Can somebody please explain me how this zvs driver works? / General Science and Electronics / Forums

Forums

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

« Previous topic | Next topic »

Can somebody please explain me how this zvs driver works?

1 2 3 next

Move Thread

LAN_403

Electorials

Mon Jun 25 2012, 07:29PM

Registered Member #5457 Joined: Mon Jun 25 2012, 06:42PM
Location:
Posts: 14

Hello,

I've been looking everywhere to find out how this circuit manages to switch the mosfets when the voltage across them is 0V.

This is the circuit:

It is designed by Vladimiro Mazzilli.

All I can make up from the schematic is that when one mosfet turns on, it brings the gate of the other mosfet to ground to turn it off.
I don't see how this will make them switch at 0v?

If you don't wish to explain how it works, it's also just nice for me if you can post me a link to where the way this circuit works, is explained.

Thanks in advance,
Electorials

Ben Solon

Mon Jun 25 2012, 08:47PM

Registered Member #3900 Joined: Thu May 19 2011, 08:28PM
Location:
Posts: 600

It doesn't get much more simple than you describe. The gates are linked to the terminals of the capacitor. The capacitor and the coil resonate, so when one end of the cap is high, the other is low. The low end pulls the charge out of the gate of the opposite MOSFET, so the circuit oscillates in tune with he resonant frequency of the lc components.

Electorials

Mon Jun 25 2012, 09:53PM

Registered Member #5457 Joined: Mon Jun 25 2012, 06:42PM
Location:
Posts: 14

Thank you for the reply!

But I still don't understand something:
If you remove the capacitor and coils from the circuit,
then I think I can say that the operating frequency will be dependent on the turn on and off time of the mosfets
(since the one mosfet turns off when the other has turned on).

So why would adding a capacitor and coil to it, change that frequency to suddenly operate at the resonance frequency?
I mean, the two wires with diodes are still connected to the same point, right after the mosfets
so those connections will still shut down the mosfets as soon as the other mosfet turns on.
(I don't see why the frequency would suddenly depend on the newly connected components instead of just staying the same)

And also:
The lower mosfet from the circuit will turn Off when the voltage across the Upper mosfet is 0.
you say that if the one capacitor side (top side) is 0v there will be a higher voltage at the other side.
This all means that the lower mosfet turns off while the voltage across it is not zero

and that is not 0v switching :( so what's wrong with what I said here?

Ash Small

Mon Jun 25 2012, 10:43PM

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

I think the point is that they 'are' connected to the LC tank circuit.

They possibly would resonate at the 'turn on/turn off' frequency of the 'FETs, but the tank circuit prevents this, and the 'delay' corresponds to the 'resonant frequency' of the LC circuit, thus you get 'zero voltage switching'.

Maybe you should google 'resonant LC tank circuits'? (or something similar)

(I'm also still trying to fully understand this circuit, and I'm looking forward to replies from others)

Ben Solon

Tue Jun 26 2012, 12:17AM

Registered Member #3900 Joined: Thu May 19 2011, 08:28PM
Location:
Posts: 600

They won't resonate when there is no lc circuit, because then there would be no turn off signal to pull the gates low.

To step through each event when the circuit begins:
The gate power supply charges both mosfets through the gate resistors. Due to small asymmetries, one MOSFET turns on faster, dominating the other. This MOSFET pulls power through half of the coil, charging the cap. This cap starts "ringing" to use the resonant term. When this ringing signal approximates a decaying sine wave if left alone. But due to the diodes, when this sine wave dips below the gate power supply voltage, it pulls the gate voltage down with it. Because of this pull down, the gate resistor on this fet is unable to recharge the gate. But the other one is able to charge through the resistor because the diode on that fet is reverse biased. This fet turns on and pulls be current in the lc circuit the other direction, adding to the already oscillating lc circuit. And the cycle repeates...

Hth

Electorials

Tue Jun 26 2012, 07:05AM

Registered Member #5457 Joined: Mon Jun 25 2012, 06:42PM
Location:
Posts: 14

ben123324 wrote ...

..
To step through each event when the circuit begins:

That is just exactly what I needed!
Thanks a lot for explaining!

So as Ash Small suggested, I looked a little up about the LC tank to know how exactly this works
and when knowing that, I was also thinking that it would work this way as you now described.

So basically it's the LC circuit that does all the work, and because the oscillation would die out,
the mosfets keep on powering the oscillation at the correct timing to pump extra power into it?

And the 0v switching will be the gate votlage, so if the gate switches at 6v, the mosfet will switch when it has 6v across it right?
(mosfets will switch on when the capacitor voltage rises above 6v, and will switch off when it drops below 6v)

Now, are the thick 2W resistors necessary if I let the control circuit work at a fixed 12v supply and use a separate supply for the load?
or will I be able to use smaller resistors there if I use separate supplies?

Ash Small wrote ...

I think the point is that they 'are' connected to the LC tank circuit.

It now indeed seems that this circuit won't work without the LC tank.
Thanks for the reply!

Ash Small

Tue Jun 26 2012, 07:59AM

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

ben123324 wrote ...

They won't resonate when there is no lc circuit, because then there would be no turn off signal to pull the gates low.

To step through each event when the circuit begins:
The gate power supply charges both mosfets through the gate resistors. Due to small asymmetries, one MOSFET turns on faster, dominating the other. This MOSFET pulls power through half of the coil, charging the cap. This cap starts "ringing" to use the resonant term. When this ringing signal approximates a decaying sine wave if left alone. But due to the diodes, when this sine wave dips below the gate power supply voltage, it pulls the gate voltage down with it. Because of this pull down, the gate resistor on this fet is unable to recharge the gate. But the other one is able to charge through the resistor because the diode on that fet is reverse biased. This fet turns on and pulls be current in the lc circuit the other direction, adding to the already oscillating lc circuit. And the cycle repeates...

Hth

I can see that the circuit won't oscillate without the LC circuit. Whichever 'FET turns on first will stay on (as often happens when the circuit doesn't operate correctly), but what frequency does the cap 'ring' at?

I assume it's not the resonant frequency of the LC circuit. (or is it?)

That's the bit I don't fully understand.

EDIT: I always assumed it switched when the voltage difference between the gate and source dropped to less than the gate 'turn on' voltage, or something, but could never work out the exact mechanism.

Steve Conner

Tue Jun 26 2012, 09:55AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

The ZVS is a temperamental circuit, operation is quite sensitive to the threshold voltages and Rds(on) of the MOSFETs.

It may help if you consider that both MOSFETs turn partly on as the tank voltage passes through the zero crossing. At the instant of the zero crossing, the MOSFETs have whatever gate voltage needed to pass half of the DC link current each, and a drain voltage the same as this, give or take a diode drop.

The LC tank circuit then has to provide the electrical "momentum" to push through this metastable state and turn one of the devices off. If the loaded Q is too low, or the Rds(on) of the FETs is too high for the current level you're trying to run, this state can become stable. In other words the circuit latches up with both devices fully saturated and dies a fiery death.

Another possibility is that it keeps oscillating, but the period with both devices on grows to take up a sizeable part of the cycle. During this period the device that's supposed to be "off" operates as a linear amplifier and bursts into VHF parasitic oscillations, with the circuit going to a fiery death shortly after.

Since Rds(on) increases with temperature, and threshold voltage decreases, the ZVS can often seem to work fine at first, but when it warms up it can go into thermal runaway and explode.

As it is a push-pull circuit, you would have to replace the LC tank with some other 3-terminal network to make it complete. You can use a transformer with a saturable core, and delete the DC link choke, to get the ferroresonant version. In this, the oscillation frequency is determined by the supply voltage and the volt-second capacity of the transformer windings.

Or with a little more tweaking, you could turn it into the Eccles-Jordan astable circuit, where the oscillation frequency is determined by a couple of RC time constants in the circuit.

I can't think of any other method that would make it work without the LC tank.

Electorials

Tue Jun 26 2012, 12:22PM

Registered Member #5457 Joined: Mon Jun 25 2012, 06:42PM
Location:
Posts: 14

Ash Small wrote ...

I assume it's not the resonant frequency of the LC circuit. (or is it?)

It is :)

Steve,
Thanks for the explanation
but I think I'm going to stick with the original circuit from the picture.

I'm going to order a coil and capacitor now. What capacitance should I order? is it the higher, the better?
but I think higher capacitance will also decrease the operating frequency
so what should I do?

Ben Solon

Tue Jun 26 2012, 01:28PM

Registered Member #3900 Joined: Thu May 19 2011, 08:28PM
Location:
Posts: 600

I assume it's not the resonant frequency of the LC circuit. (or is it?)

I said ringing because I usually refer to a circuit resonating when power is being applied at exactly the correct times. But ringing is just when you change the state of the lc circuit too fast and the current oscillates down on its own. And that is in fact the resonant frequency. I have actually used this as a method to find the resonant frequency of a tank before I had a function gen. Just tap the leads of a power source across the lc circuit while having an oscope trigger and hold. From that data you can just measure out a cycle to get the ~fres.

1 2 3 next

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