Details determine success or failure Homemade 150W short arc xenon lamp

150W short arc xenon lamp? By the way, this article is about how to make this circuit!

First of all, the basic requirements of the lamp are constant current and the no-load voltage is as high as possible. Considering that the power is relatively large, the flyback circuit may not be able to handle it and the volume will also be large, so a single-tube forward circuit was chosen.

Secondly, only the current loop is considered in the circuit. The voltage is open-loop, so the no-load voltage is equal to the input voltage divided by the turns ratio, and has nothing to do with the duty cycle. Taking into account the influence of leakage inductance spikes, the actual measurement input is 234VAC and the output is 100V DC with no load.

In addition, in order to ensure the safety of capacitor voltage when the mains power is high, a 160V capacitor is selected so that the voltage is sufficient. The frequency is compromised to choose 50KHz, the switching loss is not too large, and the magnetic core does not need to be very large to output power.

Below is the final circuit diagram. The parameters are accurate, and the components with question marks are not actually installed.

At first glance, this circuit seems to have nothing special. However, details determine the success or failure of the entire production. The following discusses the problems and solutions in design and production:

1. Does the auxiliary winding adopt the forward or flyback form?

When the auxiliary winding adopts forward pulse, peak rectification is generally used. In this way, as long as the duty cycle is greater than 0, the auxiliary power supply voltage is always in a turns ratio relationship with the DC high voltage of the previous stage.

When the auxiliary winding adopts flyback, the voltage changes greatly with the duty cycle and load, and there is a possibility of failure to start.

Given that the input voltage here is 180-260V, the auxiliary voltage change is 13-20V, which is acceptable for both IC and MOS. The actual result is also quite consistent. However, it is a little higher than expected. Although it also changes with load changes, the change is very small and basically does not affect the work of competing for production.

PS: If a solution with APFC is used, it is strongly recommended that the forward auxiliary power supply voltage should be more stable.

2. How to reset

The forward transformer has no reset capability and needs to be reset passively to work properly.

Common solutions include reset winding reset, RCD reset, LCD reset, active clamp reset, and resonant reset.

Winding resetting will increase the complexity of the transformer and put forward higher requirements on the transformer's withstand voltage, and the duty cycle cannot be greater than 50;

RCD reset is relatively simple, the duty cycle can be greater than 50, and the switch tube voltage stress is relatively low, but all the excitation energy and leakage inductance energy are consumed by the resistor, and the efficiency will be slightly worse;

LCD reset is slightly better than RCD, and can achieve basically lossless absorption and return energy to high-voltage capacitors, but there are relatively few articles introducing it, so I haven't been able to understand it in depth;

Active clamping requires a dedicated IC, which can achieve the highest efficiency and a duty cycle greater than 50, but it increases cost and complexity;

Resonant reset increases the voltage or current stress of the switch tube and is not considered.

After comprehensive consideration, RCD reset was finally chosen, but the duty cycle was not designed to be greater than 50.

3. Should the transformer be breathed in?

The forward transformer theoretically does not require energy storage, so theoretically no breath is required.

At the beginning, there was no breath open loop test. There was a sound of the core closing when the power was turned on, but it ran very well with a 300W load and there was no sound.

Then the tragedy began with the closed loop. After closing the loop, it was found that the current was constant and the accuracy was sufficient, but the transformer was still making noises and generating heat.

Using an oscilloscope to look at the waveform, I found that it was working intermittently, so I thought of the feedback loop problem (see questions 4 and 5 below). After the repair, the duty cycle became continuous, but there was still jitter, and the transformer still heated up and made noises.

Is the transformer saturated? Looking at the waveform on the sampling resistor confirmed my point of view. In the messy waveform, I can vaguely see that the winding current rises sharply after some cycles, which is clearly a sign of saturation.

So, based on my previous experience and intuition, I tore off a piece of paper, which was 0.08mm thick, and put one on each side of the magnetic core to add some air. At the same time, I changed C109 from electrolytic 400V4.7uF to CBB400V0.1uF as shown in the figure. As a result, the beeping stopped and the problem of the transformer overheating was solved.

Later, I guessed that the primary inductance of the transformer was 46mH, which caused the reset current to be too small. Adding air could reduce the primary inductance and reduce the residual magnetism. Although the reset resistor was a little hotter than before, the key point was that the transformer could work reliably!

It can be seen that increasing the breath has a positive effect on improving the anti-saturation ability of the transformer.

4. Misunderstandings about UC384x and current mode

The third pin of UC384x is the current feedback pin. We all know that current feedback can be used for protection, but in fact this pin plays a more important role, which is PWM modulation. Its role is similar to the sawtooth wave from the oscillator in voltage mode PWM, but this point is ignored by many people.

I often see people asking why the duty cycle of UC384x is always the largest after pin 3 is grounded, and pins 1 and 2 cannot control the duty cycle at all. Now you should understand; after pin 3 is grounded, the output of the error amplifier controlled by pins 1 and 2 is always greater than the voltage of pin 3, and there is no opportunity to reduce the duty cycle.

Therefore, the waveform of pin 3 of UC384x is related to whether the entire power supply can work normally.

5. What does slope compensation compensate for? Why is slope compensation necessary? When is compensation needed?

I guess many beginners have heard of this, which is the function of the capacitor between pins 3 and 4, but they only have a vague understanding of this term.

Slope compensation compensates for the voltage change slope of the third foot of UC384x.

Why and when is compensation needed?

When the slope of pin 3 is insufficient, compensation is required. If the slope of pin 3 is too small, a slight change in the feedback voltage will result in a very large adjustment, causing fan failure.

Normally, DCM flyback does not require compensation because the voltage slope at pin 3 is proportional to the primary current slope and should be relatively large.

CCM flyback and forward (single/double) and even CCM boost are needed. Because the primary current changes little when the tube is turned on, although the causes are different, the manifestations are similar, resulting in a small change slope at pin 3, which is prone to convulsions. At this time, slope compensation is needed. For forward, slope compensation is a must!