Prev: TeeVee over Fibre
Next: Low noise guitar pre-amp / ADC
From: D Yuniskis on 14 Apr 2010 13:38
whit3rd wrote:
> On Apr 12, 3:39 pm, D Yuniskis <not.going.to...(a)seen.com> wrote:
>
>> I've got an external SATA-USB cradle that gave up
>> the ghost.
>>
>> 12VDC supply (used by the spindle motor) which is
>> then dropped down to 5V (for the logic) via a
>> buck converter ...
>
>> Voltmeter shows recirculating diode is shorted.
>
> Yes, that's a common failure mode. Use an ultrafast
> recovery diode (they're cheap enough), unless the
> markings indicate this was a Schottky type (more
> expensive). SATA only includes modern-ish drives,
> never going to require more than 2A @5V, but that's enough
> to put some thermal stress on the rectifier; it may need to

Of course, the drive's power consumption will drive the
*average* current through the diode -- no idea what the
ripple will be.

> heat-sink to the printed wiring board.

<frown> That would be hard. There doesn't appear to be much
unused real estate to exploit. E.g., moving to a larger package
may prove to be a problem in itself. Maybe try to go to a
thru-hole part more heavily derated and let it "fly"?
From: D Yuniskis on 14 Apr 2010 13:57
Hi Hammy,

Hammy wrote:
> On Tue, 13 Apr 2010 06:18:26 -0700, D Yuniskis
> <not.going.to.be(a)seen.com> wrote:
>
>> It really looks like a cheap design. Nothing that protects
>> the device from a defective drive (e.g., the 12V supply is
>> just fed directly through from the wall wart; the 5V supply
>> doesn't *appear* to have any current limiting, etc.)
>
> The controller you mentioned (FP6101) does have current limiting. It

Well, it has "short circuit" protection but I don't see
any way to limit the current (e.g.,setting it to only
deliver 2A)

> seems to be a Chinese or Taiwanese controller which may be the

<grin>

> problem. If their web site is any indication of the reliability of
> their products.
>
> http://www.micro-bridge.com/FP6101.asp

I've not been able to find anything that wasn't written in
chinese. I'll see if I can trick google into a translation.

> If all they have is ceramic input caps that may be causing the
> problem. The chip might not be failing from an over voltage transient
> due to some integrated protection? But the diode might not be so
> robust from OV.

I think the chip is designed for < 20V. Diode is 40.
Of course, as you say, the chip might be able to tolerate
stresses as long as it hasn't turned the FET on...

> "Ceramic Input Capacitors Can Cause Over voltage Transients"
>
> This is I think what Joerg was referring to.
>
> http://cds.linear.com/docs/Application%20Note/an88f.pdf

Cool! Actually, from eyeballing the foils on this board,
there is no way to avoid the scenario outlined in the AN.
The power switch has nothing on the upstream side -- besides
the wall wart. I.e., turning power on after connecting the
wall wart is just the same as turning power on *before*
attaching the wall wart.

The only 'lytics I see are on the output of the converter.

So, tacking something across the input "barrel connector"
would be a win -- as well as finding a heftier diode.

> You may also want to scope for ringing on the diode under various line
> load conditions it may need a snubber.

Have to replace the diode before I can even power it up.

> Or its possible unstable loop, saturating inductor.

But, that would have toasted the load already, right?
(since the load expects to see 5V, not the 12V supplied).
First  |  Prev  | 
Pages: 1 2 3
Prev: TeeVee over Fibre
Next: Low noise guitar pre-amp / ADC