Tinkering with flyback generator for ozone

I've been recently playing around with concepts for possible ozonolysis equipment. I devised a simple driving circuit to breathe life into an ancient high voltage transformer from an old tv set. A real old one....vaccuum tube based model from the early seventies I guess.

Circuit diagram below:



The brains of the operation is a UC3843 chip, which drives a IRFP450 mosfet in "somewhat resonant" mode and allows for the regulation of driving frequency and duty cycle to be able to select proper operating conditions for the transformer and the ozone generator. The circuit has no feedback from the secondary side and is hence a rather cropped version of a fully capable UC3843 based design. But for the sake of simplicity and safety I opted to omit any feedback from HV side. Besides, the switching waveforms are highly cluttered and would confuse the chip anyway. The main reason for the noise is very probably the bad design of the transformer and a rather high leakage inductance which wreaks havoc in forms of paracitic oscillations and inductive kickback.

Proto board with the drive circuit:



Simple ozone generator (DBD type "dielectric barrier discharge"):



Soft corona dischage at about 25W power:



It all seems to function more or less respectably from the side of electronics, but what gives me surprising grief is the longevity, or rather the lack thereof, of the dielectric in the ozone generator. I've tried regular glass, borosilicate glass and several plastics (PE, PP, PET) and all of them tend to fail after a rather short time of operation which can be measured in terms of hours for the glass and in terms of mere minutes for the plastic. The generator is not overdriven to high gear (15-20W for constant operation with glass) and is not overheating, but nonetheless a crack will appear after some time in the dielectric and short out the generator. What surprises me even more is that a complete ebay unit had the same type of failure after about 10 hour operation....the ozone tube failed and shorted out the power supply. Even though the bought unit was operationg at only 7W. Has anyone else been battling with a similar situation?

Your dischargelooks a bit hot for ozone synth. ─ and it's hard to see how you've configured the air-space in your unit, but hot arcing will produce nitric oxides which you'll recognise by its pungency rather than the clean smell of ozone.

Air-fed generators need practically anhydrous air or compressed, dry oxygen to obviate this.

Quote: Originally posted by Twospoons

I'm surprised your MOSFET hasn't died.

Thanks for the useful tips and suggestions, everybody I'll certainly bring the system into a more "civilized" form and try out the suggestions.
I must admit that the underlying principle of "resonant style" voltage converters seems to elude me and I find it hard to work with such solutions. Hence I have preferred the quick and dirty hardswitching solutions for push pull and flyback topologies. In fact if the leakage induction of the transformer is kept low with appropriate design and the drive circuitry is tuned to a frequency sweet spot, then the results are very good and snubbers can be omitted quite often. But that usually calls for the use of a a factory produced transformer or very meticulate and often technically complicated execution of winding geometry on a diy transformer.
I've played around with push pull converters (12V to 500V dc dc) for quite a bit during the years and there was a really interesting push pull desing from a russian source that utilised the principle of resonating the leakage inductance of the transformer with an external capacitance at about 1,4 times the switching frequency if I remeber correctly. If executed properly the resulting device totally eliminated inductive voltage spikes at mosfets and provided intrinsic current limitation at short circuit conditions of secondary voltage stage without any additional components. The fact that inductive voltage spikes were nonexistant allowed for the use of mosfets with very low Vds (and in conjunction very low Rds) granting a exceptionally high efficacy of the circuit and minimal heatsinking at the switches. In fact a 150W converter dissipated only a few watts worth of heat at the mosfets and external heatsinking was not required at all. The downside was that it created enormous current spikes in the input stabilisation capacitors...something that the electrolytic caps tend to hadle very poorly. In concequence the average quality input stablilsation capacitors overheated very fast and failed if not paralleled in numerous pairs.

Quote: Originally posted by Twospoons

The basic idea of resonant converters is to take a problem - uncontrolled ringing due to parasitics - and turn it into a useful feature. Every transformer primary has leakage inductance and winding capacitance so this forms a tuned loop that rings at some unknown frequency out of sync with the switching. The solution is to swamp the parasitics with large external components ( usually C, but sometimes a fixed L as well) and bring the ringing frequency down to something known and usable. The controller then aims to turn the power switching device on or off at a known point on the ringing waveform. In a ZVS (Zero Voltage Switching) converter, the idea is to turn the MOSFET on when the drain voltage rings down to zero, reducing switching losses, then off again at some pre-defined drain current that will (a) provide enough energy to the output and (b) provide enough energy to the primary L-C tank that it will ring down to zero volts on the next cycle. Does that help, conceptually? The main benefits of resonant converters are reduced stress on the switching devices, reduced power loss in the switching devices and a reduction in electromagnetic emissions (which can be a real pain in a commercial product).

Thanks for the insight....my vague understanding was along the same lines actually
Although I must admit that there are certainly quite complicated details about resonating circuitry which still boggle my mind. One of which is the short circuit current limiting effect of the design with the resonant tank. I guess what happens is that the resonant tank allows only a certain maximum amount of energy to be transferred through the primary mosfets (and transformer) during a switch cycle. Am I on the right track? This amount of energy is predetermined by the parameters of the resonating components (paracitic or deliberate) and chosen as such to able to be handled by the switching elements without damaging them (current through mosfets can not get "sky high" during switch cycle). Reactance of resonant tank will act as a current limiting factor under short circuit conditions? Quite hard to wrap my mind around these effects intuitively....

On another note I conducted two 10hour low power runs with the ozone generator during the weekend and surprisingly the dielectric held up so far (borosilicate glass section from a high volume burette). The glass was actually very thin...0,8mm in average and worked quite effectively for generating ozone (based on cautious organoleptic assertion). The only change I made to the setup (apart from replacing the defective borosilicate glass with a new and thinner one) was to regulate the power setting down to an average of 10W for the generator. Seems that perhaps I was still overdriving the setup in too high power setting on previous attempts and this accounted as the main factor for the failures in dielectric....time and testing will tell.