Blue Shark Advisory.

[dat2zip]

I have seen a higher than normal number of emails with customers not achieving full output under high power factor conditions.

I have found that the current batch of Blue Sharks were loaded with a inductor of wrong value. The inductor used is a 6.8uH inductor and it should be a 4.6uH inductor.

If you plan on driving with an input to output ratio higher than 2:1 and expect input currents to rise above 2A you should return any unused converter boards and get an exchange for new ones.

There is nothing wrong with using this inductor except for the fact that the 6.8uH inductor generates more heat and this additional heat on the board will heat up the IC and cause it to thermally shut down. Continuous thermally shutting down will eventually cause the IC to fail.

Effective today June 11, 2010 any Blue Shark shipped will have the correct inductor installed.

If you have any questions you can post here or send us an email.

To date an estimated number of Blue Sharks shipped is somewhere in the ballpark of 30 or so plus some quantities shipped to other customers with larger volume.

Wayne

 

[Justin Case]

What should one do with installed Blue Sharks that are thermally shutting down, presumably because they have a 6.8uH inductor? I have such a situation, and the Blue Shark is installed with some top side copper pieces to help manage the heat. But I still get shutdown. I can't see the inductor markings because of the top side copper thermally glued to the inductor to pull away the heat.

Voltage ratio is prob >2:1 (3x17670 driving 7 Cree XR-Es at 950mA) and current draw is over 2A.

Any issues to be aware of if I want to de-solder the 6.8uH inductor and replace it with the proper one? Is it a Vishay-Dale IHLP-2525-CZ-01 (I think I have the part # right from memory). It looks like two pins on the switcher IC run to the inductor trace. Any risk of overheating those pins?

 

[Alan B]

My Blue Sharks are not yet tested or installed. They were shipped June 20, 2009. Do they have any issues?

Thanks,

 

[fyrstormer]

Is fully-potting the board in thermal epoxy sufficient to prevent thermal shutdown when running at full specified load?

 

[dat2zip]

More information and update 10JUN2010.

I have been doing more bench testing and the inductor is not the problem. The current batch doesn't seem to get past ~2.5A on the input side.

This is very different than previous blue sharks and I do not have any answers at the moment. Right now all Blue Sharks are on hold and we are not shipping these at the moment till I can resolve this.

Sorry about the inconvenience.

Wayne

 

[Justin Case]

Wayne, I am getting petrev's 6S x 17500 battery holder for the SureFire M6 to drive my LED Zeppelin MZXR-7. That should give me a much more favorable Vbatt to Vload ratio (very close to 1:1). Hopefully, that solves my thermal issues independent of what you eventually discover to be the problem.

I've previously measured Vload at max drive current to be about 23.9V. Five 17500 cells should be less than Vload at all times, even when freshly charged. However, six 17500 cells can start out at 6*4.20V = 25.2V for a few seconds, and then sag down to a quasi-steady state Vbatt of 6*3.7V = 22.2V.

Do you think that having Vbatt > Vload for a brief time running six 17500s is an issue, and that I should avoid it by using one dummy cell and running five 17500s?

 

[Justin Case]

On one Blue Shark that shut down within a few minutes on me, I had used Arctic Alumina to glue a small copper heat sink/tab to the top of the diode. The copper piece had one fin going straight up and a small tab that was thermal glued to the thermal vias next to the diode. The copper piece looked like a chair with a tall seat back (the fin) and a small foot rest (the tab).

On inspection after thermal shutdown, the copper fin was seriously discolored, clearly from surface oxidation at high temp.

 

[fyrstormer]

Yikes. That definitely sounds like it was overloaded, possibly by a partial short-circuit.

 

[Justin Case]

Yikes. That definitely sounds like it was overloaded, possibly by a partial short-circuit.

I'm unclear where this partial short would come from. Maybe I'm doing something wrong, but when I run the Vishay inductor loss calculator, the inductor current exceeds the saturation value when Vbatt falls to around 11.6V (Vload is set to 22.0V and drive current is 0.8A).

 

[fyrstormer]

Oh, I wasn't saying you did anything wrong; I was just surmising that if the inductor isn't the problem as Wayne says, and you're not overloading the driver, then there must be something else causing more drain than anticipated. A wiring or soldering fault seems like the simplest hypothesis.

 

[dat2zip]

Wayne, I am getting petrev's 6S x 17500 battery holder for the SureFire M6 to drive my LED Zeppelin MZXR-7. That should give me a much more favorable Vbatt to Vload ratio (very close to 1:1). Hopefully, that solves my thermal issues independent of what you eventually discover to be the problem.

I've previously measured Vload at max drive current to be about 23.9V. Five 17500 cells should be less than Vload at all times, even when freshly charged. However, six 17500 cells can start out at 6*4.20V = 25.2V for a few seconds, and then sag down to a quasi-steady state Vbatt of 6*3.7V = 22.2V.

Do you think that having Vbatt > Vload for a brief time running six 17500s is an issue, and that I should avoid it by using one dummy cell and running five 17500s?

Having Vbattery higher than Vload is not a terrible condition if under that condition the DD to the LEDs are low. I have builds that work this way and do not get low low with a fresh battery. Only after some initial use does low go down to the proper level. Under that condition the current in DD is not something to be worried about since you might be driving the LEDs at 100mA or so.

Wayne

 

[dat2zip]

On one Blue Shark that shut down within a few minutes on me, I had used Arctic Alumina to glue a small copper heat sink/tab to the top of the diode. The copper piece had one fin going straight up and a small tab that was thermal glued to the thermal vias next to the diode. The copper piece looked like a chair with a tall seat back (the fin) and a small foot rest (the tab).

On inspection after thermal shutdown, the copper fin was seriously discolored, clearly from surface oxidation at high temp.

The diode heat dissipation is dependant on the Vin to Vout ratio.

What was that configuration for battery voltage and LED Vf load?

How long ago was this? We changed the diode to a more efficient diode and I don't think we have had heat issues with the new diode.

Wayne

 

[dat2zip]

If you input to output ratio is higher than 2:1 and you plan on driving maximum output of 1A to the load you may experience thermal shutdown with low battery condtions as the vin to vout ratio increases.

With the current Blue Shark issues I am recommending you turn the output down 10-20% to avoid this issue. Turning the output down will put less demand on the converter board and will be less prone to thermal overload.

In the meantime I have ordered a new small batch of boards that will be called the Red Shark for the red soldermask color that will be used.

The Red Shark will have components moved around a bit. Basically the same design. Jumper J1 will be eliminated. The interface to the remora will be the full 3 pin connector and mate with the three larger holes in the Remora. The physical mount alignment will remain the same as the holes are now in alignment with the offset of the three and two holes on the Remora. The larger holes allow for the small connector header to be used or regular wire.

The input power holes will be moved towards the edge of the board and farther from the inductor to eliminate the inductor sitting over the Vin hole.

Output capacitor size will now be a 1206 and allow for a 35-50V capacitor rating since any configuration currently over 25V on the output is over the current 0805 voltage rating.

The Red Shark will be 2 oz copper and a special technology to improve the transfer from U1 to the copper C. This hopefully will solve once and for all times the thermal issues with this design.


I will post updates on this thread regarding any status or information regarding testing results of the Red Shark.

Until then if you can turn down the Blue Shark to put it in the safe operating zone would be the recommendation until the Red Sharks become available.

Wayne

 

[dat2zip]

I am thinking of re-doing the Remora as well. This will correct a few minor issues and make full PWM control available.

If the Remora is re-done I am thinking of making some additional code enhancements and make one input for thermal sensing using a thermister. The thermister could be mounted to the IC or to the heatsink to monitor the heat from the LED whichever is worse. This will alllow thermal foldback.

Also, some mechanism for setting the low will be implemented.

Low battery detection will be available, but, not clear how to provide for different battery setups. I don't like setting it via the UI. I would prefer cutting jumper traces on the board. If three jumpers are provided that would allow for 8 configurations.

This is just in the thought process. If you have suggestions or ideas I would like to hear about them.

Wayne

 

[Justin Case]

Having Vbattery higher than Vload is not a terrible condition if under that condition the DD to the LEDs are low. I have builds that work this way and do not get low low with a fresh battery. Only after some initial use does low go down to the proper level. Under that condition the current in DD is not something to be worried about since you might be driving the LEDs at 100mA or so.

Wayne

Do you mean that I might drive my LEDs at some level just slightly greater than 1000mA for a short time, or do you really mean 100mA? I would think the former since the nominal Vbatt=25.2V is only about 1.3V above Vload (or about 0.2V per LED).

 

[Justin Case]

The diode heat dissipation is dependant on the Vin to Vout ratio.

What was that configuration for battery voltage and LED Vf load?

How long ago was this? We changed the diode to a more efficient diode and I don't think we have had heat issues with the new diode.

Wayne

Vbatt:Vload ratio is not the most favorable -- 3xLi-ion (let's say 11.1V) vs about 23V for total Vf. I turned down the trim pot to about 800mA drive current to get to that total Vf (about 3.3V per LED).

Bought the Blue Shark maybe a couple of months ago?

Diode marking is "S4" if that helps.

 

[dat2zip]

Do you mean that I might drive my LEDs at some level just slightly greater than 1000mA for a short time, or do you really mean 100mA? I would think the former since the nominal Vbatt=25.2V is only about 1.3V above Vload (or about 0.2V per LED).

Sorry, Didn't read it very carefully.

In my case Vin is around 18V and Vload at full drive is around 23V. But, under light loads Vf drops to around 16V or so. This puts the converter in DD mode at light load configuration.

Your configuration is different. Your DD configuration will be pushing the LEDs pretty hard. It depends on how hard your DD configuration is and for how long. How long is it 25.2V before it sags. If it is a couple of seconds that shouldn't be a problem. If it is more than a few minutes that could be a problem. It all depends on how hard DD is under this condition. If you are pushing more than 4A for more than a few seconds you might make the diode overheat.

S4 marking is the current diode. I'll have to take another look at this and see if there is still a heat issue with the diode.

Wayne

 

[Justin Case]

I'll have to discharge test an AW17500, which is my planned battery source, to see what it does at roughly 1A draw, which is what I'd ballpark estimate the tail current would be when using six 17500 cells in series to drive six XR-Es at full power (980mA nominal drive current).

But based on Silverfox's testing here, it looks like 17500 cells sag to 4.0V quite fast (within seconds), and down to 3.9V within about 3 min.

4.0V per cell would be just barely greater than my total Vf of about 23.9V at full power drive current. 3.9V per cell would be less than Vload.

 

[dat2zip]

I'll have to discharge test an AW17500, which is my planned battery source, to see what it does at roughly 1A draw, which is what I'd ballpark estimate the tail current would be when using six 17500 cells in series to drive six XR-Es at full power (980mA nominal drive current).

But based on Silverfox's testing here, it looks like 17500 cells sag to 4.0V quite fast (within seconds), and down to 3.9V within about 3 min.

4.0V per cell would be just barely greater than my total Vf of about 23.9V at full power drive current. 3.9V per cell would be less than Vload.

That sounds safe. Just don't do this with hot LEDs and batteries directly off the chargers. Hot LED Vf will be lower than room temperature and will draw more current under this condition.

Wayne

 

[Justin Case]

So I decided to remain conservative and kept the Blue Shark drive current dialed down to about 810mA. The total Vf at that drive current is about 23.1V. I loaded my SureFire M6 with 6SxAW17500 and measured a tail current draw of about 0.9A.

Thus, I charged a test AW17500 cell to 4.10V (24.40V total nominal Vbatt) with my hobby charger and then discharge tested that cell at 0.90A. The voltage dropped to 4.0V within 3 sec, and it took that long probably because the hobby charger takes a finite time to ramp up the discharge rate to the 0.90A setting. Within one minute, the voltage sagged to 3.90V, or about 23.40V total, which is close to the measured total Vf.

When I charged my test AW17500 cell to 4.20V (25.20V total nominal Vbatt), it took about 5 sec for the cell to sag to 4.0V, again probably due to the finite time it takes the hobby charger to ramp to the 0.90A discharge rate. After 3.5 min, the cell voltage dropped to 3.90V. So it looks like my voltage sag estimates based on Silverfox's data were quite close.

Your caveat about avoiding the combination of hot LEDs (due to lower Vf) and Li-ions fresh off the charger suggests that going with a 4.10V terminating voltage charge might be the safer approach.

BTW, using the significantly more favorable Vbatt:Vload ratio has clearly reduced the waste heat generation significantly. Previously, with 3SxAW17670, the light would suffer thermal shutdown after about 13 min of continuous running. Using the 6Sx17500 batt config, I ran the light for 23 min continuous before I terminated the test from boredom. The light was hot but it was chugging along.