4060 counter not accurate?

[kong]

Wondering if I'm doing this correctly....have a 4000 series 4060 divider/oscillator hooked up in the typical RC configuration seen in the datasheet. Rt is 470k and Ct is 0.22uf. R1 is 2M. The circuit oscillates, but the frequency seems quite a bit off. I have a frequency meter hooked up to pin 7 (divide by 16). My frequency, when calculated, is 4.20Hz. Divide by 16 gives me 0.26hz at pin 7. When measured with my frequency meter I get 0.323hz. Is this much error normal with an RC configuration? I understand the RC setup is typically pretty inaccurate, and the frequency can vary with temperature and supply voltage, but this seems a little extreme. I have had accurate results with a watch crystal which I used as the heart of a digital clock I built. But I was hoping the simplicity of an RC network would give me better results for my new application which is a simple cycle timer and doesn't require pinpoint accuracy. This still seems like a ways off. Any advice? Do you think adding a fixed voltage regulator on the input would give me better results? Like a 7812?

I've used polarized electrolytics for experimentation purposes only and I would get longer than calculated delays. Some frequently published timer schematics using this chip recommend putting electrolytics in in series to half the capacitance and create a non-polarised high capacitance Ct. I've tried those circuits and they seem to be more accurate, using a 500k variable instead of a fixed resistor, and a much higher value resistor connected to pin 11...like 4M7.

Any explanations would be welcome, thanks!

 

[hevans1944]

Pinpoint accuracy is a relative term. Typical accuracy of an RC timer with off-the-shelf components can vary quite widely and is usually temperature and voltage sensitive, depending on component selection and how the RC charge/discharge switching levels are determined. Probably the most accurate RC circuit uses the ubiquitous 555 timer whose timing is NOT sensitive to power supply variations, within reasonable limits. If you insist on using an RC timer oscillator, start with a 555 series IC and select precision resistors and capacitors with low temperature coefficients.

Better yet, use a crystal or ceramic resonator to construct your oscillator and divide the frequency down to whatever range you need for timing. RC oscillators have their uses, blinking LEDs in toys and such, but serious timing requires a predictable and stable oscillator. Sorry 'bout that.

 

[(*steve*)]

The other issue is that depending on how you probe the oscillator, the frequency could be changed when you try to measure its frequency. This is another sort of variability (stray capacitance, loading effects) you need to consider with oscillators, with RC oscillators being potentially quite vulnerable.

 

[hevans1944]

I almost forgot... an inexpensive and very accurate (long term) timing source is the mains frequency. This is corrected by the utility every day to ensure the total number of cycles per day is always the same, meaning line-synchronized clocks do not gain or lose time.

You can start with a higher frequency generated by a VCO, divide that down to the power line frequency, compare the divided oscillator frequency with the power line frequency, and use the difference in a PLL circuit to control the VFO. There are ICs made for this purpose, and a simple RC network in the feedback path to the VCO will guarantee a stable control loop response. A nice thing about this approach is you can program the divider to produce a range of output frequencies, all phase-locked to the power line frequency.

Of course, if your project is battery operated and off-line from the mains, you need another reference source for the PLL. In addition to the crystal and ceramic resonators suggested in post #2 above, I would recommend using a GPS receiver to get a very accurate 1 Hz reference clock. I believe there are also MEMS (micro electro mechanical systems) resonators available for timing purposes, and their accuracy and stability now compete with quartz crystals. Not sure about the cost of MEMS resonators in onesie twosie quantities.

There is a plethora of stable and accurate timing sources that are probably out of scope for your project, things like rubidium or cesium atomic clocks. Some dedicated radio amateurs have purchased rubidium clocks from eBay, just because they can, and there is a certain thrill in knowing an oscillator frequency to such high accuracy. In fact, the ARRL (American Radio Relay League) sponsors a yearly frequency measuring contest that encourages such "best efforts" on the part of interested parties world wide.

So there you go... a whole spectrum of accuracy and precision is available to the electronics enthusiast , all at pretty reasonable cost.

 

[kong]

Very nice information thank you. I like the GPS idea, especially with a PLL. Although, aren't they pretty accurate just by themselves? I thought I read somewhere that the satellites they talk to communicate with atomic clocks and derive their accuracy ultimately from them.

 

[KrisBlueNZ]

The PLL is used if you need to derive a higher frequency from a very accurate lower frequency. A mains frequency of 50 or 60 Hz is too low for many applications, hence the PLL. Some GPS receivers provide a 1PPS (one pulse per second) output which is even slower. Some provide a 1 kHz signal as well, and these would be better for timekeeping applications.

 

[kong]

Oh I discovered my error. Calculated times for any output on the IC are for the combined H/L cycle. I was comparing the calculations with just the high portion and wondering why everything was off by 50% or more. Bonehead mistake. After looking at my results and dividing by 2 everything works out appropriately. RC still isn't accurate, but more like what I was expecting.