Timing Circuit

Go to

THANKS! This chip looks perfect. Good ol TI always seems to come through for me.

The 74hc4060 is an attractive chip with its on-board oscillator, but you should be aware cmos inverter-oscillators aren't very predictable (from IC to IC), nor very stable after the timing components are adjusted for your time delay. If you're only looking for a rough 10 minutes, fine, but if you want a semi- accurate 10 minutes you may prefer to use a second oscillator IC, such as a cmos 555, etc. Also, the 'hc4060 isn't the most attractive choice, because its 2^(14-1) = 8192-cycle timing requires a pretty low oscillator frequency for 10 minute timeout (13.65 Hz). Other parts have a higher divider ratio, such as Motorola's (now ON Semiconductor) mc14536, with 24 bits.

The mc14536 also has an onboard oscillator, in case you still want to give that a try. I first started using the '4536 in the early 70s, and shipped thousands of instruments that relied on its cool 4-bit divider-select feature. We also featured the '4536 in our "hour of power" circuit in AoE (page 972), although I'm not happy with circuit we show there. $0.76 at DigiKey, very nice.

BTW, TI also makes the '4536, as their cd4536B, if you prefer them. ST calls it the hcf4536B. The TI and ST datasheets, which were modeled after the Intersil/Harris datasheet, do a better job of showing how to use the powerful timer functions,

Alternately, if you want to stick with a 16-bit IC, ON Semi's mc14541 is more attractive than the '4060, because it features an output flip flop with a polarity selection feature (and it also has an onboard oscillator), only $0.50 to 0.58 at DigiKey.

BTW, you can get free samples of both ICs from ON Semi. And TI.

Thanks for the info. As far as accuracy, I'm looking for something that will work with a +/- 30 seconds or so. I'm assuming that any change in performance based upon manufacturer and/or temperature fluctations will not be a worry. Also, I don't really care as to the number of bits the chip has for an output. I'm thinking I'll just run the desired outputs through an AND that sets a S-R flip/flop. I'm sure there is a better way of doing this and once I start with the design maybe I'll come across some features that can help me cut down the need for extra circuirty. Perhaps that's what the mc14541 does. Doing this stuff in actual practice is quite a bit different than what I did for school. It seems that there are so many permutations of doing the same thing it's dizzing. It seems almost impossible to find the absolute cost effective and most efficent way of doing something. Too many options and not enough time.

THANKS FOR YOUR HELP! Jimbo

Win,

Is there a "system" for the naming of the IC out there? I keep reading things that tend to "hint" that there is a systematic way of naming these things. Isn't the formation usually "LETTERS" + "NUMBER" + "LETTERS". Have any idea as to the method these things are named?

That is only 5%.

You don't need to do that unless you want the timeout to be programmable, in which case none of the suggested parts are that good.

Once you get close, the differences in implementation become negligible.

This has to be a classic job for a baby microcontroller. Typically a tiny PIC.

That involves learning about microcontrollers and doing programming though. Doing it in hardware is 'easier' for a beginner but will use lots of parts in comparison. It *won't* be 'just an IC' although the microcontroller method can be.

Graham

Acckk!! You're right, I mentally added in an extra factor of 60 and came up with a moderately-painful 1:1200 design. Hah!

========== Oops, I made a mistaken post, with mental calculations off by 60x. Let's leave it here, below, but for sake of discussion let's assume that Jimbo's requirement was for a 0.5-second, 0.1% accuracy. :-) In practice, with his 5% spec, he can use a '4541 IC with a 294k 1% resistor and a 27nF 5% film capacitor. Or better, 2% caps like the nice Panasonic parts offered by DigiKey. That way tolerances will be in the IC, and Jimbo may be able to dispense with adjustments. ===========

Jimbo, your 30-second spec is one part in 1200, within the range of analog relaxation-oscillator stability with RC components, provided they are properly chosen. A 16-bit divider (15-bits in timer mode) like a '4541 will have its oscillator running at 32768/600 = 54.6Hz, which isn't bad. Using f = 1 / 2.3 RC, and R = 249k plus a 100k pot, we get C = 0.027uF, a moderate-value capacitor for which small stable versions may be practical. Still, I'm tempted to suggest instead an 8-pin uP with a 32kHz resonator. Have you considered that possibility?

Check out the 4536 as well. But a single 8-pin microcontroller can give you the required accuracy with no external parts, or 10x better with one external part. You would have to program it.

Best regards, Spehro Pefhany

There is next to no sense at all in it.

The first few numbers/letters may tell you something but not a lot.

Graham

The first 1 or 2 digits is usually the heater voltage.

Regards PN2222A

The first 1 or 2 digits is usually the heater voltage.

Regards PN2222A

so you have a 300mA heater??

-- Sell your surplus electronic components at

On 30/01/2006 the venerable PN2222A etched in runes:

Actually the first letter is the heater voltage and the next one or two are the construction. For example ECC83 is a 6.3VAC heater dual triode and ECF86 is 6.3VAC heater triode/pentode.

...

Actually, I take that back, Jimbo no doubt will need a pot, even if he does use a 2% capacitor. The chip's timing factor of 2.3 has a tolerance as well, although none of the manufacturers will give you a hint as to what it is. NSC's cd4541B datasheet does have a frequency-deviation graph from which one can see about 2% drop vs supply from 15 to 10V, and another -8% for 5V, along with a 1% increase for a 50-deg C increase in temperature. One useful thing we can take from this graph is that for operation near 5V one should use a constant of 2.5 instead of 2.3 in the equation.

microcontroller? I've programmed them before but I don't know of any that are as cheap as these other chips. Also, I'm doing this to pay my way through grad school and I'm working with a very small upstart so any development boards/software is out of the question becuase of the price. This is going to be a "large volume"/"low margin" product so every extra dollar we can squeeze out of the design will have a huge impact on the bottom line. Are you thinking about something like the