Ferrite noise - screech


I am using ETD44. With EF90 cores. Good up to 200kHz.
Frequency is between 100kHz and 200kHz (resonant system with ST Microelectronics L6599 controller).
Primary turns is 23.
Secondary turns 46. I want 350V dc.
Resonant series capacitor is 22nF.
Resonant frequency is 150kHz.
Leakage inductance measured as 50uH (micro henries)
I think the primary inductance is around 2.3mH.

It works ok with 60V on the Xformer. I think I know why it blew up yesterday.:D
I have to test it again with 320V dc. Flak jacket, safety gear...and switch it on, baby.:hot:



But with 100kHz or more wires resonate so fast even if loosened, that you can't even hear then, at least not from switching... You will hear core make noise only if it is satureted, this also means it has to be loaded, not idle

Resonant converter are a bit harder to make, since you have to know some thing as you seem to know about trafo

Good luck, post some pic will you...
Thanks Eva.
Make sense, as I was not using a power supply middle point (i.e. two caps in series with balancing resistors). One end of the xformer is connected to the transistors middle point (half-bridge) the other was to the negative rail. This applies 330V plus to one transistor. The xformer is calculated with +/- 170V.
The transistors I use now are only rated at 400V. Soooo 330V plus noise and spikes is not good for their health.

I will try again with middle point.



2005-08-14 3:01 pm

i had a similar problem with noise and then later fire and smoke and plasma bursts as i increased the current/load on a power supply i am working on building.

well after much time and transistors i discovered i have been using a ferrite toroid for the inductor. ft-240 of 77 material...

so i ordered some iron core ft-240 8 and 23 material toriod cores.

i have been on the road for a couple of months and my wife told me they arrived.

SO....I thought I would wind it with 30 wires of 26 gauge .

BUT..... I read that litz type wire is unnecessary.

SO........ one solid wire or multi-wire-enameled? for the inductor...............
The good thing with Litz wire is that it is so flexible. That is what I have been using in my xformer. It works between 100 and 200kHz.

Which by the way is now performing quite well. I can change the frequency of the oscillator without any bang!!!:D
No screech! Just a hum 50 and 100Hz picked-up from our 20kW UPS.
I'll be moving away from the thing for the next test.

I am waiting for a high voltage cap for the resonant circuit.
Unfortunately, I had to rewind the main secondary (900 and 1000V) with wrapping wire. As the old copper was letting arcs here and there.
Hopefully, the insualtion - Kynar, whatever that is - will stand the voltage. I insulated each layer, and came back to the same side for each layer.


2005-08-14 3:01 pm
kynar wire

kynar wire is only 0.004" to 0.005" insulation thickness which would only be good for 200 to 300 volts, also it is only rated for 105C to 135C sooo.....

if you wrapped the transformer in such a way that the cold end turns and the hot end turns are not close to each other and the transformer does not get tooo hot it may work
Serge66 said:

Unfortunately, I had to rewind the main secondary (900 and 1000V) with wrapping wire. As the old copper was letting arcs here and there.
Hopefully, the insualtion - Kynar, whatever that is - will stand the voltage. I insulated each layer, and came back to the same side for each layer.

Kynar will certainly not solve your problem.You have to use high voltage winding techniques. Some hints:

- Leave 3mm (or so) margin at the sides of all winding layers (from outer turn to coil former) to prevent each HV winding layer from arcing to the layers above or below. This creates a good 6mm path once mylar tape is added. You may fill the margin gaps with electrical tape hand cut to the right width.

- Separate each layer of HV winding from other layers with three layers of mylar tape or more to avoid breakdown through the tape. Avoid stretching the tape locally too much, like when there are two perpendicular magnet wires with the tape inbetween (this has been the source of most of my breakdown problems).

- Use gentle thick plastic tubing to protect from arcing the ends of every winding where they enter and leave the coil. Let the tubing extend to where the first turn begins.

- Allow for proper spacing between the HV pins of the coil former and all other pins.

That should be enough to get 1KV quite reliably.
Hi megajocke,
Yes. I have divided the bobbin in two equal parts by means of a glass fibre (epoxy board without copper) partition.
Primary on one side, secondaries on the other.
I am thinking of shelving this transformer and experiment with a smaller one giving low voltage outputs (10V).
It won't be that scary to do the testing.

I have used the AN2450 application note from ST microelectronics on the L6599 circuit.


2003-01-11 8:01 pm
Did you have any luck with the arcing?

I found this very interesting on resonant converters:

They compared different converter topologies for a 48V front end converter for computer servers or something like that. They chose a LLC converter in the end for its good efficiency at high input voltage and boost capability for hold-up.

I guess this is for a tube amp? It seems like a good topology for that Myself I am looking for something with high no-load efficiency as it is for a solid state class B thing. I'm starting to like the simplicity of a PWM half bridge actually. I'll probably use pretty low freqency and BJTs too... :smash:

I haven't totally ruled out resonant though - I'd like to read up on ZCS switching and how much this helps BJT switching losses. I wonder what the base drive is supposed to look like for optimal switching in this case. The biggest problem seems to be storage time spread and temperature dependece when trying to operate at higher freqencies. I read someting stating that storage time in an emitter switched bipolar could be made pretty low. This might be an interesting switch type if that is true.


2003-01-11 8:01 pm
ZCS isn't as favourable for MOSFETs as ZVS is so this is probably why it is so hard to find info about it. It seemed to be quite popular in the eighties though even for MOSFETs - probably because high voltage MOSFETs were quite slow back then compared with what is available today. ZCS doesn't solve the energy loss from discharging the drain-to-source capacitance which will still cause loss even if switching is really fast so ZVS is usually better for MOSFETs.

ZCS for bipolars though might be useful to reduce switching losses. But noone uses bipolars today so it's hard to find info. ;) Storage time needs to be controlled though to be able to work at high freqency without unneccesary dead time.