Planning:
I needed finer control of Vpp. Since this is an experimental design, I decided to make some calculated guesses to choose the right values.
To start with, I believe that putting a resistor between the emitter and ground of the feedback transistor will limit the gain of the transistor. The voltage of this resistor will rise as the current through it increases limiting the current gain of the circuit. The first guess I tried was calculating the resistor as if the transistor wasn’t there(same as fully saturated) for my max voltage (25 V).
My voltage divider is 25V * R4/(R1+R4) =1.23V. I applied a little algebra and get R4 = 0.0517R1
R4 = 0.0517*100K = 5.17K (The closest I have is a 4.7K Resistor)
Putting the values back into the original equation, I got 25*4.7/(100+4.7)= 1.12 Volts This means I can go a little higher than 25 volts. That’s okay, a little higher max voltage still allows me the full range I want.
Modifying the physical circuit:
I un-soldered the transistor from the pads it was on and re-soldered the collector to the pad it was on. I soldered a 4.7 K resistor to the GND pad the emitter was soldered to and added a jumper from the resistor to the emitter.
Testing:
I hooked up the power and started testing like last week.
| ΣΔ Duty | Filtered ΣΔ Voltage (V) | Measured Vpp (V) |
|---|---|---|
| 25 | 0.40 | 4.56 |
| 51 | 0.72 | 4.60 |
| 76 | 1.04 | 6.28 |
| 102 | 1.28 | 10.4 |
| 128 | 1.60 | 14.5 |
| 153 | 1.92 | 19.6 |
| 179 | 2.24 | 22.4 |
| 204 | 2.48 | 23.2 |
| 230 | 2.72 | 23.8 |
I charted these values in LibreOffice Calc and I could see that the curve is fairly linear from about 6 to 20 V.
Starting with 6.28 V for ΣΔ duty of 76 and ending with 19.6 with ΣΔ duty of 153, I get a range of 13.3 Volts over a Range of 77 for ΣΔ duty. I should get about 0.2V per step of ΣΔ. I am happy with these results.
Looking ahead:
I still haven’t decided whether a level shifter or a co-processor are best for this design. Since I have solved this problem, I think I am now leaning towards the level shifter(it is a less complex solution). I think I should try this design with a much higher gain transistor to see how the curve changes with different transistors.
I need to update the schematic with this change.
Feedback:
Do you think this is a good solution? Do you see any pitfalls? Do you think this design is gain dependent?