Electronic – How to improve the Freq to Voltage circuit

circuit analysisfrequency-measurementtroubleshooting

I had a term project which was based around Analog-to-Digital convertors.
We used anLM2907 Freq-toVoltage convertor to take a 150mVpp sine wave (0- 1kHz) and convert that to a voltage. The voltage was sent to a PIC 18F4685 mcu that ran on a Northmicro101 proto board and through some embedded code would take the anaolg voltage and convert it to a 10bit value. This value was translated back to a frequency value and displayed on the LCD.

It wasn't meant to be a bullet proof set-up, more about learning and implimenting the concepts learned through the term. However what we found was that the frequency on the LCD would creep upwards 1 or 2 Hz after about 15-20 minutes and would keep on doing a very slow creep the longer we ran the circuit.

My question is what would be the most likely cause for this creep and how would we eliminate it in a future version?

The F2V circuit was tied to the NM101 protoboards +9V power supply and the output of the F2V circuit was set for +5V output max for PIC input pin protection.

enter image description here

Best Answer

I suspect a big part of the drift you're seeing is related to this equation in the LM2907 datasheet:

enter image description here

This shows that the LM2907 output voltage will drift linearly if your Vcc drifts. If you are not using an extremely good regulator to power the LM2907, this could easily account for 0.2% of drift.

If you are using a precision reference circuit for your ADC (or a bandgap reference internal to the uC), you might try switching to using Vcc as the reference. With a Vcc-referenced ADC any affect of Vcc drift on the LM2907 output will be compensated by the drift in the ADC operation.

Edit: I remove the prior advice because I see you are powering the LM2907 from 9 V, while the PIC is powered from 5 V; therefore, you can't expect the two power supplies to drift equally.

Related Topic