555 Is This Feasible?

[@rt]

Hi,
I want an oscillator with a square waveform duty cycle 9 mins low (approx), and 500ms high (approx).
It is important that the low duration doesn’t exceed 10 minutes, so 9 minutes is just a generous error margin.

It appears the only difference between t1 and t2 in the diagram is achieved by R2.
Is it feasible to expect the desired outcome with the same capacitor C1 responsible for both durations?
It occurs to me that I could replace R2 with a very high value resistor to increase t2,
but can I keep t1 very low at the same time? The capacitor can’t be so large that t1 blows out.

[ledtester]

In this circuit the charging of the cap is determined only by R1 and the discharging of the cap only by R2.

If you set C1 to 100uF you can get a 9 minute off period with R2 set to:

R2 = 9*60/.693/(100*10^-6) = 7.8 M ohms.

and for 0.5 seconds:

R1 = 0.5/.693/(100*10^-6) = 7.2K ohms

If you have a 1000 uF capacitor you can divide R2 and R1 by a factor of 10, i.e. 780K ohms and 720R respectively.

[james_s]

9 *minutes*? It may be possible, but I would not expect it to be very precise. It would be trivial to do this with something like a attiny12 microcontroller though.

[RoGeorge]

Minutes long timing might not be reliable with a 555, would drift with temperature/humidity, and if you need more than one timer, then you'll need to calibrate each one manually, which would be very time consuming considering you'll have to wait about 10 minutes for each pulse.

A microcontroller would be better, and for very small ones it might cost about the same if not cheaper than a 555.  Many microcontrollers have an internal calibrated clock which doesn't need any external RC or Quartz.  Usually the error of the internal oscillator is less than 10%, guaranteed over all the range of temps and voltages, and that can be calibrated for less than 1% if needed (without having to wait 10 minutes to check the calibration results).

[jpanhalt]

You might consider the CMOS version ICM7555 for longer time constants or the CMOS programmable version CSS555C
(http://www.customsiliconsolutions.com/downloads/Revised%20Standard%20products/CSS555C_Spec.pdf) for extremely long periods.  The latter will also work for 9 minutes.

[@rt]

Hmm... well for everyone else... It’s 10 minutes max, but can be arbitrary.... as I said in OP, there’s a minute error margin built into that time.
The 9 minutes could even be lower.
500ms is minumum, but I have mitigated error already with slightly higher value resistor.

Micro I could do easy, but is also a bit overkill... I know not on price, but on complexity.

[@rt]

Introspection.

Nobody else would know that the autism here won’t accept a million transistors doing the job of an astable multivibrator. So I should actually thank everybody. Thanks
No micro in this case for a hobby project anyway. For a larger run for commercial purpose, the autism would allow cost to dictate method.
The initial question really is about freq drift.. ie. someone else who has done something similar please sanity check that temp coefficient of discrete components won’t ruin it.
Now that I have constructed physically, I’ll accept a shorter low time (t2) until I am sure the 10 minutes won’t elapse.
 

[MikeK]

Capacitors have a typical tolerance of 20%.  That affects your timing precision, of course.  If 10 minutes is a hard limit you're going to want better precision than 20%.  Many microcontrollers have 1-2% internal oscillator accuracy.

[Circlotron]

I've seen 555 circuits with a 100uF cap and several megs of charging resistor and sometimes the leakage of the cap would be such that it would only charge partway and never get to the timing threshold.

[CaptDon]

The 1000uf / 780K / 720 answer should not be used as I think 1K was the minimum data sheet value for that resistor. The suggestion of using the CMOS version is even recommended by the 555 datasheet for longer time delays. You can get much longer time delays using the CMOS version while avoiding using high leakage unstable electrolytic capacitors. I believe the CMOS version of the 555 datasheet even says right in the bold print 'Delays from minutes to hours'.

[MrAl]

Hi,
I want an oscillator with a square waveform duty cycle 9 mins low (approx), and 500ms high (approx).
It is important that the low duration doesn’t exceed 10 minutes, so 9 minutes is just a generous error margin.

It appears the only difference between t1 and t2 in the diagram is achieved by R2.
Is it feasible to expect the desired outcome with the same capacitor C1 responsible for both durations?
It occurs to me that I could replace R2 with a very high value resistor to increase t2,
but can I keep t1 very low at the same time? The capacitor can’t be so large that t1 blows out.

Hello there,

You should know right off the bat that this is not a good way to do it even if it does work.  So get off the bat and on the ball and find another way to do it or one day you'll be stuck out in left field, and if i can think of any more baseball metaphors i'll get back here later :-)

But really RC timing with larger caps is a little problematic both in initial set points and long term age factors.  There are other ways.

One way in particular that doesnt involve a micro controller is to use a timer like the 555 with reasonable cap and resistor values and a digital divider chip that divides the 555 frequency down to a much lower frequency.  You can go pretty far with this idea because then the 555 will be operating normally and digital dividers are very very very repeatable in their divide ratios, and in fact may be flawless for the most part.  This technique is actually used in very important applications where a change in timing could blow out very expensive equipment. In fact in those cases even an LM339 can be used as the base frequency oscillator with the dividers following.

If you really MUST go with a 555, then perhaps you can try multiple 555's.  If you have one 555 trigger the next 555 you get a longer delay time using lower value caps and more reasonable resistor values.  If you use more than two 555's you can probably go pretty far with this but you have to have the room for the chips on the board.  Dont know if you are into SMD or not but the 555's also come in a dual package type so you get two 555's on one chip, and two chips means you have four 555's to work with.  In fact, i would bet there is a better solution using LM339 comparator chips with a little thought.  I might give that some thought myself it may be interesting.

I've done delays that go into the days and one i remember best is a special kind of battery charger that had to charge the cells once every 12 hours.  The timing was around 11.75 hours off, then 15 minutes on.  I just have to say it was pretty darn easy to do that using a very very cheap microcontroller chip at the time made by MicroChip.  That takes a bit of learning to do if you never programmed a microcontroller before, but today there is no excuse because of products like the Arduino.  The Arduino microcontroller is used by hobby folks all over the world and with good reason. There is a host of information on the web and sample programs that are so easy to change you'd wonder why you never used one before.
With the Arduino IDE you dont have to deal with the chip itself directly, dont even have to know how to solder.  You can get jumpers that you plug into the Ard board and run a program that came with the IDE.  From there a minor mod to the program and you can get delays that probably last longer than the age of the universe (ha ha).  Seriously though you can get hours, days, months, years, you name it.  You just need a 5v supply or even a 9v DC wall wart.
The cheapest ones are the Nano and they are small too.  The entire board is something like 1 inch by 3 inches and all the parts are on board already.  If you need an output transistor that's really all you need to add except for maybe a resistor or two.  If you dont know how to mod the program im sure someone here would surely help you and it would not be much to change either, maybe a few lines of code which is very human readable too.  This kind of solution should get you a sure home run.

Ok no more baseball metaphors for now, good luck with it and if you can please come back and let us know how you made out with it, whichever way you decided to do it.

[Terry Bites]

I bumped ito this few weeks back. No nasty diode temcpco issues. www.edn.com/add-one-resistor-to-give-bipolar-lm555-oscillator-a-5050-duty-cycle/

[james_s]

I think a microcontroller will be cheaper, and I'm certain it will have a lower component count. This is an application where even one of those 3 cent Padauk micros would do just fine, although the cheapest mainstream micro you can find is going to be a lot easier to develop for. Digikey has the ATTINY10-TS8R for 47 cents each in singles, SOT23-6 package is smaller than a 555.

[Zero999]

Micro I could do easy, but is also a bit overkill... I know not on price, but on complexity.

What leads you to this conclusion? An 8 or 6 pin microcontroller, a decoupling capacitor and some code is much simpler, than a 555 circuit.

[james_s]

Micro I could do easy, but is also a bit overkill... I know not on price, but on complexity.

What leads you to this conclusion? An 8 or 6 pin microcontroller, a decoupling capacitor and some code is much simpler, than a 555 circuit.

Probably lower power consumption too, possibly dramatically lower. The micro could spend the vast majority of the time in deep sleep mode drawing a few microamps, it could potentially run for years from a coin cell.

[Bud]

Micro I could do easy, but is also a bit overkill... I know not on price, but on complexity.

What leads you to this conclusion? An 8 or 6 pin microcontroller, a decoupling capacitor and some code is much simpler, than a 555 circuit.

How it is simpler if it requires writing code and a programmer to burn it in. The 555 does not need any of that.

[james_s]

Micro I could do easy, but is also a bit overkill... I know not on price, but on complexity.

What leads you to this conclusion? An 8 or 6 pin microcontroller, a decoupling capacitor and some code is much simpler, than a 555 circuit.

How it is simpler if it requires writing code and a programmer to burn it in. The 555 does not need any of that.

It has less parts on the board, that's the definition of simpler. Device programmers are cheap and most people who are serious about electronics will already have one. Writing the code for a project like this is trivial, I could probably do it in 5 minutes and I'm not much of a programmer. The 555 needs a pile of external components, the capacitor alone will cost more than the entire circuit using a microcontroller. (MCU and decoupling cap)

[Nusa]

Capacitors have a typical tolerance of 20%.  That affects your timing precision, of course.  If 10 minutes is a hard limit you're going to want better precision than 20%.  Many microcontrollers have 1-2% internal oscillator accuracy.

All the components have tolerance values, but that just represents factory binning. After you've received them and tested actual values to be acceptable, the tolerance marking loses meaning. All that's left is environmental drift and long-term aging.

[Vincent]

This may be a completely stupid idea, but say OP decides not to the uC route, what about a 555 with a shorter clock cycle running a binary counter like a CD4040 or something? Timing components for the 555 could be higher precision and/or more stable ones.

[jpanhalt]

Post #5 refers to the CSS555 chip -- a 555 with built-in programmable counter.

Here are links to CSS's application notes and to the specific note that applies to programming. It looks fairly simple, but then, if you need to write an MCU program to program a 555, why use a 555?  I suspect a development board is or was available to do that for you.  No idea of the cost.

https://www.customsiliconsolutions.com/products-for-ASIC-solutions/standard-IC-products_no.aspx
https://www.customsiliconsolutions.com/downloads/Revised%20Standard%20products/CSS555_App_Note1_Serial_Interface.pdf

[MrAl]

Micro I could do easy, but is also a bit overkill... I know not on price, but on complexity.

What leads you to this conclusion? An 8 or 6 pin microcontroller, a decoupling capacitor and some code is much simpler, than a 555 circuit.

How it is simpler if it requires writing code and a programmer to burn it in. The 555 does not need any of that.

The Arduino has code that is already written for pulses that would be very simple to modify.  Anyone here could easily do that i think.

[MrAl]

This may be a completely stupid idea, but say OP decides not to the uC route, what about a 555 with a shorter clock cycle running a binary counter like a CD4040 or something? Timing components for the 555 could be higher precision and/or more stable ones.

No it's not stupid at all, and in fact it was already mentioned in reply #10 in this thread.
There are even counters that divide by a large number in the thousands so that brings the components down to the normal size and value.  Perhaps one 555 timer chip and one counter chip would do it.  You do have to reset the counter at the right time though with possibly a logic gate.  Since there are two time settings it may be necessary to use two counters and some glue logic.

[MarkF]

Post #5 refers to the CSS555 chip -- a 555 with built-in programmable counter.

Here are links to CSS's application notes and to the specific note that applies to programming. It looks fairly simple, but then, if you need to write an MCU program to program a 555, why use a 555?  I suspect a development board is or was available to do that for you.  No idea of the cost.

https://www.customsiliconsolutions.com/products-for-ASIC-solutions/standard-IC-products_no.aspx
https://www.customsiliconsolutions.com/downloads/Revised%20Standard%20products/CSS555_App_Note1_Serial_Interface.pdf

Am I correct that the serial clock to program the EEPROM is between 2 MHz and 5 MHz?
Not very user friendly if you have to have a microcontroller to program it.
It would be nice if you could program it from a standard serial port at normal baud rates (i.e. in the low KHz).



There is a lot of talk of meeting the OPs tolerances...
He says he'll accept anything between 500ms and 10min.
Just how hard can it be to meet a tolerance range like that?   

Hmm... well for everyone else... It’s 10 minutes max, but can be arbitrary.... as I said in OP, there’s a minute error margin built into that time.
The 9 minutes could even be lower.
500ms is minumum, but I have mitigated error already with slightly higher value resistor.

Micro I could do easy, but is also a bit overkill... I know not on price, but on complexity.

[ledtester]

Am I correct that the serial clock to program the EEPROM is between 2 MHz and 5 MHz?
Not very user friendly if you have to have a microcontroller to program it.
It would be nice if you could program it from a standard serial port at normal baud rates (i.e. in the low KHz).

Here is the app note on programming the chip:

https://www.customsiliconsolutions.com/downloads/Revised%20Standard%20products/CSS555_App_Note1_Serial_Interface.pdf

Looks like just a standard serial EEPROM. There's no mention of a maximum clock period. The minimum and typical durations are 500ns and 200ns respectively which corresponds to 2MHz and 5MHz but that's an upper bound on the clock speed.

[magic]

CD4060? It's like 555 but with built-in frequency divider. It would probably need to trigger an ordinary 555 (or some discrete contraption) for the 500ms thing.

I have seen those buggers used for multi-hour timeout in dumb NiCd chargers.