Byron said:
Take a read here:
http://www.allaboutcircuits.com/vol_4/chpt_12/4.html
The 555 simply isn't specified to operate with such large resistor and/or
cap values. Caps are leaky.
Also you'll have a lot less to go wrong with the shift register because
there will only be two parts in the circuit instead of 6.
Probably. You can get more detail on the 555 here:
http://tinyurl.com/859tf
But the shift register will give you more repeatable results with less
effort.
BAJ
Hi, Jeff. Looking at the LM555 data sheet,
http://cache.national.com/ds/LM/LM555.pdf
you can see that the leakage current from the threshold pin (0.1uA
typ., 0.25uA max) is the limiting factor. If you've got a 12VDC power
supply, and a 1.6 meg series resistor, you've got 7.5uA charging the
100uF cap. With a 5V supply, you've got about 3uA. I can see how the
0.4uA max. leakage current will cause inaccuracies, but it doesn't make
the chip inoperative, even at 5V.
Of course, the OP can use a CMOS 555, which has almost no leakage
current at the threshold pin, making this a non-issue.
Of more importance is the leakage current of the cap. Some
electrolytic caps are specified as having several uA of leakage current
maximum new out of the factory. That's more of a problem.
The OP should use an electrolytic which is specified as having low
leakage, such as one available from Mouser (140-LLRL16V100). These
caps ($0.34 ea.) are specified as having 0.4uA max leakage.
Actually, one of the major problems here that nobody is talking about
is the power supply. The 555 changes state on 1/3Vcc and 2/3Vcc. If
the supply is unregulated or poorly regulated, the power supply sags
are going to change things quite a bit, too.
I believe the OP can get within 20% of his timing requirements with the
originally specified parts, assuming he has a well-regulated supply.
But, if it would make anyone happier, the OP can just multiply the cap
value by 10 (1000uF 16V, Mouser p/n 140-LLRL16V1000, $0.77 ea.), and
divide the cap values by 10 (start with 27K for the 30 seconds, and go
up from there to 160K). That woould make it a bit easier on the
leakage current budget, since 10X as much current is charging up the
cap. All legal, and according to the data sheet, permissible.
Since we're drawing this out at length, there's also a bit of a glitch
with the 555 in the astable configuration (oscillator) that nobody's
mentioned here. Usually you figure pulse high (charge) time as about
0.7 * (Ra + Rb) * C. However, on turn-on, the cap isn't charging from
1/3Vcc, it's charging from 0V. That means the equation for the first
output logic high is closer to the monostable constant, 1.1 * (Ra + Rb)
* C. In other words, the first high pulse is going to be 57% too long,
and the logic high is more than half the total period, so the first
whole cycle of the 555 is going to be at least 23% too long, even with
a mm value for Rb. That's out of spec here.
In days of yore, when men were men and women were glad of it,
4000-series CMOS used to rule the earth (mid '70s). Time delay relays
were improved significantly by replacing the old clunky PUT-based
timing circuits with a 4060, which is a divide-by-2^14 counter with a
built-in oscillator circuit, which only requires two Rs and a C, kind
of like the 555. The idea was that, even if one timing cycle went bad
because of a power glitch or noise from the relay, well, you had over
8,000 of them to timeout (the 2^14th output went high to drive the
relay, so it had 2^13, or 8,192 counts. Who cares? It's still
accurate.
Anyway, even with a divide-by-2^14 counter following it, I believe the
oscillations would be too slow for the 4060 oscillator. So I would add
a 4020 or 4040 counter after the 4060 to allow us to get the 4060
oscillator up to speed, and go with that. You would then have Mr.
Popelish's 4015 shift register after, with the last output locking out
additional clocks.
I would guess the OP has enough information to go on. With ST CMOS
555s in thru-hole going for $0.40 ea. at Mouser, and the 1000uF low
leakage caps going for $0.77 ea., I would think the OP could just pull
the trigger and put something together for about $7 in parts, and see
if it works. If not, he can get back with another post.
Good luck
Chris