I have max 8 MHz sensor signal and I would like to detect the at least 8 Hz frequency change and collect the data. Is there anybody who could suggest me what to use for such high precision? PS. The change in amplitude is not important. Only thing i need is to detect the frequency change. Low power should be considered so I can not use the solutions that requires high power.
Frequency Change Detector
- Thread starter meceng
- Start date
It seemed like that high resolution is not possible with frequency counter. Do you have any specific products i could us?
I do not have knowledge on it so i will check it. Thanks for the suggestion. If you have any source i can get the most accurate information i would love to hear from you.
Hello,
Here is a link:
https://aa7ee.wordpress.com/tag/direct-conversion-receiver/
When you use a 8Mhz crystal, the output will be round 0 Hz, when the input signal is 8 Mhz.
When the frequency is higher or lower as the crystal frequency, a signal will be outputed.
You can also use an offset, say 8Mhz + 1000 Hz.
When the frequency is exact 8 Mhz, 1000 Hz will be present at the output.
When the frequency is a couple of Hz higher or lower, it will be added or subtracted from the 1000 Hz.
Bertus
Here is a link:
https://aa7ee.wordpress.com/tag/direct-conversion-receiver/
When you use a 8Mhz crystal, the output will be round 0 Hz, when the input signal is 8 Mhz.
When the frequency is higher or lower as the crystal frequency, a signal will be outputed.
You can also use an offset, say 8Mhz + 1000 Hz.
When the frequency is exact 8 Mhz, 1000 Hz will be present at the output.
When the frequency is a couple of Hz higher or lower, it will be added or subtracted from the 1000 Hz.
Bertus
Probably not the resolution you need - but there's frequency to voltage chips. You can follow that with a data slicing comparator, you can find examples from old disk drive service manuals. The average input voltage charges up a capacitor via a resistor and provides Vref for the comparator - the input goes directly to the other comparator input. You probably need a cross between that and a window comparator to detect deviations of both directions.
BobaMosfet
- Joined Jul 1, 2009
- 2,113
Use an appropriate divider to input the 8Mhz into a CD4046 PLL, and the feedback loop. You can then monitor the deviation.
It kinda depends on how you define your data. If detecting 8 hertz/parts change out of 8 million......is all you need......I believe one could do it with low power. Even come close to quantifying the change.
But if you have to precisely measure the frequency that close.....then no......not with low power.
But if you have to precisely measure the frequency that close.....then no......not with low power.
Whether it is a direct conversion circuit (very cute) or a frequency counter, you are trying to resolve a frequency change of 1 part per million. To do that you will need a frequency reference (crystal oscillator, etc.) with better precision than that. That is expensive. In ascending order of price, a double-oven crystal oscillator, a GPS-disciplined oscillator, a rubidium standard. The last two output 10 MHz, so this would have to be divided or phase locked down to 8 MHz for the receiver, or up to 100 MHz for a frequency counter.
Or something like that.
ak
Or something like that.
ak
Don't some some of these ultra low power and super fast controllers run at and above 80 MHz now. Some even higher.
And don't most have built in high precision and stable oscillators now? Not at 1 ppm of course...but still....low rates of drift? Run time counters are common also.
With these clock rates and slow drift.....can't we make some RELATIVE measurements at 8 MHz?
The oven of course....kills low power.
And don't most have built in high precision and stable oscillators now? Not at 1 ppm of course...but still....low rates of drift? Run time counters are common also.
With these clock rates and slow drift.....can't we make some RELATIVE measurements at 8 MHz?
The oven of course....kills low power.
As I see it, to resolve 8Hz in 8MHz, however you do it, implies a crystal (or other) reference frequency accurate (or stable) to better than 1 ppm, at least in the short term: but if both the DUT and the reference frequency were to drift in sympathy in the long term it should be possible to detect a short-term frequency shift. That is why I asked how quickly the DUT frequency was likely to change.
AFAICR: the 4046 falls over somewhere around 2MHz - but there are faster types.
There are probably PLL chips for TV tuners that go to nearly 1GHz. But they might use ECL prescalers on the front end.
