removed, I cut the wires and soldered in the disconnect. I double checked after soldering to ensure no wires were accidentally crossed. After the batteries were reinstalled, the LEDs lit up nice and bright for less than a second and then went out. I don't think it's a heat issue; there wasn't enough time for any heat buildup, especially since the Shark was attached to Download's multisink at the time. After that, there would be a very short duration, very dim ficker of the LEDs when the light was first switched on and then nothing. Troubleshooting showed that the Shark was no longer boosting voltage. Vout = Vin. Further inspection, both visual with a 20x stereo microscope and what electrical checks were possible without detailed knowledge of the circuit, was unable to find any cause. I finally gave up and replaced it entirely.

[dat2zip]

I would like to run 2 parallel strings of 4-series Edison-Opto KLC8 (Cree chip) LEDs from the Shark. I've done the below calculations, can you please correct me on anything I have wrong, and answer the extra questions at the bottom?

J1 Jumpered for LEDs
External 20K pot
Shark will provide 980mA equally divided by the 2 strings:
490mA per string max
50mA per string min

LED Characteristics:
2.875V at 50mA/string
3.5V at 500mA/string

String Characteristics:
MinV10 (To be sure not fry shark from DD, used 2.5V-per-LED figure)
MaxV14.8

Shark powering characteristics:
Max Vin 10V (More comfortably 9V)
Min Vin 5V (Nominally >7.5V)

At 5Vin boosting to 14.8Vout ≈ 80%eff ≈ max3.7Iin (Too much?)
At 6Vin would be ≈ max3.08Iin (Acceptable?)

At 9Vin ≈ max2.05Iin - min1.38Iin (This is the preferable V/Iin, right?)


Application questions;

Q1.If running from a regulated power supply, is it preferable to have the shark performing the least voltage boost possible, or does it really not matter if the input is 1/3:1V, 1/2:1V or .9:1V as long as the input is constant?
(Q2.Is any Vin configuration better from the perspective of heat generation?)

At max throughput, the Shark would be running .98A * 3.7V*4 = 14.5W ≈ 2.9W heat.
From previous posts I gather that requires heat-sinking.
Q5.How much heat-sinking would be required for the ability to constant-run without damaging the converter? The application is preferably capable of up to 12 hours. Space is not of much concern, however a passive sink is necessary. Would sinking to the inside of a die-cast Aluminium enclosure suffice?

Thanks
-Xiorcal

Q1.If running from a regulated power supply, is it preferable to have the shark performing the least voltage boost possible, or does it really not matter if the input is 1/3:1V, 1/2:1V or .9:1V as long as the input is constant?
Yes, it matters. The less the Shark has to boost the easier it is on the Shark as far as heat generation from the Shark is concerned. The more current the Shark has to pull from the battery the more heat is generates.
The highest you can go is the the lowest output setting since that is when Vout is minimum and you need to be less than that to always be in boost mode.



(Q2.Is any Vin configuration better from the perspective of heat generation?)
Maximize Vin. In your case running Vin = LED Vmin = ~10V would the ideal Vin to minimize heat generation from the Shark. The Shark always need to be bonded to another heat sink of some shape, box, housing.

While the data sheet for the IC used on the Shark says it is good for 3A the heat sinking needed requires both top of IC and bottom of IC heat sinking in order to keep the IC happy. Realistically, the max battery current is around 2A tops.

I'll be offering copper heat sink for the Shark very shortly. The copper parts arrived earlier this week and I have done some testing.
Here is a picture of the two copper parts. One is soldered to the bottom of the board to the two grids of vias. The grid of vias under the IC must be completely filled with solder or the bottom side heat transfer is not efficient at all even with the copper "C" soldered on. The copper "C" provides a thermal path from the board to another heat sink. The copper "C" by itself does nothing.

The smaller piece is soldered next to the diode and is thermally greased to the diode to suck the heat off the diode. Not necessary in most cases and it causes additional potential problems shorting to the underside of the Remora board. They were not as expensive to make and since I had a shop make the one it was only a small additional charge to make these.



To get the maximum output from the Shark the top side needs to be heat sinked as well. I'll be offering a limited finned heat sink that can be glued to the top of the IC which in conjuction with the bottom side thermally attached to a heat sink will provide maximum operation output.


Mock up of the heat sink on top of the IC. In the picture the heat sink was placed there (not glued down).



Since the finned heat sink has to be cut from a larger heat sink and since it is such a small part it is a real PIA to cut even with good tools.

Wayne