Singing flyback transformer.

[I_think]

Hello,

My flyback transformer is making a noise. and its loud that I can here it in the next room. This flyback transformer is a part of smps based on UC3842. with TL431 feedback. I have connected lamp as a load. It glows, but I am not able to measure its voltage, because the digital multimeter shows some random reading (The meter is ok.) Looks like a lot of radiation.

To add, this transformer is hand made. And primary wound in the sandwich pattern. winding are kept tight as possible. I kept this on for 3 hours, but it was working fine.

Summary:

Having two problems. 1) Noise from flyback transformer 2) Not able to take any reading on multimeter.

Requesting you all to guide me for solution.

 

[BradtheRad]

To reduce audible noise, it is common to operate switched transformers at frequencies above the human range of hearing.

Is it possible that the transformer is contacting something rigid which amplifies the vibrations? It may help to isolate the transformer somehow. Perhaps even to avoid using rigid wires. Cushioning it is an obvious solution but you should be careful to allow heat to radiate from it.

My DMM does not give a sensible reading with square waves.

My analog meter gives a consistent reading with square waves, but it is not correct. The meter is calibrated to read a sine wave.

 

[thylacine1975]

Yup - I second Brad's comments about operating frequency and measurements. You may have more luck with a so called "true RMS" meter, or better yet - an oscilloscope.

Audible noise from SMPS inductors/transformers operating in the frequency range of human hearing is generally unavoidable due to magnetostriction (https://en.wikipedia.org/wiki/Magnetostriction). You *might* be able to _improve_ (but not fix entirely...) things [especially if your core is gapped] by potting the transformer assembly in epoxy/varnish. It does make the modifying the core later a whole let less fun though!

 

[I_think]

Thank you, Bard and Lacine.

I think I have put insufficient information. My SMPS operates at 100Khz (97.6Khz measured). I was measuring the voltage at the output of the rectifier and filter. So its ok to expect clean DC with some ripples. But unable to read this voltage on my DMM.

Some things that I doubt are, wrong design of feedback network ( TL431, optocoupler) this enabling and disabling the UC3842. which make that audible sound.

secondly, either my transformer is radiating ,or my switching transistor is radiating so high , that my DMM is not able to take readings.

- - - Updated - - -

Some more things, wont potting the flyback transformer ( with gapped core) change its inductance?

 

[BradtheRad]

Some things that I doubt are, wrong design of feedback network ( TL431, optocoupler) this enabling and disabling the UC3842. which make that audible sound.

I guess this is the condition where either the coil or capacitor has developed excess charge on it, causing a skip in the switching function, until a cycle or two has gone by. I've seen this called 'hiccup mode'.

If it occurs every 10 cycles at 100 kHz, then it can create a noise within our hearing range.

secondly, either my transformer is radiating ,or my switching transistor is radiating so high , that my DMM is not able to take readings.

I have thought the same thing. A DMM may be influenced by rf, ambient or through its leads.

Or another factor. The DMM takes a sample every so often. Say you are reading pulsed DC (on a DC range setting). The incoming volt level can vary while the sample is taken. The reading will be unpredictable. It won't be an average. It won't be a minimum or a maximum.

Similar thing with how a DMM measures AC. I tried to measure inverter waveforms and got nonsense readings. They were squared waveforms, not sine waves.

 

[I_think]

I guess this is the condition where either the coil or capacitor has developed excess charge on it, causing a skip in the switching function, until a cycle or two has gone by. I've seen this called 'hiccup mode'.

Yes, I could see such hiccups during simulation. ( the momentary switching off of UC3842). So to check this trying to get direct feedback ( non- optoisolated)

I have thought the same thing. A DMM may be influenced by rf, ambient or through its leads.

What Can I do for this ? it means this is totally noisy.

 

[I_think]

Audible noise from SMPS inductors/transformers operating in the frequency range of human hearing is generally unavoidable due to magnetostriction (https://en.wikipedia.org/wiki/Magnetostriction). You *might* be able to _improve_ (but not fix entirely...) things [especially if your core is gapped] by potting the transformer assembly in epoxy/varnish. It does make the modifying the core later a whole let less fun though!

I wonder why dont we hear them from any commercial SMPS flyback?

- - - Updated - - -

the sound is between 5Khz to 6Khz ( measured). altering the filter and the tl431 does not make any difference.

so can I confirm that its as thylacine said ?

but still, how to clean the radiations?

 

[szhighstar]

To reduce noise for transformer:
1.vacuum varnish
2.even wound wire for all windings
3.connect core using epoxy glue
4.fix coil and core using epoxy glue

 

[I_think]

Thanks szhighstar,

I got EMI resolved. I used Y-capacitor.( still not sure how to choose the value). and the noise disappeared , but not sure why. Its only 2 hrs back i got it resolved. Haven't loaded it 100%. Will update you.

more knowledge and suggestions are welcome.