Monostable timing components - range of values

[IconicPCB]

Please see the attached data sheet  - CD4538 CMOS monostable.
Data sheet claims ide pulse width range, 

 Wide pulse-width range        1 ms to infinity

Further it states on page 4 under Operating Conditions minimum timing resistance value of 5K whilest maximum value is not specified , how ever it is limited by a clause which explains:

 "...The maximum usable resistance R X is a function of the leakage of the Capacitor C X , leakage of the CD4538B, and leakage due to board layout, surface resistance, etc. ..."

At the same time Cx is not defined at all.

What maximum pulse width might be reasonably and repeatably achieved using current technology tantalum capacitors as part of timing circuit?

[MagicSmoker]

Setting aside the kludge factor for trying to get long pulse widths out of any kind of monostable, your main limiting factor is the leakage current spec of the capacitor. If you need to generate pulses more than 10 seconds long then an X7R ceramic is probably going to be a better choice, especially if ambient is >35C (leakage follows the Arrhenius equation, so doubles every 10-11C increase in ambient temp).

[jfiresto]

... At the same time Cx is not defined at all.

What maximum pulse width might be reasonably and repeatably achieved using current technology tantalum capacitors as part of timing circuit?

Some manufacturers specify a minimum 5 nF Cx value, but I wonder if that is to preserve the simple t = RC timing formula. Most suggest adding a discharge protection diode/resistor if Cx is 500nF or more: to protect the (1)4538's internal ESD network should VDD plummet on powering off.

A polyester film capacitor would also be low leakage and I believe would be more stable than an X7R ceramic. I might look at the CD4541 to produce a really long pulse.

[IconicPCB]

Magic Smoker,

Thank You for the comment.
I had a look at Tantalum capacitor low leakage specs and came to the same conclusion.
Leakage current is is typically specified as some constant  * capacitance * voltage.

It will be in the low microamps for high value capacitor .  The constant in above expression is reduced by operating the capacitor at reduced voltage. Typically 25 to 40% of rated value for a halving in leakage current.

Based on the above and the effect of leakage current i estimate 1 Meg and may be up to 20uF night be the largest working combination of timing components where the leffects of leakage current  might be acceptable in terms of stability and repeatability of pulse width. 

jfiresto,

A good suggestion; I had seriously considered it. A major problem with CD4541 is that it has

"... A setup time equal to the width of the one-shot output is required immediately following initial power up, during which time the output will be high..."

If the time in question is in the order of  a hundred seconds the powerup set up becomes untenable.