I completed the circuit design of Push Pull TL494 Inverter to Modified Sine Wave Inverter, but during simulation I saw several issues.
One of the major issues was with the output. The output was not a continuous stream of square waves but rather "bursts" of activity followed by "dead zones" where the output fell to 0V. It was Intermittent PWM Shutdown or Pulse Bursting as shown in the figure below.
I tried to find the problem for hours, checked the transformer, the snubber circuit, the diode bridge, output filtering capacitor etc. The TL494 IC (U1) was like "panicking" due to high-frequency noise and an overly sensitive feedback loop, causing it to shut down its outputs ($E1/E2$) repeatedly to protect the circuit.
But as shown in figure below, the sudden shutdown of output from TL494 had still not gone.
Its primary roles in the circuit are:
Error Amplifier Stability: It controls the "gain" and "speed" of the TL494's internal error amplifier to prevent it from reacting too violently to small changes in voltage.
Preventing Oscillation: Without this network, the controller would rapidly switch between 0% and 100% duty cycle, causing the "pulsing" or "breaks" seen in your earlier waveforms.
Filtering High-Frequency Noise: It acts as a low-pass filter so that the switching spikes from the transformer do not trick the TL494 into thinking the output voltage is higher than it actually is.
I increased their values from R16= 10kΩ to 100kΩ and C3=100nF to 1μF. With this I improved the response time which had previously "slowed down" the controller.
Calculating the values for $R_{16}$ and $C_3$ involves designing the Type II Compensation Network for the TL494's internal error amplifier. This is essentially a PI (Proportional-Integral) Controller.
The goal is to set the Mid-band Gain and the Zero Frequency to ensure the system is fast enough to regulate voltage but slow enough to ignore switching noise.
1. Determining $R_{16}$ (Proportional Gain)
$R_{16}$ works with the input resistor (which, in your feedback divider, is effectively the parallel combination of the divider resistors, but we'll simplify). The "Mid-band Gain" ($A_m$) of the error amplifier is roughly:
Why we chose 100kΩ: With $10\text{k}\Omega$, your gain was too high. Every tiny ripple on the output was being amplified by the TL494, causing it to "snap" the PWM to 0% duty cycle (the bursts/breaks).
The Result: Increasing $R_{16}$ to $100\text{k}\Omega$ lowered the "sensitivity" (gain). It tells the TL494: "Don't overreact to small changes; just provide a steady average."
2. Determining $C_3$ (The "Zero" Frequency)
$C_3$ creates a "Zero" in the frequency response. This is the point where the integrator starts to work. The formula for the zero frequency ($f_z$) is:
Calculation with your new values ($100\text{k}\Omega$ and $1\mu\text{F}$):
$$f_z = \frac{1}{2\pi \times 100,000 \times 0.000001} \approx \mathbf{1.59\text{Hz}}$$Why 1.59Hz? For a DC-DC converter, you want the feedback loop to be very slow compared to the switching frequency ($23.8\text{kHz}$). By setting the zero at $1.59\text{Hz}$, the TL494 ignores the $23.8\text{kHz}$ ripples entirely and only looks at the long-term average of the DC voltage.
3. Summary of the Transition
| Component | Old Value (10k/100nF) | New Value (100k/1μF) | Effect |
| Gain | High | Low | Stops the PWM from "panic-stopping" (fixes the gaps). |
| Zero Freq | $159\text{Hz}$ | $1.59\text{Hz}$ | Smooths out the response; the blue line becomes a flat DC. |
And so together with removal of the C16 capacitor and changes to the negative feedback compensation network R16 and C3 successfully stopped the TL494 intermittent shutdowns.
Thinking it was the negative feedback components issue, I reconnected the C16 capacitor, but the problem persisted and so all three changes were required.
You can download the proteus project file from the link below:
Download TL494 Modified Sine Wave Inverter Proteus Project/Schematic/Code
Related Useful links:
