Circuit Maker Simulation of Modified "Mazzilli" Driver / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

Circuit Maker Simulation of Modified "Mazzilli" Driver

previous 1 2 3 next

Move Thread

LAN_403

jpsmith123

Tue Dec 06 2016, 02:59PM

Registered Member #1321 Joined: Sat Feb 16 2008, 03:22AM
Location:
Posts: 843

Probe at drain of Q2:

Edit: I forgot to say, now that the amazing fact that it's oscillating in the first place - has worn off - it doesn't look right to me either. Hopefully later today or tomorrow I'll get a chance to try it with another simulator and see what happens.

Sulaiman

Tue Dec 06 2016, 05:51PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

O.K. ... now I'm confused because the drain waceform is what I'd expect,
maybe the 100 nF is resonating with the transformer leakge inductance ?
(in parallel with 2x 100 uH)

jpsmith123

Wed Dec 07 2016, 05:41AM

Registered Member #1321 Joined: Sat Feb 16 2008, 03:22AM
Location:
Posts: 843

That's what I was thinking also.

But since CircuitMaker seems generally "messed up" in ways I don't understand, I'm going to force myself to stop wasting time with it and get more familiar with other simulators such as Pspice.

Right now I'm experimenting a little bit with LTspice. Hopefully by tomorrow I will be able to compare its result to that of CircuitMaker.

(Basically what I want to do is to run one of my big ferrite HV transformers with a ZVS driver - preferably powered directly from the 120 volt wall socket. So I'm trying to get a good simulation of the basic Mazzilli circuit as a baseline, and then possibly modify it, if need be.

Sulaiman

Wed Dec 07 2016, 03:30PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

this type of zvs topology has many limitations;
if you want a continuous fixed secondary voltage (e.g. no short-circuits or capacitor charging) ok
what sort of things do you expect to do with the output ?

jpsmith123

Thu Dec 08 2016, 02:27AM

Registered Member #1321 Joined: Sat Feb 16 2008, 03:22AM
Location:
Posts: 843

I was seriously considering using a Mazzilli ZVS driver to drive a ferrite transformer (@25 to 30 kHz and 20 to 22 volts/turn) followed by a voltage multiplier - as a 50 kv CCPS.

The simplicity of the circuit is an attractive feature. Also, some people e.g., on youtube, are claiming to have successfully operated them from the 240V mains.

I was thinking that if I can get good simulations, I can experiment with different gate drive modifications to the circuit.

Also, if I try to run a ZVS driver and ferrite transformer from a DC bus voltage of 330 volts or something like that, I may need to gap the core to get a higher Q, right? Simulations might shed light on this too.

OTOH with a half-bridge SLR driver for example I wouldn't have that complication. So maybe I should forget the ZVS driver for this purpose?

hen918

Thu Dec 08 2016, 01:30PM

Registered Member #11591 Joined: Wed Mar 20 2013, 08:20PM
Location: UK
Posts: 556

The problem with high voltage Mazzillis is the much higher current flowing. It causes problems with the gate diodes not switching quick enough, and causing a FET to remain on, and explode. The is also a higher inductance on the primary required, (and I'm talking much higher) and higher voltage FETs, or, better, IGBTs, required (at least 1200 V). At 3.14*340 V (1070 V) isolation becomes important on the primary. High voltage FETs or IGBTs and high currents = high losses.

An interesting Mazzilli equivalent for high voltages is John's Induction Heater circuit:

Even this is temperamental.

There are reasons there aren't many high voltage Mazzillis!

I personally avoid them, I never get even low voltage induction heaters / wireless chargers to work properly with a low coil inductance, although all my flyback drivers have worked fine, even charging electrolytics.

jpsmith123

Thu Dec 08 2016, 02:34PM

Registered Member #1321 Joined: Sat Feb 16 2008, 03:22AM
Location:
Posts: 843

I could be wrong, but I don't necessarily see those issues as fundamentally limiting it to low voltage operation, but at some point it may
be more of a hassle than it's worth, IMO.

BTW have you seen this video on youtube?

hen918

Thu Dec 08 2016, 05:05PM

Registered Member #11591 Joined: Wed Mar 20 2013, 08:20PM
Location: UK
Posts: 556

jpsmith123 wrote ...

I could be wrong, but I don't necessarily see those issues as fundamentally limiting it to low voltage operation, but at some point it may
be more of a hassle than it's worth, IMO.

BTW have you seen this video on youtube?

That's not a normal Mazzilli:

But even so, impressive.

jpsmith123

Thu Dec 08 2016, 09:54PM

Registered Member #1321 Joined: Sat Feb 16 2008, 03:22AM
Location:
Posts: 843

hen918 wrote ...

That's not a normal Mazzilli:

But even so, impressive.

Well as I see it, it's a Mazzilli driver with a modified gate drive circuit - which may or may not be necessary. (It seems no two "Mazzilli driver" circuits are exactly the same these days, so they seem to be collectively known as generic "ZVS drivers").

Anyway, IMO, the point of going to higher DC supply voltage - for a given power throughput - is to reduce the primary current. Of course this requires more turns on the primary, which will increase the inductance and lower the Q.

At the bottom of that schematic diagram you linked to there's a statement that the Q must be several times > 1 even when under heavy load (arcing). What happens if it's not? (It would be nice to have more detailed information about his circuit, especially the transformer and capacitor).

With the big TSC ferrite core transformer I have ("50all" ferrite material) I get something like 7800 nH/(turn)^2, so for a primary running at a drive level of 20 to 22 volts per turn with a 330 volt DC supply voltage, my primary inductance will be somewhere close to 2 mH. So what I want to know is: How do I get a Q > 1 under load with so much inductance?

This is one of the things I was hoping to explore with simulation, i.e., to see how the circuit behaves under various load conditions vs. Q.

###########################################

E dit: I keep forgetting that the primary voltage on the Mazzilli driver is 3.33 (or something like that) * DC supply voltage, so apparently I would need many more turns than I mentioned above. Anyway, I'm going to forget about it and take Suliaman's hint and just go with a SG3525 based half-bridge SLR circuit, which CircuitMaker seems to simulate properly.

hen918

Fri Dec 09 2016, 02:36PM

Registered Member #11591 Joined: Wed Mar 20 2013, 08:20PM
Location: UK
Posts: 556

I was going to mention that comment at the bottom of the diagram.
the voltage across the LC tank should be Pi * Vin, so 330*3.14 = 1040 V
1040 / 20 = 52 turns
according to some playing around with ExellentIT, the magnetising inductance is going to be about 6mH with an ungapped core. This will be the approximate inductance unloaded.

The capacitance to achieve a frequency of 25kHz is 6.8 nF.
The Q factor is going to be approximately 0.001, depending on the series resistance of your inductor (I have assumed a resistance of 1 ohm.)

When loading increases the Q factor also increases as the effective inductance drops, to become the total leakage inductance of the transformer with a 0 ohm load. This leakage inductance will be in the 10s of uHs, or much greater with a gapped core.

The only issue now is the huge voltage across such a small inductance. MASSIVE currents now flow.
X(L) = 2 * Pi * f * L
2 * 3.14 * 25000 * 0.00001 = 1.57 ohms.
V(RMS) = 1040/sqrt(2) = 735 V
735/1.57 = 470 A

if R(L) = 1 ohm, P(Dissipated) = 470^2 = 220 kW
oops!
This is why the core is gapped. But Q will still be very low, and the circuit probably won't oscillate.

previous 1 2 3 next

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