Small DIY plasma globe using ZVS oscillator

[davekni]

This is my latest little project - only 20 watts.  Made a plasma ball from a plastic wine glass filled with about 15kPa of argon.  Works some with air at lower pressure, around 5kPa.  Experimented with argon/helium mixtures, but plain argon was the brightest.  Don't have neon nor xenon around (due to expense), the normal gasses inside plasma balls.  Mine isn't as bright and doesn't have multiple colors.  Still fun.



For the HV transformer, I used a pair of small already-potted transformers from EBay:
https://www.ebay.com/itm/20KV-high-frequency-high-voltage-transformer-ignition-coil-inverter-driver-bNWUS/323974762353?hash=item4b6e668f71:m:mg4RAb8mo-3R0Me2JUcxcfw

The pair are electrically and magnetically in series, both primary and secondary.  Magnetic path is closed with some NiZn ferrite pieces (low electrical conductivity type).  Still some gap due to the potting.  Driving circuit consists of a normal hard-switched buck converter using the same inductor as the ZVS input inductor, 19V to 15V(adjustable), with ~1.2A current limit.  Separate buck and ZVS boards, still connected with clip leads in the pictures:





The HV winding capacitance and load (plasma ball) capacitance alone makes a ~25kHz resonant frequency.  So, I added only a little bit more primary-side capacitance.  This creates an issue with the two possible resonant modes, one at ~25kHz and another (with primary and secondary voltages opposite) at ~530kHz.  To keep the ZVS oscillation from locking to 530kHz, I added a series LRC notch filter across the ZVS output (transformer primary).  Schematic is:


Here are a few waveforms of the ZVS outputs (transformer primary terminals).  First is with low load (nothing touching the plasma ball).  Second and third are with load (hand on plasma ball), showing lower voltage (due to buck-converter current limit) and a cycle or two of high-frequency mode ring at each zero-crossing.  Without the notch filter, the ZVS locks to the high frequency and stays there.




Any suggestions for improvements?

2/2020 update:  After a longer run, the little transformers developed an arc path between the two, burning through the potting material.  I've replaced them with two identical parts glued end-to-end instead, avoiding the series HV being adjacent.  Haven't ran that configuration for long yet.

[John123]

Very cool! Wish I could help with your question for suggestions but I'm not at that level of skill yet, maybe some ferrite beads could help with the spiking on the primary waveforms?

Aw I hate it when hv transformers have insulation breakdown during the fun.

[ritaismyconscience]

I bought the same ferrite transformers a while ago. Since the transformer needed a pulse of around 100Vpp to form the arc in flyback mode, I needed to step up the output of the ZVS using another transformer.

I got a power supply ferrite core and made a 5+5:60 transformer. I used the ZVS circuit but with 30V 60A mosfets and 24V zeners for protection. When I powered the ZVS with 5V, I got around 60Vpp at the output of the transformer, so not as good as the flyback circuit. This was then fed into two transformers connected in series (I think they're rated for DC, not AC)

Here's the result, circuit uses 40W at 5V.

I can stretch the arc over an inch.