High Voltage Planar Ferrite transformers. ( Intial Experiments ) / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

High Voltage Planar Ferrite transformers. ( Intial Experiments )

first previous 9 10 11 12 next last

Move Thread

LAN_403

Uspring

Mon Oct 20 2014, 04:11AM

Registered Member #3988 Joined: Thu Jul 07 2011, 03:25PM
Location:
Posts: 711

Electra wrote:

The narrow pulses you see, is I think a result of the value of L being small...

Precisely. When you apply a voltage to an inductance it will cause a linear rise in current. This rise will finally be limited by the series resistance. Quantitatively:
dI/dt = V/L and
Imax = V/R
The time the current will rise, which is about the width of the secondary voltage peak is then roughly
t = Imax / (dI/dt) = L/R
That is Electras equation.

As far as I can see from the scope shot, the voltage peak is maybe about 1us long. In order to obtain a reasonably faithful reproduction of the 10us input pulse, t should be at least 10 times as large either by increasing L as Sulaiman suggests or by decreasing R as Electra does.

Eventually you'll want to get rid of a current limiting resistor, so the only way of keeping the current within limits is to increase inductance. Inductance can be greatly reduced by gaps in the ferrite. You have to be extremely careful of avoiding that when glueing the pieces together. A preliminary ballpark estimate I made seems to suggest that you have a considerably lower inductance than expected considering the high u material you are using.

Patrick

Mon Oct 20 2014, 05:14AM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

Uspring wrote ...

Electra wrote:

The narrow pulses you see, is I think a result of the value of L being small...

Ah yes i see, the constant slope (nearly) triangle ive seen before.

OK so let me try reducing the resistance and decreasing the time interval to 5 uS, that should help, right? Im also using passive reset, which i gather you all figured out the long off times were for. (The whole point of this project is high power at high effceincy, so the resistance is unwanted, eventually. Im using it to avoid blowing components though.) sulaimans idea worries me as i was trying to get away from fractional turns.

here in this thread:

(typical TV flyback)

Current profile through flyback and 0.5 Ohm resistor, (Primary current), peaks at 5.8 V, so 5.8V / 0.5 Ohms = 11.6 peak amps.

Uspring

Mon Oct 20 2014, 05:56AM

Registered Member #3988 Joined: Thu Jul 07 2011, 03:25PM
Location:
Posts: 711

Yes, that should help. But, as said, you'll want to get rid of the series resistor, since it'll eat up lots of power. From your scope shot my guess at the primary inductance is from

a) Current rise speed: 0.4A/us
b) Applied voltage 13V (?)
c) L = V / (dI/dt)

That will amount to roughly 30uH inductance. If you run that with 100kHz frequency, reactance will be 2*pi*f*L or about 20 ohms. Assuming a 100V peak to peak or 50V peak input, the magnetizing current will be 2.5A peak. This is with an unloaded secondary. As soon as you load the secondary the current will increase. If you finally want to work with 50A peak (i.e. with loaded secondary), the amount of magnetizing current alone looks reasonable.

You have a rapid drop of current in the primary. You must be generating a huge voltage peak somewhere.

Patrick

Mon Oct 20 2014, 06:07AM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

Uspring wrote ...

You must be generating a huge voltage peak somewhere.

Im hoping its at the 3 turn secondary, i was getting 63 volts or so across a 680 ohm resistor. I disabled the gate and source-drain TVS, to prevent "missing voltage" but the internal body zener is beyond reach.

Ash Small

Mon Oct 20 2014, 06:21AM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

Patrick wrote ...

sulaimans idea worries me as i was trying to get away from fractional turns.

I'm not sure it is 'fractional turns'. It passes through both winding windows. For example, with a toroidal core, the wire only needs to pass through the toroid to be a 'full turn'.

What I'd do is move the MOSFET and the positive feed as close as possible to the window itself. In the circuit I'm working on (which operates in 'flyback mode') I'll have a capacitor between the drain on the MOSFET and the primary. (EDITED, I made a mistake. I think it's correct now, but I've been up all night)

I'm not sure if all this is applicable to your circuit (I'm planning on having three or four MOSFET's per primary turn, and a solid aluminium primary, but I'm using toroidal cores). My primary pulse length will be around 100nS, I hope.

Patrick

Mon Oct 20 2014, 06:31AM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

Im fine with redesigning the layout, given the great progress so far. in fact ill get rid of the gate driver IC and 555. the next version will have a single UC28023DW from TI.

Patrick

Tue Oct 21 2014, 02:48AM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

I bought some common wire wound resistors, 0.47 and 1 ohm, +- 10%.

After "liberating" the wire from the mineral wool, it measures 5.6 inches for 0.47 ohms.

Im wanting 0.15 ohms. So, 0.47 / 5.6 = 0.084 ohms per inch. 0.15 / 0.084 = 1.79 inchs.

Primary voltage between Drain and lower transformer terminal, not sure why its mysteriously higher than 13.5 Vin. These two primary and secvondary waveforms do look like each other, so that's an improvement.

Seen here the new oscillogram, 5 uS and 0.15 ohms in series. reaches 113 V on the secondary. It absolutely squeals when it runs. when I tried this at 10uS and 0.15 ohms, the resistor started to glow bright nuclear orange. repeating with new resistor at 5uS did not reveal much heat, the mosfet and copper were noticeable warm, but I don't know if that heat came from the internal diode or the resistor.

Resistor that went incandescent is outlined in blue with arrow.

im not real sure what the oscillogram means though, it looks like a more defined version of the previous posts. so I assume the previous points you all have made are right, the t=L/r thing is dominating here. I just don't see why the shape is a spike. or why there is a square-ish dip before the spike. Whats exciting is the secondary dip and spike are proportional to the 1:3 turns rate at the primary, as it should be.

A thought for this primary side pic, if I take more power out of the secondary, will the sharp 26 volt peak start to diminish?

Patrick

Wed Oct 22 2014, 03:05AM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

no added resistor, just the primary and MOSFET. it ran just fine for a while, then blew the MOSFET while making a bad measurement.
I don't know why that dip is in front. it hurts my brain.

1413951972 2431 FT1630 Flyback Converter Pss Kneepoint Trace

Im wondering if its mode is close to that of a flybaack.

EDIT: Im wandering if there should be a diode to prevent the reverse voltage. a diode above the transistor but below the coil.

STTH30L06G-TR

Shrad

Wed Oct 22 2014, 10:32AM

Registered Member #3215 Joined: Sun Sept 19 2010, 08:42PM
Location:
Posts: 780

that's just a bet, but what about your three cores arranged 120Â° apart at one side of the PCB and the three mosfets the other side, with the secondary in the same form as a triskel?

Uspring

Wed Oct 22 2014, 10:53AM

Registered Member #3988 Joined: Thu Jul 07 2011, 03:25PM
Location:
Posts: 711

If that's secondary voltage you are looking at, primary voltage probably looks quite like it.

When you apply a positive voltage pulse to the primary, the current will rise until the pulse ends. Then the current will stay on and drive your input negative. If you don't clamp it, voltage will go to a value where something gives and will clamp it.

The secondary will just reflect this behaviour, i.e. first a positive pulse with the width of the input pulse and then a negative one.

first previous 9 10 11 12 next last

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob