[Milkymist-devel] proposed M1 reset circuit improvement
werner at almesberger.net
Mon Jul 25 04:15:18 PDT 2011
The reset circuit caused some trouble in M1 rc3, due to the high
capacitance of the diodes and their reverse leakage current, which
makes the real-life performance of the diodes very different from
that of ideal diodes.
An alternative solution would be to use logic gates. They provide
a very clean separation between their input and their output.
The 74AUP family can operate at voltages as low as 0.8 V, so the
gates should already be fully operational before most of the
system has even quite realized that it has power.
The following schematics show three circuit variants:
Upper left and lower left differ in the use of chips in SC-70 vs.
uQFN packages. The one on the right uses a more analog approach,
allowing the use of the same chip for both functions. Sebastien
prefers the fully digital solution, i.e., the left-hand side.
As a reminder, this is the current circuit in M1 rc3 (after
Regarding sourcing, the 74AUP1G07 and 74AUP1G08 are available from
TI and NXP:
They come in a number of different packages. The following charts
compare the TI and NXP packages and group those with compatible
In order to avoid unpleasant surprises during sourcing, it's
desirable to use a package that is available from more than one
manufacturer. Furthermore, the chips in question should be stocked
by major distributors. (Lead time from the manufacturers would be
painfully long 2-3 months.) The following charts show the
availability as of last weekend:
We can see that the DCK/GW and the DSF/GF packages have good to
reasonable availability. Unless space becomes a concern, the SC-70
packages (DCK/GW) would be the preferred choice, due to better
availabilty and being easier to solder and rework.
The KiCad schematics, along with a BOOKSHELF file, are here:
More information about the Devel