Hi Ken!

Nice list.

> Ensure all devices follow the Rocket Names naming guidelines. Ensure
> that all lines in the schematic have properly descriptive names. Run
> BOM and prepare a parts list (ensure all parts carried over from
> prior design are still available).

Yep! You should do this.

> 1) U281 Binary ripple counter used to divide the 12MHz clock down to
> 1.5MHz for sync of the SPS and HAP switching regulators. Issue:  At
> 5V for VDD the part can only handle a max input frequency of 3.5MHz.
> The specified part can handle up to 12MHz, but only with Vdd=15V. 
> Solution: Replace with a 3.3V, 12MHz (or higher) divider circuit.
> Perhaps similar to the SN54LV163A (binary counter, fmax=50MHz
> @VDD=3.3V, use Qb and clock) from TI.

*Great* catch. We don't need the 163's preset capability, so the 161 is
OK as well. However, they're all 16 pins, and the only one Digikey has
which is in a QFN is your 163, so yep, go nuts:


> 2) CAN transceiver: MCP251x Issue: Not a 3.3V part. Solution: Replace
> with 3.3V CAN transceiver.  These are hard to find, but it looks like
> TI makes a few, though none in a package smaller than 8pin SOIC. A
> good choice appears to be the SN65HVD234 from TI

Right, you won't find a CAN transceiver in anything *but* an 8-soic. And
that's because they're all descended from a phillips part, and they're
all pin compatible. Note that the MCP2551 (not the MCP251x, that's a
different chip entirely) does require 5V input but is compatible with
3.3V logic. So if we decided to rely on the fact that all nodes have 5V,
we could use it. But you're right, let's not, and use a 3.3V chip instead.

The only feature which I'd like to find in a CAN transceiver is the
stuck-dominant-bit-goes-bus-off feature - this is where if the
transmitting node goes nuts and transmits a dominant bit on the bus for
more than a CAN message time, the transceiver stops transmitting.

Else the '234 is fine.

> 3) U283 Reset controller, used to hold the reset signal until the SPS
> or HAP has reached stable output Issue: Is this part and its
> associated circuitry needed? Solution: Just like the LPC1248, the
> LPC2368 has brownout circuitry built into the IC. That said the
> justification given by the 2006 Capstone team still appears to be
> valid.

Meh. I've go back and forth. Keep it, but let's rethink it carefully
before we make a decision.

> 4) U282 USB transient noise suppressor Issue: The part is an SOT23
> package, not DFN. That said the SOT-6 is still a pretty small part,
> just not leadless. Solution: Possible to look up a new part available
> in a DFN package. The existing part is available in SOT23, BGA, SSOP,
> and PDIP.

Nah: SOT23's are small enough. Keep it, unless you're itching to go find
new parts :)

> 5) Small "cut-able" jumpers Issue: in several places, there are small
> cut-able jumpers in the schematic, is there a standard symbol for
> this? Is it a good idea to cut traces on the board in the first
> place? Solution: Use two zero ohm resistors, mark one as a do not
> place. If desired, resistor can be removed and soldered on in the
> alternate position, resulting in lower possibility of board damage
> and higher reliability during use.

Actually, cutting wires on a board is pretty reliable. In fact, I just
did it today more than 8 times while hunting down a short in one of my
PCBs. And the point of the jumpers is that we probably never actually do
want to cut it.. that's sort of the point. If we were seriously
considering changing it, then I agree with you, zero ohm resistors is a
good idea. The jumpers are a statement that says "we're probably not
going to do this", as opposed to the resistors which say "we might want
to do this".

> 6) USB High Speed Select Issue: Can the LPC2368 drive the voltage and
> current needed to get the bus to act in High Speed Mode? Solution:
> Probably, Need to look up the USB2.0 spec to be sure. The LPC2368 can
> output +2.9V (VDD-0.4V) at -4mA on its GPIO pins.

The LPC2368 is only able to do USB 2.0 "Full Speed", not "High Speed",
so this isn't an issue. We thought about this and decided that each
individual node does not have to be high speed, as long as the hub could
pipe the full speed streams to a high speed flight computer. We'll do
that "one day" since our current FC is only 1.1 compliant :(

> 7) Everything Else Issue: Time consuming Solution: Time machine, I
> checked, TI still doesn't make one, neither do Micrel, Linear,
> On-semi, IBM, or Intel.

Did you check Yoyodyne?


PS. I'm meeting Dave at 10:30am tomorrow in the CS lounge, so I'd
recommend coming late, say 11:30am tomorrow to talk schematics.


       Andrew Greenberg - and...@thetovacompany.com - 503.788.1343

Andrew Greenberg

Portland State Aerospace Society (http://psas.pdx.edu/)
and...@psas.pdx.edu  P: 503.788.1343  C: 503.708.7711

psas-avionics mailing list

Reply via email to