55V/1.3A TOPSwitch-JX design

Posted by: Carlos Mori on 12/09/2013

Dear Colleagues, I am experiencing an issue in my power supply design and I'd like to ask you some help.Schematic and transformer diagram are attached and the circuit was completly designed using PI Expert 9.The input is universal (85V - 265V / 60Hz) and the output is 55V / 1.3A.I set an input voltage of 120Vrms and the output voltage is 36VDC with no load. If a 100ohms load is used, the output voltage drops to 24VDC.Is there any design problem? Are the circuit and the tranformer OK?I'd like to be confident on the design so I can continue debugging on components, transformer construction, PCI assembling, etc.Thanks in advance, Carlos.

Files

SCH_CV-only_55V_1.3A.pdf (2.8 MB)

XFMR_CV-only_55V_1.3A.pdf (1.29 MB)

Comments

Dear Colleagues,

I did some measures on my circuit and the results are attached.

These curves are drain voltage. Are they correct?

Thanks in advance,

Carlos.

What I think is happening is that your feedback winding is charging C8 to the leakage inductance spike value. What you may need to do is to add a resistor in series with your bias rectifier. Add about 5 ohms and I think the situation will improve. Make sure your transformer is having a right phase polarity in each winding. Make sure your place your windings as showing in the picture. First: primary winding, second: secondary winding and last: the Bias winding. This arrangement improves the magnetic coupling from secondary winding to bias.

Thanks for your feedback!

Unfortunately, there was no change on the results introducing the 5 ohms resistor in series with the bias rectifier. It seems there is no regulation.

I did some measures using a DC electronic load and as the current increases the output voltage decreases. For a load of 100mA the output voltage was 32V. For 1.3A, output voltage was 16.2V. I measure VDrain during this process and the duty cycle changes with the current change. So it seems that it works, but with no regulation.

I really don´t know what else I can do to fix this power supply. Apparently, there isn´t a design error. The PCI assembly seems to be OK and the transformer was constructed correctly. So, I really didin´t know what to do next. Any ideas?

You may check the Bias voltage , The ripple needs to be very small and the DC voltage should agree with the equation VF/NF = VOUT/NOUT. where N mean the number of turns you have in feedback winding and output winding.

I checked the bias voltage and the results are in figures attached.

The "Vbias_ripple.jpg" figure shows the voltage at the bias winding (anode of bias diode) and there is a ripple about 1V. This voltage sweeps from 4.8V to 5.8V. It seems that 5.8V level is saturated, doesn´t it? Is this a problem?

I noticed that occurs an unexpected behavior for me on Vbias voltage as shown in figure "Vbias_fail.jpg". This behavior is periodic and occurs each 2.3 seconds. Is this a normal behavior?

The third figure named "Vcontrol_1A_load.jpg" is the voltage at TOPSwitch control pin with a load of 1A. If the load changes the duty cycle changes as well.

The last figure named "Vout_1A_load.jpg" is the output voltage at 1A load.

So the equation VF / NF = 5.9 / 2 = 2.95 and the equation VOUT / NOUT = 31.7 / 10 = 3.2. The results are coherent.

Did these measures help us?

I think you are injecting too much current into your control pin. Your transformer NS -= 2 Turns and Nb 2 Turns too.

What I would do if you need to use primary feedback is to add a zener in series with the feedback branch that feed the control pin of U1. You can see AN-32 figure 4 for appreciation how to implement this feedback