Sinking GDT waveform / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Sinking GDT waveform

previous 1 2

Move Thread

LAN_403

Weston

Sun May 17 2009, 06:50AM

Registered Member #1316 Joined: Thu Feb 14 2008, 03:35AM
Location: Cambridge, MA
Posts: 365

Here is a pic of running the GDT at 80 kHz ish.

The data sheet says that the GDT is good down to 10 KHz though.....

...

Sun May 17 2009, 08:06AM

Registered Member #56 Joined: Thu Feb 09 2006, 05:02AM
Location: Southern Califorina, USA
Posts: 2445

What is the p/n of the core you are using? It really does have too few turns for what you are using it for, I am guessing that the 10KHz number was probably quoted for a lower voltage. Can you run the bridge at a higher frequency?

GeordieBoy

Sun May 17 2009, 01:44PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

Weston, that first waveform you showed with the steps half way along the peaks and troughs is normal when bipolars are used for GDT drive.

What is happening is that the triangular mangetising current of the GDT primary is commutating between the switches and diodes in the GDT's drive circuitry.

Draw out your H-bridge of transistors with its free-wheel diodes that makes up the GDT circuit. Then draw out the triangular current waveform that you would expect the GDT primary to draw in the presence of a square wave of applied voltage. Finally work out where the current flows in your schematic at each point in time. You will see that the little steps in the tops and bottoms of the voltage waveform are caused when the magnetising current passes through zero and switches from diodes to transistors.

When the current is flowing through the diodes the voltage across the GDT is equal to the DC supply voltage plus two diode drops. When the current switches to the transistors, the voltage across the GDT is now equal to the DC supply voltage minus two times Vce. The size of the voltage step should be equal to 2x(Vd+Vce) This is usually about a couple of volts and posses no problem for MOSFETs provided it occurs deep in their "turned-on" region.

If you are seeing a larger voltage step, then you need to check your free-wheel diodes and also check the bipolar transistors aren't damaged.

As a side note you won't see this behaviour when a bridge of MOSFETs is used for GDT drive because they are able to support conduction in both directions when enhanced.

I hope this helps,

-Richie,

syntroniks

Sun May 17 2009, 03:28PM

Registered Member #1530 Joined: Tue Jun 10 2008, 03:34PM
Location:
Posts: 32

I can send you a good GDT core if you'd like. I have a coilcraft "Gate drive transformer" here next to me and it doesn't really perform for this application. I haven't used this other one yet, but it has rather high permeability. 3E6 material from Ferroxcube. Let me know if interested. Not sure how it performs in low frequency (I work an order of magnitude higher freq).

Check GeordieBoy's comment. Also, are you designing this SMPS yourself or going off an existing design? Either way, rock on and be safe!

Weston

Sun May 17 2009, 05:23PM

Registered Member #1316 Joined: Thu Feb 14 2008, 03:35AM
Location: Cambridge, MA
Posts: 365

Oh! I forgot the free-wheel diodes

. They are across emitter and the collector but reversed biased during normal operation, right?

GeordieBoy

Sun May 17 2009, 06:05PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

> Oh! I forgot the free-wheel diodes.

Ah, that will do it!!! I wondered if you might have, but wanted to give you the benfit of the doubt.

Without the free-wheel diodes (or clamp diodes) across the biploar transistors, the transistors will probably undergo reverse breakdown as the GDT primary inductance tries to force current through them in the wrong direction!!! This is likely the cause of the particularly large steps in your waveform.

Yes, the free-wheel diodes go across the bipolar transistors between C and E so that they are normally reverse biased. On a schematic with +V bus at the top, all clamp diodes point upwards. Schottky diodes are the best choice as the voltage and current requirements are low. UF4001 will suffice if you have some handy though. 1N4148/914 may struggle with the current unless the magnetising inductance is large for 20kHz,

-Richie,

Weston

Sun May 17 2009, 10:54PM

Registered Member #1316 Joined: Thu Feb 14 2008, 03:35AM
Location: Cambridge, MA
Posts: 365

I tried using UF4007 diodes as clamp diodes, still no improvement of the waveform.

. Any more suggestions?

Firefox

Mon May 18 2009, 04:38AM

Registered Member #1389 Joined: Thu Mar 13 2008, 12:50AM
Location: Pittsburgh, PA
Posts: 346

Here is the simulator waveform you asked for. I used voltage sources for the drive signals and circled the points where each wave is scoped from. Enjoy!

1242621341 1389 FT69582 2n3904 Waveform Sim Jpg

previous 1 2

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