Thanks again to those who have expressed interest in my project. The objective is a high performance circuit for the measurement of frequency or pulse width. Also, to investigate different methods of processing signals for time-based measurements. Before you start ripping me a new one, please remember this is a work in progress. At present, it's just an idea being simulated using generic components. I still have alot of work to do before it's a finished product. Also, before telling to "just use a PIC", I want to say this project isn't targeted as a stand alone system; rather, it's targeted as an embedded subsystem for other projects where measurement of rapidly changing frequency is desired.
Most of the analog frequency measurement circuits or tachometers I've seen rely on some combination of charge pumps and averaging networks. Those components can cause a lag between fast frequency changes and the eventual accurate measurement output. My project attempts to rapidly change the output in response to frequency changes. In the attached diagram, U1 outputs a signal which is proportional to the integral of the signal's "low" period (green trace). The input network insures the integrator is reset to measure each pulse independently. The first FET follows the integrator’s output during the measurement period, and holds the peak value during the integrator’s "reset" period (blue trace). C2 keeps the value, which is in turn sampled by the next FET/capacitor. This sampled value is buffered by output by U4 as the final measurement signal. In the simulation, the output is at 97% of the final value on the very first sample.
For frequency measurement, the signal needs to have a 50% duty cycle (or 50% mark/space). This can be accomplished by dividing the input signal with a flip-flop, and scaling the output to account for the division. Alternatively, the circuit can measure the pulse width of a periodic signal without pre-processing.
There are several issues with the present circuit layout. For one, care must be taken to not overload U1, as that messes up the measurement totally. In the simulation, I use a pretty unrealistic signal source, until I have a good solution for that. Also, I have to find the right combination of amplifies and "hold" capacitors to keep the accuracy high. The values of the caps have all been arbitrarily chosen, and will ultimately depend on the frequency of the input signal. The idea here is to make the circuit independent to frequency, as much as possible. But for truly wide banc measurements, a selection of caps using rotary switches will be necessary. Also, R16 and R17 are not real resistors, but meant to model the opamps input currents for the simple components being used.
Thanks for reading. Any suggestions or help is appreciated.
Most of the analog frequency measurement circuits or tachometers I've seen rely on some combination of charge pumps and averaging networks. Those components can cause a lag between fast frequency changes and the eventual accurate measurement output. My project attempts to rapidly change the output in response to frequency changes. In the attached diagram, U1 outputs a signal which is proportional to the integral of the signal's "low" period (green trace). The input network insures the integrator is reset to measure each pulse independently. The first FET follows the integrator’s output during the measurement period, and holds the peak value during the integrator’s "reset" period (blue trace). C2 keeps the value, which is in turn sampled by the next FET/capacitor. This sampled value is buffered by output by U4 as the final measurement signal. In the simulation, the output is at 97% of the final value on the very first sample.
For frequency measurement, the signal needs to have a 50% duty cycle (or 50% mark/space). This can be accomplished by dividing the input signal with a flip-flop, and scaling the output to account for the division. Alternatively, the circuit can measure the pulse width of a periodic signal without pre-processing.
There are several issues with the present circuit layout. For one, care must be taken to not overload U1, as that messes up the measurement totally. In the simulation, I use a pretty unrealistic signal source, until I have a good solution for that. Also, I have to find the right combination of amplifies and "hold" capacitors to keep the accuracy high. The values of the caps have all been arbitrarily chosen, and will ultimately depend on the frequency of the input signal. The idea here is to make the circuit independent to frequency, as much as possible. But for truly wide banc measurements, a selection of caps using rotary switches will be necessary. Also, R16 and R17 are not real resistors, but meant to model the opamps input currents for the simple components being used.
Thanks for reading. Any suggestions or help is appreciated.
Attachments
Last edited:
