[Milkymist-devel] DDC short circuit protection
werner at almesberger.net
Wed Feb 22 16:18:19 PST 2012
Adam Wang wrote:
> d) PTC, holds 50mA at 20?C, 1.5s time to trip when 150mA at 20?C, like
> Tyco p/n: MICROSMD005F-2, Rmin = 3.6 Ohm, R1max = 50 Ohm, resettable way,
> but recovery time is long, needs to carefully check few parameters to
> ensure 0.305 usd/100pcs.
Yes, the R1max here worries me a bit. I haven't been able to
find any information on the conditions under which R1max is
reached (besides the fuse having tripped at least once).
In our scenario, we could only afford a maximum resistance of
10 Ohm or a device drawing the maximum allowed 50 mA would
drop more than 0.5 V (10%) of those 5 V. (Assuming here that
a drop of 10% would still be acceptable, but I don't have any
reference to back this up.)
> Regards to that why selected 50mA, check  to meet two cases:
> * If monitor is powered off, it may not draw more than 50mA.
> * If monitor is powered on, it may not draw more than 10mA.
In fact, we should deliver up to 55 mA, according to page 15 of
"Note: The DDC required +5 volt power pin must be able to supply
a minimum of 55 mA."
Not sure where the 5 mA difference is supposed to go, though.
Also not sure it we really care about distinguishing between
50 mA and 55 mA. The MIC2090's Ilimit is from 50-100 mA anyway,
so it will almost certainly let a little bit more than 50 mA
Let's just hope monitors don't expect they can exceed their 50
mA by a lot. Well, the chip has a fault indicator. So at least
M1r4 could tell us very quickly if it ever meets such device.
> so indirectly shows in VGA system, draw current is
> low even under 10 mA too.
So we have one sample (out of a possible hundreds is not
thousands) that would not need all that much effort :)
Unless we can be sure monitors/TVs/projectors/etc. that draw
more than just a few mA are extremely rare, it wouldn't be safe
to design for less than at least 50 mA.
I'm also not sure what would be the consequences of not
providing a correct DCC5V supply. E.g., would the monitor
refuse to power on ? Would is malfunction in some other way ?
Or would this only affect the host's ability to read the
monitor's characteristics, which should be a comparably minor
issue. (As far as I know, we currently don't use any of this
> So Werner and me thought that candidate with Micrel part can work for it.
> so will order few and test soon.
Yes, I think this is the safest solution. Spending a few more
cents on getting this right certainly beats having to worry
about DCC5V as the potential cause each time a monitor problem
More information about the Devel