Transformer noisy and efficiency

Hello junchan1122,

The DC input rail is directly rectified from the AC input or from a power factor correction circuit? The input voltage ripple looks quite normal if you have a bridge rectifier and some EMI filters on the input side. The noise you here from the transformer it is not related to this voltage ripple, except you will see a lower ripple amplitude if you improve the efficiency.
The transformer audio noise is not the cause of low efficiency. Actually, it is the other way around. I suspect something really wrong with the reset circuit or the transformer design to be the reason for low efficiency and audio noise.
One first validation test will be to measure the transformer inductance (primary side) and the leakage. If the inductance value is a lot lower than you expected maybe the core somehow cracked. If the leakage inductance is way too high, I assume the transformer construction is miserable and you have to validate the person who built the transformer actually cared for the job is doing.
Another very usual failure is in the reset circuit. Make sure the reset diode has no reverse current leakage and the resistors around are not destroyed. At least for a validation test use overrated resistors (1W or 4W) and fast, very high voltage diodes (at least 1kV). After you isolate and fix the problem you can go back to low power resistors and the usual diode.
Also, make sure you use the right output rectification diode on the secondary side. You must measure the reverse voltage (directly on the diode, with an oscilloscope) and use schottky diodes rated for that voltage. Only after you push the efficiency around 70% - 80% for flyback or 80% - 90% for forward topology you can try to replace schottky with some fast diodes for rectification, but that is coming with a penalty on efficiency.
The 50% efficiency shows with no doubt something really wrong and the points discussed here are most of the time enough to isolate the problem.

Cheers,
PI_Crusher

Thanks a lot for your prompt response and useful informations.

The schematic diagram and some drain current waveforms are attached (fig.8 - 11). Would you please give me some comments. Except the high switching spikes during the switch on and off, I just found the drain current is quite irregular when the +/- 15V O/P are loaded with 3A (fig.8 and 9). The waveform seems distroyed, I don't know why this happened.

For the reset circuit, the clamping zener, blocking doide and resistor are used as recommanded from the design report. Indeed, I used 6W for the clamping resistor instead. I woundered why the clamping circuit seemed not working properly. And, the O/P didoes are already been schottky type.

For the DC rail, actually it was quite stable up to about 175VAC I/P which could be seen from the scope proir to the chip started to switching. When I disconnected the feedback loop from the O/P, I could hear the switching sound inside the transformer when the chip trying to start and restart the unit. It seemed a high magnetic energy existed insdie the air gap and generating the sound when the switch was closed. Since the whole PS was constructed by me the first time. I have a lot of problems have to cope with.

According to the design report, the effective core gap is 0.432mm. Therefore, I used 0.215 gapping paper on both halves to form the core. I am afraid the core gap might not be setup correctly although I think I did. Should I increase the core gap further a bit to reduce the peameability? Would this be helpful to increase the gapped core effective inductance and the primary inductance?

On the other hand, I suspect the chip had been running with Dmax once it was powered up and created such a high switching energy brust. Will there be anything else could lead to this?

When I connected a stable 5.8VDC from an external bench PS, the unit did not start up. The chip seems so intelligent to know that. I will try again feeding it from the O/P instead. Besides, it was also odd that the pre-load resistor RPL from the circuit diagram seems playing an important role from the circuit. As when I removed this resistor, the PS unit went out of regulation even I reduced the bias resistor to 120 ohm to the opto-diode.

Anyway, I will try to measure the primary inductance as shown in appTV.

Thank again and best regards.

Jun.

You must mesure the inductance of the transformer and make absolutelly sure you stay between 10% tolerance (use the software to calculate the nominal value). Anything else is not acceptable.
Cheers,
PI_Crusher

Hi PI-Experts,

Thanks for the advice.

Unfortunately my RLC meter had been broken a long time ago. According to the drain voltage/current estimation method, I found that it was so difficult to obtain the precise ramp current from the oscillating shape waveform (fig. 11 and 12). After a rough estimation, the priminary inductance is only about 100uH, which is well below the 261uH from the design report.

Therefore, I adjusted the VOR from 100V to 135V from the software. This lead to the turn ratio changed from 6.3 to 8.8 that is the primary turns increased from 32T to 52T. However, after the transformer was re-constructed. The oscillations during the switching still occurred. And also, the transformer became more hot than b4. Attached Fig.13 and 14 are the drain current waveform for the 52T primary. Would you please give me some comments about the shape of these waveform especially the damping shape and why the current did not ramp up linearly when the switch was closed?

Besides, I also changed to use ultra fast schottky diodes for the secondary but the effect was similar.

Would the stray capacitance from the copper traces or the fransformer itself creating the damping shape waveform? Please advise.

Thank you very mcuh.

Brgds.,
Jun.

The current waveforms 13 and 14 are too unusual to discuss about any possible “tune-up”. Actually, I am not sure I see any drain current slope there. Check the transformer polarity (beginning and end of the inductors with the same direction of winding) and diodes polarity, something must be really wrong.
Also, there is no going around transformer inductance measurement; you need to know this parameter. You do not adjust the inductance by playing with VOR; you have to incrementally grind the core (the middle part of a single half) to adjust the inductance within 10% of the requested value. In your case there is no current slope you can use as reference since your circuit is not actually working. For inductance measurement you need the transformer to be completed, with the assembly very tight.
A precise inductance measurement is possible using a resonant circuit with your transformer winding, a resistor, and an additional capacitor (serial or parallel resonant circuit). You need a sine-wave signal generator to sweep the frequency and identify the resonant frequency. Since you know the value of the capacitor and the resonant frequency, you can calculate the inductance.
Cheers,
PI_Crusher

Thanks PI-Expert,,

The primary inductance was measured to be very very low, having only about 18uH for 43T which was splitted into 2 sections. The measuring circuitry was attached. (fig.16) I don't know what's wrong with this windings as I have wound exactly 43 turns on the bobbin with the same polority and direction. Besides, the resonant frequency was also strange that the corresponding Xl and Xc were differed a lot. I just woundered would the windings' capacitance altered the mesaurement!

The drain current waveform seems to be more "normal" with this 43T transformer (fig15). Could you please give me some comments?

1.Clamping circuit
The RCD clamp seemed to be in-adequated. Could you please re-post the pdf file "PI-DG-101_ClampSizing.pdf" such that I can re-design the clamp circuit.
The previous one in another post was corrupted.

2.Feedback loop oscillating/unstable
The racing behaviour of the feedback loop was confirmed the cause of the audiable nosie from the transformer. Therefore, I changed the gain resistor from 270ohm to 510ohm as in figure 12 of AN-14 but the improvement was very little. Would there be other parameters, i.e. phase/gain margins I can adjust to make the feedback loop more stable?

3.Primary leakage inductance seems too large
The primary current still continued to flow (decaying) even the switch was opened. Is this related to the windings's coupling.
And, any different between using gapping insulator and grinding the center leg.

Thank you very much.

Brgds.
Jun.

Dear PI-Experts,

Most of the problems of my +/- 15V 120W PS had been solved after I grinded the center leg of the core within 10% of the deisgn report and replaced the clamping capacitor with the film type capacitor.

The PS works well with excellent load regulation. (within 1% with 3.5A loading) The efficiency is measured to be around 80% that had been sufficient enough for me. Although there is still a little flyback sound which I think is due to the high AC flux in the air gap, I think it can be reduced anyway.

The last problem hopefully is the high voltage spike at the output which I think is due to the high leakage inductance of the windings. I did try a lot winding layout, sequence and no. of turn....etc in order to increase the coupling of the windings and mininize the stray capacitance as well but the effect is flustrating.

Finally, I added a very small value capacitor between the drain/source of the TOP258EN and also at the both output rectifiers, the output voltage spike could be reduced from 6.3Vp-p to 1.4Vp-p as shown in fig.11 and fig.12. I have some questions hoping the PI-Experts could give me some comments.

1. Is there any other factors creating such a large leakage inductance? Should I use copper foil to increase the magnetic coupling?

2. Would there any side effect by adding a capacitor between the Drain/Source of the TOP258EN and the output diodes?

3. What can I do to mininize the output voltage spike further?

Thank in advance.

Brgds.

Jun.

Follow the ripple measurement example from page 32:
http://www.powerint.com/sites/default/files/PDFFiles/rdr158.pdf

Cheers,
PI_Crusher

Hi PI_Crusher,

The o/p ripple was measured to be 50mV without load and 0.2V with 3.5A loading.
I think this is within normal range. The voltage spike I mean is the capacitive discharge at the drain node when the switch is turning off. This is observed prior to the oscillation by the primary leakage inductance and stray capacitance.

The primary leakage inductance is bout 3.5uH while the stray capacitance is about 73pF. This figures I think should be normal. I just wounder the PCB layout will create such a voltage spike at the o/p. Anyway, I will try another PCB layout to see the effect.

Besides, in order to minimize the o/p noise. I connected a 0.1uF capacitor to both the secondary return paths to the chasis ground. Is this connection acceptable?

Thank you very much.

Brgds.
Jun.

Dear PI-Crusher,

I just found the audio noise from the transformer might due to the pluse bunching effect in the feedback loop. The PS-University showed that a feedback loop accelerator may help to fix this problem. However, it seems the circuit shown in the video is for Tiny Switch PSonly . Could you help me how to modify the connection to use in the TOPSwitch series.

Thank you very much.

Brgds.
Jun.