A few small schematic improvements
Posted: Fri Feb 15, 2019 1:29 am
I've been looking at the schematic (rev D3_E) and there are a few small improvements that I've noticed might be made. Obviously if it works, it works, these are just things I noticed:
The power supply section:
The power supply section:
- Holy war material: C3, C4, C5, C6 are specified as tantalum, but MLCC capacitors would work just as well if not better -- these are linear regulators, not switching regulators -- and they won't fail short
- PWR_JACK says 7-9V, and the manual (rev 1.2) says not to connect more than 9VDC, but it seems that D2 would block any higher voltage, and the LM1117 can take up to 15V.
- Z1 - not sure why this is needed, considering the above.
- The USB 2.0 spec doesn't want more than 10u capacitance on VBUS. Higher than that and you're going to have higher inrush current for more time, which could cause a well-behaved USB host to not connect (again, if it works, it works, but still...). You have 2.2u after the fuse, then a diode, then 10u on the LM1117 input. And if the 5V_VR jumper is selected you have an additional 47u. You could instead use a TPS2112, for example. This serves also as a switch between two supplies, so you could also eliminate D5 and reduce C10 to 100n. It also has a current limiter, so you could also eliminate F1.
- C12 could probably stand being bumped up to 4.7u. This is the example value given in the AT91SAM7S datasheet section 5.3.
- It looks like R6 and R7 are there to provide the minimum load current required by the LM1117-ADJ. However, the fixed-reference versions don't need such a minimum. You could eliminate R6, R7, R5, R10, R37, R4, and R9 by using the LM1117-5.0 and LM1117-3.3.
- It's likely that D1 and D4 aren't necessary. The LM1117 datasheet says that these are needed when the output capacitance is extremely high (>1000uF).
- R19 doesn't need to be there. The AT91SAM7S datasheet (section 6.2) says that there's already an integrated 15k pulldown on the TST pin.
- Likewise with R17: section 6.3 says there's already a pullup.
- ADVREF and AD4-AD7 should probably be connected to ground -- I'm assuming the ADC isn't being used.
- R26 could be increased to 27k and R32 to 47k. This would match what's in the AT91SAM7S manual (section 35.4) and also reduce the load drawn from VBUS -- not by much, but still
- C19: Holy wars have broken out over this. My take is that if you do separate system ground from shield ground, then the shield ground should connect to a surrounding metal enclosure. There's no such thing here, so you may as well connect shield to system ground.