Idiosyncrasies of the Mazzilli Driver / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

Idiosyncrasies of the Mazzilli Driver

1 2 next

Move Thread

LAN_403

quicksilver

Sun Jan 23 2011, 11:06PM

Registered Member #1408 Joined: Fri Mar 21 2008, 03:49PM
Location: Oracle, AZ
Posts: 679

With help from the folks here I have found some unique aspects of the Mazzelli driver. I had not fully understood the workings of this driver and the more I learned; the more I experimented. Now I recognize that the "area of activity" is limited - this begs the question of IF there is some other issue that may help (aside from extremely expensive FETs and associated boosting of all passive components).

Obviously the more current and voltage input: the better the results - up to a point. I have found that the type of the capacitor and the winding of the primary have enormous effect on the outcome. The Mylar and the "gumdrop" cap are generally the choice; however the CDE Polypropylene Film Capacitors have a great ability as well. The level of voltage per se' (as long as it's above 250Vdc) is not too much of an influence (as far as I can tell) but they (CDE's) appear to be a slight bit better in terms of the results. The wire used for the primary and number of turns DOES have a profound effect.
Thus far it appears that the resultant arc drops when 22awg or smaller is taken over 6 turns; at 9 (on each side of the tap) the results appear to plummet. 18ga wire is about as thick as I have gone so as to make the coils as clean and neat as possible. The issue of sharp edges on the core are worth addressing as they can possibly cut into the insulation when making a tight coil. A sloppy coil does degenerate the performance from my brief experimentation.

The actual design of the core appears to be a contributing factor as well. It seems that the care taken in making the coil clean and tight equal better efficiency. The driver (as originally designed) doesn't like lower current what-so-ever. Even dipping to five to eight amps will degenerate the output. This driver appears to like very high current almost to the exclusion of the voltage itself (12 @ 20a yields a fine arc but 36 @ 10; not so great).

I have ironed out some problems with the help of folks here but I would love to hear any further input as to aspects of increasing performance aside from those basics I had described. I fairly sure someone has found a combination of efficiency and input energy that yields higher performance & since the combinations are substantial I wondered if someone found the "golden fleece".....Diode alteration may be next on the list however, it appears that the "area of activity" is the primary source of yield. .....Thoughts?

Arcstarter

Mon Jan 24 2011, 03:59AM

Registered Member #1225 Joined: Sat Jan 12 2008, 01:24AM
Location: Beaumont, Texas, USA
Posts: 2253

Well most of these factors apply to any inverter driving transformers, but to what extent each variable matters is different for the numerous inverter circuits.

I know that if i where to make a ZVS driver for the highest efficiency possible, i would replace the gate pullup resistors with a 1k resistor, and 12v into that, and have that drive a gate driver to drive the gates. That would reduce switching losses, make the resistive losses practically vanish, and would reduce stress on the diodes, not that that was a problem in the first place... I just wonder if it would start oscillating on turn on as well. The UC

Also, as for the transformer itself, litz would be better than solid or stranded copper.

Any measurements by visual arc inspection are probably inaccurate as far as efficiency is concerned. Bigger arcs with tighter windings makes sense, the tighter they are, the less leakage inductance, thus the more current it can pull. The same with the wire size, the smaller the wire, the more resistive losses, and the less current it can pull.

The circuit can definitely be made to 'prefer' voltage. It goes by Ohm's law like any other circuit. Just add more turns for higher voltage and less current at the same wattage. But, for any testing of that theory to be valid, you would have to take into account that the resonant frequency changes as you increase the tank circuit's inductance.

Gah, sooo many variables!

Proud Mary

Mon Jan 24 2011, 12:08PM

Registered Member #543 Joined: Tue Feb 20 2007, 04:26PM
Location: UK
Posts: 4992

A quick Google will find you many descriptions and explanations of the astable multivibrator.

In an ordinary astable or 'free running' multivibrator, one transistor switches on while the other is switched off.

It will keep on going indefinitely - a pendulum going backwards and forwards tick tock - which is why it is called 'free running.'
(There be other monostable multivibrators that are 'single shot' - they will change state, or flop once just once, after the power is switched on)

Now the rate at which this pendulum goes backwards and forwards is controlled by R and C - that is to say, the rate at which a timing capacitor is charged up through a resistor. Once the charge on the timing capacitor has reached a certain value , the circuit will change state, and the pendulum will start to go back in the other direction.

Since the circuit does nothing but switch backwards and forward, on and off, tick and tock, flip and flop, its output is a square wave. One moment on, one moment off, with nothing in between (in an ideal world!)

In the so-called 'Mazzilli ZVS' - which I would call a tuned drain astable multivibrator - the timing function performed by R and C in the ordinary multivibrator is taken over by L and C - inductance and capacitance - which together form a tuned circuit.

The tuned circuit is like a bell or chime which goes on ringing - resonating - at one particular frequency long after it is struck.

In the tuned drain astable multivibrator - your Mazzilli thingy - the striking of the 'bell' - the tuned circuit - is performed by the aggressive switching action of the power transistors going on an off.

quicksilver

Mon Jan 24 2011, 06:55PM

Registered Member #1408 Joined: Fri Mar 21 2008, 03:49PM
Location: Oracle, AZ
Posts: 679

Thank you both.
You have given me many things to consider.

The lowering of resistance makes quite a bit of sense. The element of a "tuned drain astable multi-vibrator" as well gives a clearer picture of it's functionality.

The experimentation appears near endless to coincide with what level of energy is applied. As diagrammed I can see why the originator deemed it suitable for very high current; yet his generality when applied to the primary was an area in which he could have further expounded upon.

- .....appreciate the input.

Sulaiman

Mon Jan 24 2011, 07:29PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3141

Before yet another term is coined, the zvs/marzilli/Royer is already classed as a "Current-Fed Parallel-Resonant" oscillator/inverter... CFPR
The terminology is important to me;

Current-Fed ... the dc link inductor should have high enough impedance at the operating frequency that it can be considered as a constant-current-source.
This is what allows PI x (dc input voltage) as peak primary voltage.

Parallel-Resonant implies inductor//capacitor and usually a significant 'Q' ... I vaguely remember Q>=2.something for correct operation.

Benefits of a CFPR inverter include self-running and sinewave operation.
The cost of cfpr vs. squarewave is;
dc link inductor and resonant capacitor required
the transformer is significantly larger than for a squarewave inverter.

The CFPR inverter is really good for low-power and/or reactive loads
also for low-power constant-voltage supplies with low rfi/emi emissions,
or simple low-voltage inverters,
it's not so ideal for making arcs at it's output, or charging capacitive loads.
Other topologies are more suitable (e.g. look at ccps in 4HV)
though a cfpr/zvs/Royer/Marzilli circuit is often used at high power it will rarely be operating in zvs mode ... causing the transistors to heat up.

ok ...that's today's rant..........

Proud Mary

Mon Jan 24 2011, 07:50PM

Registered Member #543 Joined: Tue Feb 20 2007, 04:26PM
Location: UK
Posts: 4992

There you go, Quicksilver! I'm sure Sulaiman has forgotten more than I ever knew about electricity, so you should pick his brains whilst he's here and able to help you.

quicksilver

Tue Jan 25 2011, 08:19PM

Registered Member #1408 Joined: Fri Mar 21 2008, 03:49PM
Location: Oracle, AZ
Posts: 679

To be honest I needed to re-read Sulaiman's post over several times and digest the points made.

If I understand correctly, the Royer driver has a certain "damned if you do; damned if you don't" aspect to it.
If driven with sufficient high current the LOPT is very likely to be destroyed; making it a poor candidate for a reliable HV PS unit. If not driven with high current I have seen no advantage over certain single transistor designs (in terms of longevity of HV output. I had thought I was alone in this observation.

In almost every case where I got a large 'flame-like" (or for that matter even fairly large) arc from the performance of the driver I killed the LOPT after a time. I had thought I was making some error in my topology, etc in that I lost a couple of LOPTs when driven to what I had come to EXPECT was their potential with this driver.

Superficially, dropping it's level of performance (via lower current) to a point where in the LOPT continually survives, I see no greater reason for using it (the Royer driver) compared to some single transistor designs IF the end result is a HV supply of reliability with lower demands of input current (lead acid battery excluded - high current supplies are not cheap).
I'm not sure that a LOPT was ever designed for a (relativity) higher current supply. And as a novice I have seen much less expensive, simpler inverters as well.

Perhaps emboldened is the wrong term but as I read Sulaiman's post it tracks with what I had thought perhaps was "using the wrong tool for the job" if longevity is expected in an HV supply.

Sulaiman

Wed Jan 26 2011, 12:17AM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3141

One of the important specifications of a LOPT is it's output current
- which is normally not published for public consumption.
'typical' crt beam current is only 1 to 3 mA ! ... 25W to 100W typical.
- this is what LOPT are designed for, even then the LOPT is one of the 'weakest' parts of a crt tv/monitor.

The other significant specification is output voltage, which is commonly specified.
The output voltage is limited (mainly) by winding insulation and diode reverse breakdown.
The other source of output voltage limitation is 'clamping' of the primary flyback voltage,
a lopt is a transformer with a Ns:Np (primary # turns) : (secondary # turns)
for the larger lopt (the ones we want) a 1500V transistor is required to allow about 1kV primary flyback voltage for about 30 kV secondary peak voltage.

If a cfpr inverter is used the output voltage is further limited because the waveform is sinusoidal ... equal positive and negative peaks.
This increases the reverse voltage stress on the diodes.

Summary: for long term reliability use a lopt well within it's limitations,
for the highest voltages the flyback topology is best,
for big fat arcs that last a few seconds use a cfpr/zvs/royer/marzilli.

quicksilver

Wed Jan 26 2011, 01:40AM

Registered Member #1408 Joined: Fri Mar 21 2008, 03:49PM
Location: Oracle, AZ
Posts: 679

I did not realize that output current was not published for public consumption but it certainly makes sense in the light of a competitive marketplace and the "edging out" of CRT oriented devices.
Frankly, I could not really find a place that had specifics on manufacturing specs on LOPTs - nor saw clearly defined model numbers on many & assumed that they were a "trade secret" of the manufacturer's as well as perhaps changed as models changed from alterations in topology wherein components also changed to fit a profit margin, features, etc.

I was unfamiliar with stress on the diodes; however that makes a great deal of sense.
Being unfamiliar with the use of an oscilloscope during construction of my "toys" - still the issue of sinusoidal waveforms is very logical (last time I even used one was an old analog when in secondary school and loved "everything radio") .

It appears the LOPT is rather delicate outside of it's design parameters - something that I should bear in mind when thinking of differing future HV supply designs......(?)

James

Wed Jan 26 2011, 02:54AM

Registered Member #3610 Joined: Thu Jan 13 2011, 03:29AM
Location: Seattle, WA
Posts: 506

The output is not usually published because it is not typically relevant. Flyback transformers are custom designed for a particular set or series of sets. They are spec'd to be adequate to drive the CRT in that application, the end user doesn't care what current the flyback is capable of as long as it performs properly in the original application.

I really like this driver. It's not the best for DC flybacks, but I've seen nothing more effective and robust for the AC type and self wound transformers. I've built a couple and pushed one as hard as 300W so far and the mosfets barely get warm to the touch with *no heatsink* at all. I've even done some experiments with induction heating by winding only 2-4 secondary turns on a ferrite transformer and using those to drive the work coil. A higher frequency than mine is set up for would work better but it does easily heat small screws orange hot in a few seconds. For driving flybacks, liberal application of silicone in critical areas or an oil bath is a must for really pushing things. I ruined one when the arc flashed over into the coil.

1 2 next

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