near exact clock frequencies

[arod]

I am building a digital voltmeter at school and my design so far relies on 2 clocks at about 10kHz and 19kHz. It seems as if it would be hard to achieve these exact frequencies with a simple 555 timer because of the tolerances of resistors and capacitors. We have a device which measures capacitance very accurately in the lab. I was thinking I could get pretty close to the desired frequencies with a couple of pot resistors that can be manually tweaked until you get the desired frequency on the oscilloscope.
I have read that crystal oscillators have exact frequencies and are used in many clocks and microcontrollers, but I am not sure how (if at all possible) one could adjust its frequency since they are all in the MHz range. I am open to any suggestions on how to get an accurate clock for the 2 counters I am using in a digital voltmeter.

 

[chemelec]

Most Digital Clocks and Watches today use a 32.768 KC Crystal, which is than probably divided down.
But Not Sure To What Frequency or Exactially How its done.

I just bought a bunch of these to play with.
But if you have an Cheap Digital watches, You could Remove this 32Kc crystal from it.

Doing a test on them, They are VERY PRECICE Crystals.
Also VERY SMALL in size.

I am building a digital voltmeter at school and my design so far relies on 2 clocks at about 10kHz and 19kHz. It seems as if it would be hard to achieve these exact frequencies with a simple 555 timer because of the tolerances of resistors and capacitors. We have a device which measures capacitance very accurately in the lab. I was thinking I could get pretty close to the desired frequencies with a couple of pot resistors that can be manually tweaked until you get the desired frequency on the oscilloscope.
I have read that crystal oscillators have exact frequencies and are used in many clocks and microcontrollers, but I am not sure how (if at all possible) one could adjust its frequency since they are all in the MHz range. I am open to any suggestions on how to get an accurate clock for the 2 counters I am using in a digital voltmeter.

 

[ericgibbs]

hi arod,

You say two freq around 10kHz and 19kHz, the 10kHz is easy, but what is the precise freq you want 'about' 19kHz ?
[it makes a difference in how its derived]

The method, if you want exact, stable, freqs, would be a crystal oscillator, connected to a divider chips.

EricG

 

[JimB]

I am building a digital voltmeter at school and my design so far relies on 2 clocks at about 10kHz and 19kHz.

Do you really need two different frequencies, and how accurate do they really need to be?

It seems as if it would be hard to achieve these exact frequencies with a simple 555 timer because of the tolerances of resistors and capacitors. We have a device which measures capacitance very accurately in the lab. I was thinking I could get pretty close to the desired frequencies with a couple of pot resistors that can be manually tweaked until you get the desired frequency on the oscilloscope.

There is little point in selecting an exact capacitance and the adjusting the resistor. If you want to go that route just accept the value of the capacitor as it is and adjust the resistor.
R-C oscillators are not the best thing for accurate and stable frequency generation, unless you use high stability resistors and capacitors, the frequency will drift.
A oscilloscope is not the best thing for measuring exact frequency, about 2 or 3% accuracy at the best.

I have read that crystal oscillators have exact frequencies and are used in many clocks and microcontrollers, but I am not sure how (if at all possible) one could adjust its frequency since they are all in the MHz range.

Yes crystals are quite accurate. The accuracy and stability does vary, depending on how much money you want to spend!
But the cheapest crystals are orders of magnitude more stable than an RC oscillator.
The frequency of a crystal can be "pulled", either by a capacitor or an inductor, but the effect is limited to 10 parts per million (a rough ball park figure).

I am open to any suggestions on how to get an accurate clock for the 2 counters I am using in a digital voltmeter.

I would suggest deriving th two frequencies from a higher frequency crystal using dividers.
But first, do the frequencies have to be 10kHz and 19kHz?

JimB

 

[arod]

The frequencies do not have to be 10k and 19k, they just have to be scaled accordingly. I want one counter counting about 1.9x faster than another counter. How much drift would one expect with an RC oscillator? A couple percent off is not a huge deal. However, it could become a problem if the oscillator's error increases or decreases with time indefinitely.

Also, cost is an issue in this project. So I really want to minimize any unnecessary costs even if it means I am sacrificing a bit of accuracy.

 

[JimB]

All real world items have a tolerance and the closer the tolerance to some specified value, the more they cost.
This applies to all things, electronic, mechanical, optical, etc.

The voltmeter you are building, I assume that this is a "learning project" rather than something with a serious long term application.
In which case I suggest that you build your oscillators using 555s, the stability should be fine for what you are doing.
Select your resistors and capacitors with regard to their stability (if you have a choice of components, or just have to use whatever is in the box).
Make one of the timing resistors in your 555 adjustable, so you can adjust to the frequency you require.
Hint: If you calculate that you need a 25k resistor for the timing, dont use a 50k variable, use a 22k fixed and a 10k variable.
You will find that the adjustment is much easier.

JimB

 

[House0Fwax]

Hint: If you calculate that you need a 25k resistor for the timing, dont use a 50k variable, use a 22k fixed and a 10k variable.
You will find that the adjustment is much easier.JimB

Also, if at all possible, make the variable resistor a multi-turn kind. Saves a lot of messing around trying to get a skeletal preset anywhere near the resistance you need it.

 

[Sceadwian]

The two biggest factors in frequency drift are temperature and voltage. If you feed the osc circuit with a well regulated voltage and perhaps put it in a small enclousure that has an elevated temperature (oven controlled oscilator) the frequency will be a lot more stable. All caps/resistors/IC's for the osc circuit should be in the temperature controlled box.