High Tide Aquatics

Trying to resurrect a Kessil #2

thesassyindian

Supporting Member
Back in September, I picked up a blown Kessil A160 from @Corallus , which was deemed unrepairable due to a blown switching regulator that had burned through multiple layers of the circuit board.
Details here.

Today, I picked up 8x A360W's from @Wlachnit to try and figure out what was going wrong.

IMG_4551.jpg
 
Last edited:
Welp! Looks like the A360's also have a distinct failure mode. How has Kessil not addressed this issue?

In all the lights I've opened, inductors L401, L400 and the ceramic capacitor C403 are blown out.

IMG_4554.jpg


In the section right next to it, which is an identical driver circuit, the corresponding passives, L300, L301 and C303 are intact!

After a gentle persuasion from my soldering tweezers, Voila!

1611115682616.png


Well, kind of :(
Here's the catch. Even though the light now turns on, the LEDs that are powered by that module don't turn on.

Original LED cluster:
IMG_4555.jpg



Notice that rows 5 and 8 are OFF.
1611115604720.png


Now, I just have to figure out the part numbers of those passives and why just those two are blowing out.
Usually, its a case of loop stability in switching regulators.
To object to my own statement above, if that was the case, all 4 driver sections should have exhibited the same behavior.

Time to investigate further!
Stay tuned!
 
Last edited:
Hmm, that burn pattern looks a lot like an improperly sized trace.

I am guessing all three parts are wired together there.
Perhaps it was laid out as a standard signal trace, not a wide power trace, or not wide enough?
Or maybe a via in the wrong spot?
You would have to look carefully at a good one to check.
 
Well if C303 is the same as C403 you should be able to just hook that up to a multimeter to find it's value. Measuring inductors isn't quite so straight forward, with an oscilloscope you absolutely can measure them indirectly, but it's a bit more work. Could also try and ask Kessil what the parts are, I'm almost sure they'll tell you nothing, but if you show them the picture, and your desire to fix it they may be a bit more forthcoming with the info, especially if you say you can measure it from the other ones just trying to save some time and not remove them.
 
@sfsuphysics we have industrial LCR meters at work so that's no big deal at all :)
I'm sure I can get compatible components from Digi-Key in my weekly order.

@rygh It doesn't seem to be a power trace. The arrangement of the inductors and capacitor looks like a filter circuit for the analog dimmer line.
I'm taking all the conformal coating out today and will reverse engineer the circuit.

Oddly enough, the LED driver circuit does not match what's suggested in the datasheet for the chip, so its going to be a lot of probing under the microscope :)
 
@sfsuphysics we have industrial LCR meters at work so that's no big deal at all :)
I'm sure I can get compatible components from Digi-Key in my weekly order.

@rygh It doesn't seem to be a power trace. The arrangement of the inductors and capacitor looks like a filter circuit for the analog dimmer line.
I'm taking all the conformal coating out today and will reverse engineer the circuit.

Oddly enough, the LED driver circuit does not match what's suggested in the datasheet for the chip, so its going to be a lot of probing under the microscope :)
Interesting.
Perhaps some sort of EMI reduction hack?
Beyond normal buck/boost power circuit, extensive use of inductors seems a bit unusual.

Curious: Which driver chip do they use?
 
Interesting.
Perhaps some sort of EMI reduction hack?
Beyond normal buck/boost power circuit, extensive use of inductors seems a bit unusual.

Curious: Which driver chip do they use?
They use 4x RT8462 as the primary drivers. What puzzles me is that their PCB layout doesn't match any of the suggested driver configurations :/
Also, it seems like the drivers blowing are caused by the LED dies themselves burning out.

I am however, able to disconnect the problematic driver and get the others back up and running!

That said, @Wlachnit , I have 3 lights working with the problematic LED driver channel disconnected!
 
They use 4x RT8462 as the primary drivers. What puzzles me is that their PCB layout doesn't match any of the suggested driver configurations :/
Also, it seems like the drivers blowing are caused by the LED dies themselves burning out.

I am however, able to disconnect the problematic driver and get the others back up and running!

That said, @Wlachnit , I have 3 lights working with the problematic LED driver channel disconnected!

As far as drivers blowing, it might make sense.
If you suddenly disconnect an inductor circuit (like when an LED blows open) you can get a large voltage on the output of the inductor.
I do not see flyback protection diodes on the example schematics, so I bet it is fair to guess there are none in the design.
And why would you bother ... it is dead at that point anyway.

Looking at the sample circuits, I still do not see where they would throw in extra inductors like that.
Just the main single inductor on the power out.
 
As far as drivers blowing, it might make sense.
If you suddenly disconnect an inductor circuit (like when an LED blows open) you can get a large voltage on the output of the inductor.
I do not see flyback protection diodes on the example schematics, so I bet it is fair to guess there are none in the design.
And why would you bother ... it is dead at that point anyway.

Looking at the sample circuits, I still do not see where they would throw in extra inductors like that.
Just the main single inductor on the power out.
Well, I think they count the MOSFET body diode itself as the flyback diode, which is a terrible design choice.

The four inductors and the capacitors seem a lot like an LC ladder filter. Why they would need that, I have no idea.
Still need to investigate further.

One of the 8 lights is blown way beyond repair, and my intention is to completely strip off the conformal coating with a solvent and retrace the driver circuit.

If it is indeed as simple as adding an airwire or deadbugging a $0.10 diode, that would be awesome!
 
Well, I think they count the MOSFET body diode itself as the flyback diode, which is a terrible design choice.

The four inductors and the capacitors seem a lot like an LC ladder filter. Why they would need that, I have no idea.
Still need to investigate further.

One of the 8 lights is blown way beyond repair, and my intention is to completely strip off the conformal coating with a solvent and retrace the driver circuit.

If it is indeed as simple as adding an airwire or deadbugging a $0.10 diode, that would be awesome!

I guess a simple test is to remove the parts, clean up the mess, simply wire across where the diodes were, and ignore the missing cap.
If that LC is something like reducing output ripple, it might not really be needed.
And that test should be "relatively" unlikely to fry anything else.
 
I guess a simple test is to remove the parts, clean up the mess, simply wire across where the diodes were, and ignore the missing cap.
If that LC is something like reducing output ripple, it might not really be needed.
And that test should be "relatively" unlikely to fry anything else.
In an ideal case, yes haha but I don’t want to try something in haste and make it worse.

So I think I’ll wait till I’m able to create a proper schematic :)
 
So capacitors blew so nothing would turn on cause it was shorting? So removing the blown capacitors made the channels without blown capacitors work.

How does removing the capacitors make it work? Are their multiple lights all on the same circuit so the additional power that should be going to the blown capacitors just goes to a different channel in the same circuit?

This is fascinating and going way over my head. Thanks for sharing all this info from your tinkering!
 
So capacitors blew so nothing would turn on cause it was shorting? So removing the blown capacitors made the channels without blown capacitors work.

How does removing the capacitors make it work? Are their multiple lights all on the same circuit so the additional power that should be going to the blown capacitors just goes to a different channel in the same circuit?

This is fascinating and going way over my head. Thanks for sharing all this info from your tinkering!
Well... kind of.

Consider a super special BAR driver circuit like below.
1611202232781.png

If the bypass capacitor malfunctions and shorts internally, yes you could desolder it and the rest of the circuit could work. There are many caveats, but in a very loose sense, this could work.

In the Kessil A360WE specifically, there are 4 independent driver circuits like below:
1611203042360.png

Since all four drivers are independent, if one channel short circuits, you can disconnect the input to just that module, and get the other 3 back up and running.

As for your statement::
so the additional power that should be going to the blown capacitors just goes to a different channel in the same circuit?
This is not the case, because fundamentally, you can only supply voltage. Current/Power is typically not supplied. It is drawn.
("Typically", because - constant current sources. That's a can of worms I'm trying my best not to open, but I'm happy to.)

Does that make sense?

My apologies if this explanation is too elementary.
 
Well damn, good for you then. We're not lucky enough to have that in the lab at school... of course we really don't need it as part of the lab is analyzing waveforms and trying to figure out what the inductance is :D
Hmm, that sounds like a fun lab.
My kid was taking physics this year, but poor teacher + covid = totally useless, plus tons of stress due to "interesting" grading.
So I am home-schooling him for the AP test as best I can. Circuits I can do. Not looking forward to quantum.
At any rate, I believe hands-on labs really make a difference, plus they are way more fun than studying,
and easier for someone who has not studied physics in 30 years, so emphasizing that.
 
Back
Top