Using a Flyback

[Trilitheum]

I have a transistor driven rectified flyback, it will draw arcs of about 1/4 inch. Are their any specific problems with running a fusor off one of these? Do they have enough juice and current output to run a demo fusor as well as a NST would?
Anyone who has experience with one of these with a fusor, your advice would be much apreciated.

Thanks, Brad

[Richard Hester]

A respectable fusor requires hubdereds of watts of HV. A transistor driven flyback will not supply this amount of power, as they are rated for about 20-30W, maybe 50W for a hot one. Even a neon sign transformer would be better.

[AllenWallace]

I agree with Richard, a single flyback is very weak compared to neon sign transformer. However, I did try a flyback and it will power a demo fusor, but weakly. The posier is plainly visible, but *very* dim compare a NST.

The Flyback in computer monitors and TV generally have build-in diodes, and the polarity is WRONG. We want negative HV for the fusor, but a TV wants positive HV. So find *very* old TV flyback without built-in diodes.

I did try hacking a TV flyback. I grounded the positive end, and floated the negative after trying insulate everything with epoxy. It sort of worked, with only some arcing from the coil to the frame, and the frame to my driver coils.

[Richard Hester]

The energy delivered by a flyback transformer is limited by the amount of energy that can be stored in the core without saturaitng the ferrite. If you treally want to hack a flyback, you can wind a push-pull primary on the core and drive it with a pair of mosfets driven by a 3525 controller chip running at full duty cycle. The output will need a high voltage fast recovery full wave bridge rectifier. With a good, big flyback, you may be able to get 150 watts or so, depending on theoutput resistance of the secondary winding. I'd try to find an x-ray tube head instead and use the transformer.

[DaveC]

You can usually get much more output from a flyback transformer, if the primary is removed and rewound as a bifilar (two wires wound as one) winding. Opposite ends of the primary windings are then joined and connected to the positive of an husky low voltage power supply. With switching transistors (either HEXFETS, IGBTs or just plain ole 3055's connected at each of the remaining ends of the primary to the supply negative, you now have a bipolar primary drive.

The transistors of course, need properly phased gate (or base) drive circuitry providing preferably square wave. Usually some modest sized (few nF ) caps across the primary reduce primary current by bringing things to psuedo-resonance.

Polarity at the secondary is mostly a matter of choosing the proper direction of the diode stack (yours, not the built-ins). Just ground the winding end closest to the core metal, as it will be the one that would otherwise flash over.

You should be good for perhaps 10 kV at least, but that depends on how many turns you wind for the primaries, and... your LV DC input. Usually you will get an effective peak to peak drive voltage across the primary that is twice (2x) the DC voltage.

As to Frequency, flyback cores are designed for 15 kHz, up to about 100 kHz . The best way to determine the optimal frequency for your setup is to apply the drive signal, with about half maximum DC from your power supply, to the transformer, and slowly increase the frequency till the output peaks. This is also usually the most efficient operating frequency, meaning least heat.

Done this way, you should be able to get at least 100 watts from a core with a 0.5" square cross-section. For more output power, just run the flybacks in parallel, but connect them at the rectifiers, rather than at the transformer HV terminals.. The rectifiers will provide isolation.

Dave Cooper

[Trilitheum]

Thanks for all the info, I suspected it lacked a little in the power area. I have two of them actually, one internially rectified and one monster from a 1967 B&W TV, that will draw out 4 inch arcs. But the positive output is going to be more trouble than it is worth, I should try to get my hands on a NST, I also have two MOT's lying around but the peak output would only be around 4 kv.

Brad

[Richard Hester]

You can ditch the experimental approach by actually calculating the flux for a given frequency and number of turns. You need to measure the cross sectional area of the core first. Do this for the smallest (least area) leg of the core. For some safety margin, you want to keep total the flux swing to about 3000 gauss or so. There are good reasons for the conservative approach, as there are a lot of pitfalls in push-pull converter design and operation for the neophyte. 25kHz is a good overall operating frequency, as it will not overtax fast recovery output diodes. If you can only get standard recovery high voltage diodes, you are better off with the 60Hz approach. Primary turns can be calculated by the formula Np = (Vin X 10^8)/(4 X B X Ae X f). Vin is the input voltage, B is the 1/2 the total flux swing in Gauss, Ae is the effective core area in cm^2, and f is the switching frequency in Hz. As an example, suppose you have an input voltage of 24V, a core with effective area of 1cm^2 (not unusual), a preliminary flux swing of 3000/2 = 1500G, and a switching frequency of 25 kHz. The number of primary turns needed to meet these conditions is (24 X 10^8)/4 X 1 X 1500 X 2.5 X 10^4) = 16 turns. The primary would actually be realized by a bifilar winding of 16 turns, to yield a closely coupled center-tapped primary with 16t on either side.
A 3525 control IC is a good way to drive the primary switching FETs, as it has totem pole outputs that will provide the necessary peak current for fast switching. I would stay away from bipolar transistors such as the 2N3055, as they are difficult to drive properly, and differences in storage time between the two drive transistors can cause the transformer core to walk over into saturation on one side and blow up a transistor. Besides, the 2N3055s are not really even rated for fast switching, which makes matters even worse. Mosfets are not that hard to get these days, and are much easier to drive. They also have no storage time problems. I would recommend oversized devices. A pair of IRF640s would make a good start. They have a minimum 200V breakdown voltage, which will help them cope with any leakage spikes. I can post a preliminary schematic if anyone is interested . This would show the control circuit, drive, snubbers and some suggestions for output rectifiers. The transformer design will of course depend on what is available to the individual. Anyway, this thread should be moved to the HV section...

[Trilitheum]

Yeah, thats an error on my part, I THOUGHT I posted it to the High Voltage Section....
A schematic would be great, its worth experimenting with.

[r_c_edgar]

I've just gone ahead and moved the entire thread into the HV section, so discussion can continue uninterrupted.

--Ryan

[Richard Hull]

The length of your spark is irrelevant. Again, the energy supported by the core is limited, If you had 50 foot arcs the same core limited energy would be there. You would just have a lot of voltage at no current capability. At fusion voltages, fusion and , thereby, neutron numbers are strictly a function of CURRENT!

Fusion demands a minimum of about 250 watts input for an easy detection scenario and about 600 watts to get activation level numbers and 1000 watts for serious neutron outputs.

Ricahrd Hull

[SteveHansen]

Richard,

Did you ever post the referenced schematic? I've done a few medium voltage/power low freq supplies (50 kHz, 3kV and several hundred watts) for running a plasma asher using adapted commercial power inverters (which use the 3525) with a largish flyback-type transformer. Eventually I want to homebrew something that uses IGBTs to get a few 10s of kV at about a kilowatt.

[Richard Hull]

I'll second Steve's request for a simple example schematic, Richard.

Richard Hull

[Richard Hester]

All things in time. The schematic fpr the HV switcher is about half done, as I've been spending some time finishing the All-In One board report and doing a lot of horrible minimum-wage labor on my old house, virtuoso stuff with shovel, ax, mattock, and sledge. I fear my weekends won't be my own until I get the old place primped and sold off.

[SteveHansen]

Thanks for the update. Nothing worse than doing minimum wage labor for no wages.