High Voltage Forum

High voltage => Voltage Multipliers => Topic started by: Andy Kay on June 06, 2018, 08:13:56 PM

Title: Voltage Multipliers keep destroying drivers
Post by: Andy Kay on June 06, 2018, 08:13:56 PM
I've been playing with Cockcroft Walton and Marx configurations, and I've been running them from a 12V car battery in order to avoid getting high voltage spikes on my mains electrical system. I've used various kinds of devices for stepping up the 12V to provide a feed for the multiplier, such as CCFL inverters, ionizer circuits, and SS Tesla coil drivers, and every time the system works for a short time before the drive stage for the transformer primary coil burns out. Is there a standard way of dealing with this problem?
Title: Re: Voltage Multipliers keep destroying drivers
Post by: FilipŠebík on June 07, 2018, 04:43:07 PM
Limit current or use 1:1 transformer as a protection for HV spikeback (if that's a thing, duh)
Title: Re: Voltage Multipliers keep destroying drivers
Post by: Mads Barnkob on June 08, 2018, 09:14:22 AM
It sounds like you have problems with either inductive kickback/EMF, lack coupling capacitor on the primary side of the transformer to avoid DC offset or you are saturating the transformer from high load and got no protection in the driver against transformer saturation.
Title: Re: Voltage Multipliers keep destroying drivers
Post by: MRMILSTAR on June 15, 2019, 05:19:35 PM
I built one of those Chinese 20-stage Marx generators that you see on E-Bay and was experiencing a similar problem. It would occasionally blow the ZVS driver which was driving the flyback transformer. The problem stumped me for a while. Then one day I noticed that when it fired I could see a dim arc jumping across the resistor isolating the flyback coil from the Marx generator first stage. Some of the discharge energy was feeding back into the flyback coil which then fed back into the ZVS driver. I replaced the 2 resistors with higher voltage units and the problem was solved. You have to check those Chinese designs.
Title: Re: Voltage Multipliers keep destroying drivers
Post by: MRMILSTAR on September 10, 2019, 10:02:25 PM
One potential fix is to place a suitably-rated varistor across the transformer primary.
Title: Re: Voltage Multipliers keep destroying drivers
Post by: johnf on October 10, 2019, 08:40:40 AM
hopefully your transformer has an electrostatic shield between primary and secondary to cancel capacitive coupling primary to secondary especially important if you have flashovers as you will impinge your output voltage as a pulse into your primary circuit
then TVS diodes acroos the drive devices set just below their max voltage and tvs 20 volt diodes gate to source / base to emitter

ignore the above and keep feeding the silicon gods your expensive sacrifices
Title: Re: Voltage Multipliers keep destroying drivers
Post by: John123 on March 05, 2020, 08:07:35 PM
Old thread I know, but I'm just going to leave this little gem I came across on 4hv as to why charging capacitors often shorts the switching devices. Source thread dates from 2008 https://4hv.org/e107_plugins/forum/forum_viewtopic.php?49027.0 (https://4hv.org/e107_plugins/forum/forum_viewtopic.php?49027.0)

It is normal for mosfets to die in a circuit like this when we attempt to draw arcs, or charge large capacitances which will both 'clamp' the output voltage to some value.

I did not understand it myself until recently, but I now only see a great amount of misconceptions being spread over and over, which I would like to stop.

It needs to be understood that any transformer or inductor can work only with AC, and will represent a dead short to any DC component. Volt-seconds in positive halfwave need to always be same as volt-seconds in negative halfwave; although their shape is irrelevant.

A flyback is just a transformer with relatively low magnetizing inductance due to large air gap introduced in it. It works by putting an amount of energy into this magnetizing inductance while switch is on and releasing it into secondary while it is off.

If we drive the switch by constant duty cycle from NE555, we have predetermined the volt-seconds we put into positive halfwave with each cycle.

The expected problem is, if we loaded the output of the transformer with an arc - there is absolutely nothing assuring that negative half-wave volt-seconds will be the same as what we predetermined.

Due to low impedance of an arc the negative voltage is actually going to be clamped to some very low value. If we attempt to charge a large HV capacitor, like to run a tesla coil, the same will happen as capacitor basically represents a short to DC output.

In any case, when positive and negative Vs aren't trimmed, the current will integrate up for their difference. This is how the DC component appears in the primary -not instantly, but over a period of several cycles. At some point the flux (which follows the integrating current) will saturate the core, decreasing the primary inductance heavily and overheating the switch.

I'm pretty sure this is what happens in all those ''NE555 flyback drivers''.

Snubber like a TVS or RCD clamp will do nothing for this. It will absorb energy put into the transformer in case output is unloaded (which will be a great deal of heat on the snubber).

In case clamp is not present this heat will simply be dissipated in the switch avalanching.

The quasi-resonant scheme is what is used in TV's, IIRC, and it is only that can actually restore energy to supply bus and increase efficiency a bit. Still it does absolutely nothing about the core saturation problem.

Sometimes the circuit can work, but relatively poorly. I think it is because, as the core approaches saturation, the output diminishes, and voltage drop across the arc increases until it gets high enough to balance the volt-seconds and stop the flux walking.

But still it usually results in hot transistor and hot ferrite core.

The industry standard way of solving this problem is called Current Mode Control, which actively prevents the current integration by cutting the cycle off when current reaches predetermined value. Look for datasheets and application notes on UC384* series of SMPS controllers.

Only with current mode control you could safely charge capacitors and draw arcs without problems.
But, still there would be the problem of what happens to the converter when unloaded - this may be solved by quasi-resonant drive, or voltage feedback. Voltage feedback would again, along with all control loop problems, probably make impossible drawing of stable arcs from the output - and who wants a flyback driver he can't draw arcs from!

And after all you have done you would just have a flyback driver, and I guess not a too powerful one.

To cut it short, building an efficient and powerful flyback converter that can perform all feats you guys want is probably one of hardest and most pointless things in electronics I can think off.

So, my advice is:

- Ignore Jan Martis, his site, and all the schematics of ''NE555 flyback drivers'', never build them. And I'm saying that solely because I'm evil. ( :D )

- For driving flybacks at low power levels (up to 500W or so), with all feats you need (drawing arcs, capacitor charging..), I can't think of anything better then ''Mazzili'' flyback driver, or a simple halfbridge of mosfets driven by SG3525, TL494 or some other controller. They each have their + and - but it's mostly about what input voltage do you want to use.

- For higher power levels, it is just that flyback transformers won't stand them for long. But if you own a dedicated transformer, my probably sole recommendation would be an SLR inverter.


Totally agree with everything Marko wrote here. This is *exactly* what happens when you short-circuit the output of a flyback converter that has fixed programmed duty ratio. The fixed on-time keeps pushing more energy packets into the core but the shorted output doesn't allow any energy to escape during the off-time. Marko's explanation is spot on.

With no load on the output the peak voltage developed across the switch immediately after the on-time will be the MOSFET avalanche rating. All of the energy put into the flyback transformer during each on-time will be dissipated in avalanching the MOSFET during the early part of the off-time.

The flyback circuits in TV receivers actually use a quasi-resonant arrangement where a resonant capacitor connected across the switch makes the energy stored in the primary winding re-circulate resonantly if there is no load on the secondary of the flyback transformer. It also fixes the peak voltage across the switch by design, so that it can not go any higher even when the HV output is open circuit. In a TV the on-time is typically something like 52us and the half-sine shaped flyback pulse is produced in the 12us of off-time. Peak switch voltage on the primary side is usually somewhere in the region of 1200V for a 150V supply, but the turns ratio of the flyback transformer steps this up to maybe 20kV on the secondary.

I've never seen a good diagram of the waveforms of a TV's flyback (line output stage) on the net but discussions about this frequently come up here, ...so i've posted my own sketch here to go along with this thread:

http://www.richieburnett.co.uk/temp/flyback.gif (http://www.richieburnett.co.uk/temp/flyback.gif)


Jan Martis then went on to develop this driver which when constructed with a good layout and tuned to your flyback will give a robust driver (the 220n needs to be selected to keep the peak drain voltage from avalanching the fet under normal operation, better higher voltage low resistance fets are around these days). Place a resistor in series with the UF4007 if the driver isn't stable, about 10 ohm is plenty.

Again, I know this thread is old but there's probably countless people searching for solutions to the same problem on the web who'll land on this page. 4hv is a great source of information but it can be hard to find it these days.
SimplePortal 2.3.6 © 2008-2014, SimplePortal