Make sure you do what the data sheet says about the 1µF input and output capacitors for the 3.3V LDO. LDO's can be tricky to keep stable, ie to keep them from oscillating. The designers have done the math and subsequent testing to give you the ranges and types of capacitors. You might save yourself a lot of grief by following them. I had an LDO in an aircraft intercom design 20+ years ago that worked fine in most aircraft, but in a few situations the aircraft's power source was exactly right to make it oscillate and damage itself, and I had to quickly re-do the board with a fix so it worked right for everyone. They have worked fine ever since, with no failures.
Just for grins, make C8 a 100µF cap instead of 10µF. The physical size difference will be minimal and you'll get a little better regulation from the 7805 as loads on the 5 volt circuit fluctuate.
The data sheet also says you should have a capacitor on the 7805's input, really close to the regulator. For digital work where a jump of 50 or even 100mV won't matter, I don't think there's any point in going overboard on the output capacitor though since beyond a few µF the regulator's response will be quick enough to compensate for sudden changes in the load. That's not to say of course that you shouldn't have ,01 or 0.1µF bypass capacitors right at each digital IC, particularly if you don't have a ground plane and a power plane. (Copper pours don't qualify.)
If anyone cares to look it up, Electronic Design magazine had a good article on LDO regulator compensation for stable operation, including selecting the right output capacitor, in the Nov 4, 1996 issue, pages 99-104, written by Chester Simpson of National Semiconductor Corp.. It seems like I saw an even better one a few years ago, further clarifying the matter of poles and zeros of the function, but I can't remember where.
I will certainly put bypass caps on all IC's. Even the clock. Just because.
I think I might go ahead and put a cap on the 7805 input line. This isn't going to be a high-volume production so I'm not too concerned on saving a few cents here or there.
Let me rephrase that...should there be a large cap supplying the "IN" pin of the 7805?
All depends on from where the 7805 is getting its juice. If the basic source is sufficiently filtered, I'd say don't bother, as the 7805 is fast enough to filter out a reasonable amount of ripple.
Let me rephrase that...should there be a large cap supplying the "IN" pin of the 7805?
All depends on from where the 7805 is getting its juice. If the basic source is sufficiently filtered, I'd say don't bother, as the 7805 is fast enough to filter out a reasonable amount of ripple.
More than likely, it would be from a 9V wall-wart that was regulated.
A cosmetic observation, but wouldn't the schematic be clearer if the GNDs were joined up? When I first glanced at the schematic I was a bit confused. I assume the two rails have a common ground; the net name is the same at both points. But for me this would look clearer if the grounds were joined with a wire.
What's the current draw? Don't forget the heat-sinks on those 78xx
