Accurate resistors and capacitors for 555

Thread Starter

jaygatsby

Joined Nov 23, 2011
182
I want to have a 555 astable trigger every 10 ms. I want this to be accurate enough so that I can have it run for up to 999,999 seconds with minimal drift. Is it possible to select resistors and capacitors for use with the 555 that have the tolerance of accuracy that I require? Or are my goals too high?

Thank you
 

SgtWookie

Joined Jul 17, 2007
22,230
I'm afraid that you will be disappointed with the accuracy over that period of time.

If you want some reasonably decent accuracy, then look at using something like a CD4040B or CD4060B and a crystal. If you want better accuracy, use a TCXO (temperature compensated crystal oscillator) or OCXO (oven controlled crystal oscillator).

If you want to get some real accuracy, look at buying a surplus rubidium frequency source. They generally output a 10MHz signal; you can use a /100k circuit to get it down to 10mS.
 

Wendy

Joined Mar 24, 2008
23,415
RC oscillators (what you are talking about) are doing good if you get 1% drift. You can tweak it with a variable resistor, but that is as good as it gets in general.

As Wook said you can get much better with a crystal oscillator. There are techniques that can improve a crystal oscillator a lot, such as a crystal ovens.

When designing with electronics you have to decide what errors are acceptable, and attach numbers to it. There will always be errors though.

Did I see a circuit here, long time ago, with a 555 and a crystal? By jpanhalt?
Yeah, but there was nothing special about it. You could pretty much do the same things with a TTL or CMOS gate. It was interesting seeing a 555 crystal oscillator, but it was not that special.
 
Last edited:

PaulEE

Joined Dec 23, 2011
474
As an aside, a crystal and a microcontroller, such as a PIC 12F683, are great for precision timing applications.
 

Thread Starter

jaygatsby

Joined Nov 23, 2011
182
I was afraid crystal would be the answer you guys would give me. To be honest, I'm scared of crystals. I've glanced at schematics and I don't understand how to use them. But I'd like to... and with the help of this forum I am confident that I can! Should I buy a 10ms crystal for my purpose?
 

SgtWookie

Joined Jul 17, 2007
22,230
Crystals are described by frequency, not by time.
10mS is 100 Hz, which would be a HUGE chunk of quartz.

What you need is a crystal that is a standard frequency, and then divide that frequency down to get your 100Hz output.

You're saying you want it to trigger a 555 monostable every 10mS. What is the duration of the output of the 555 timer?

It is quite possible to have a divide-by counter and some logic ICs output your regularly timed pulse for the desired pulse width.
 

Thread Starter

jaygatsby

Joined Nov 23, 2011
182
Ok - I have a bunch of crystals. What is a good value to look for that can be easily divided to get 10ms? I don't have any dividers on hand but I do have flip flops, shift registers, and things. I can order a divider if need be.

Thank you
 

SgtWookie

Joined Jul 17, 2007
22,230
Let's do it this way instead.

You need a pulse every 10mS.
You have not told us the desired output pulsewidth.
You have not told us what kind of tolerance you have for the output pulsewidth, or minimum/maximum for it.
We need the above two pieces of information.

Then list all of the crystals that you have.

Note that crystals are actually quite fragile. If you drop them from more than about 16" on to a hard surface, you will most likely break the crystal free from its' mountings.

You will not be able to hear the crystal rattling around inside the package, as the vibrations are a much higher frequency that humans can hear.

You will need to test each crystal before considering it for use.
 

Thread Starter

jaygatsby

Joined Nov 23, 2011
182
I see -- as usual I am learning a lot from your posts Sgt. I appreciate it.

Pulse width doesn't matter to me. I want a rising clock every 10ms to time a bunch of counter chips. Since these counters might count up to a million times, I'd like the timing source to be accurate.

I have maybe 50 crystals. Here are some examples:

66.667Mhz, 18.432, 5.76768, 40.32, 32, 26.824, 23.04(in an IC type package), 40, 29.4912, 45.....

My understanding of dividers and whatnot is limited but I think my output frequency or period is going to be an integer multiple (or fractional) of the input frequency, so the crystals that are not round integers will be useless in my case?
 

SgtWookie

Joined Jul 17, 2007
22,230
My understanding of dividers and whatnot is limited but I think my output frequency or period is going to be an integer multiple (or fractional) of the input frequency, so the crystals that are not round integers will be useless in my case?
That's not quite the case.

Let's just grab one out of the middle.
23.04MHz / 256 = 90,000 (a dual binary counter would do this)
90,000/9 = 10,000 (a decade or BCD counter could be used for this)
From there it's pretty obvious how to get to 100Hz/10mS.

How about another - 40.32Mhz.
40.32Mhz / 64 = 630,000
630,000 / 9 = 70,000
70,000 / 7 = 10,000
And there we go with the obvious again.
 
Last edited:

Audioguru

Joined Dec 20, 2007
11,248
If you have a radio station on 1000kHz then use an AM radio front end and a divider to get a very accurate time period. Does the government still transmit some very accurate frequencies?
 

Thread Starter

jaygatsby

Joined Nov 23, 2011
182
That's not quite the case.

Let's just grab one out of the middle.
23.04MHz / 256 = 90,000 (a dual binary counter would do this)
90,000/9 = 10,000 (a decade or BCD counter could be used for this)
From there it's pretty obvious how to get to 100Hz/10mS.

How about another - 40.32Mhz.
40.32Mhz / 64 = 630,000
630,000 / 9 = 70,000
70,000 / 7 = 10,000
And there we go with the obvious again.
Another consideration - I won't have much room on the board so a crystal, its supporting components, and two or three dividers may be packed tight. Also the things you're saying are obvious (after selecting the crystal) are new to me! I was hesitating to go with a crystal over 555 because I've never wired one up. I'll need to get started learning :)
 

thatoneguy

Joined Feb 19, 2009
6,359
If you have a radio station on 1000kHz then use an AM radio front end and a divider to get a very accurate time period. Does the government still transmit some very accurate frequencies?
Not anymore on AM standard band. You can still get the one second tone from WWVB via shortwave radio, and a higher frequency from one of the other short wave transmitters at the WWVB site, I forgot what all the bands/frequencies are now, but they do offer precision timebases from their bank of fountain clocks for public use.

The problem there is that tuning into the station to get \(1\times10^{-15}\) ppm accuracy would take more space than an oven controlled crystal, which is good for about 0.001 ppm, or a standard crystal which typically drifts around 20 ppm
 

Thread Starter

jaygatsby

Joined Nov 23, 2011
182
Not anymore on AM standard band. You can still get the one second tone from WWVB via shortwave radio, and a higher frequency from one of the other short wave transmitters at the WWVB site, I forgot what all the bands/frequencies are now, but they do offer precision timebases from their bank of fountain clocks for public use.

The problem there is that tuning into the station to get \(1\times10^{-15}\) ppm accuracy would take more space than an oven controlled crystal, which is good for about 0.001 ppm, or a standard crystal which typically drifts around 20 ppm
Stations on 2.5Mhz, 5, 10, 15 I think. Someone should make chips that output a 1Hz square wave based on these stations.
 

SgtWookie

Joined Jul 17, 2007
22,230
Well, if you are wanting something really compact, then you will need to use something like a microcontroller; as discrete logic IC's will be rather numerous otherwise - unless you start out with the right frequency crystal to begin with.
 

thatoneguy

Joined Feb 19, 2009
6,359
Well, if you are wanting something really compact, then you will need to use something like a microcontroller; as discrete logic IC's will be rather numerous otherwise - unless you start out with the right frequency crystal to begin with.
32.768 kHz is a great freq to start with, and widely available.

If it isn't obvious, 2^15, and you can divide it down quick using 1 binary and one decade counter, if you use TSSOP packages, it'll be pretty tiny.
 

Thread Starter

jaygatsby

Joined Nov 23, 2011
182
I see different crystal oscillator circuit designs on the net. Is there a nice simple one that I can use in my design, that you would recommend?

Thank you
 
Top