I built a 1.59KHz phase shift oscillator based on EMRFD Figure 7.25. This was an interesting exercise and will form the topic of a separate writeup. This generator gave me a clean signal with less than 2mV of noise as seen on the scope.
For an amplifier signal output voltage of 1volt and assuming a voltage gain of 24 for the amplifier, I calculated that I needed about 70mV input to the amplifier. After measuring the open circuit voltage of the generator and the output resistance of the generator, and knowing the calculated input resistance of the amplifier, I calculated the need for a 2.82k voltage dropping resistor between the generator and the amplifier. I inserted a 2.7k resistor which gave me a measured amplifier input voltage of 63mV. This was in the ballpark.
I then measured the DC bias characteristics and the signal input and output voltages of the amplifier. I also tried to measure the signal voltage on the base of the 2N2222, however this signal was really too small to measure accurately on my scope.
Next I measured all the resistors in the circuit using a Keithley 179 DMM. These results were compared to the FLUKE 8024B results and seen to be within a few percent of each other. I also measured the transistor beta which varied between 160 and 175 depending on the DMM used. I carried out these measurements twice. Once on friday evening and again on this saturday morning. This morning the temperature was 47deg F in my attic where I was working. Certainly colder than last night although I did not measure the temperature then. All measurements were consistent except the beta measurement. I used a value of 168 for the calculations. I did not measure the ESR values of the capacitors however.
I then used these measured component values in the simple model calculations and the LTspice simulations. The results are summarized in the table directly below.
The results are extremely interesting and insightful to me. I found that adding an ESR (equivalent series resistance) of 2 ohms into the emitter 100uF bypass capacitor, I was able to obtain close correlation between the measured result and the LTspice simulation. Especially when comparing the signal load voltage results. The voltage gain measurements shown below have less credibility due to the difficulty in measuring the base input voltage (used to calculate the voltage gain). It is also interesting to compare the difference between the simulation with and without the 2 ohm resistor. There is a difference of 806mV-724mV = 82mV. Meaning a difference of 33mW power output.
The shown results for the simple model also include a 2 ohm regeneration resistor. Certainly this brought this result closer to the simulated and measured result but there is still a significant variance. Why?
Perhaps a useful conclusion from this exercise is that manual calculations are great to predict a ballpark result, however the LTspice IV simulator can get one much closer to the measured result. This conclusion cannot be validated, of course, without the ability to do many more precise measurements. Certainly beyond the scope of my simple 1970's era equipment workbench. Or is it?
 |
| Selected Parameters that serve to illustrate the differences in model results compared to the measured results |
 |
| Bench test configuration |
 |
| The LTspice generated schematic showing actual measured component values |
 |
| A detailed spreadsheet showing all calculated and measured results |
 |
| The phase shift test oscillator is powered by a 9V battery. Shown on the right and coupled to the audio amp on the left with 1:1 test leads connected to the scope. No impact on the measurements, were seen from the test leads. |
 |
| Scope display showing the amplifier input and output waveforms. The waves 'look' like 'clean' sinewaves. |
