A few years ago, the kind folks here helped me to develop a versatile circuit board based on the ATMega for animating incandescent bulbs in various old-school sign projects. The board has worked wonderfully for several years now in a number of commissions, and I'd like to thank the community for helping me out.
Now, I'm hoping everyone can help me tackle a slightly different project. I'd like to adapt the original board to switch an inductive load -- specifically, a neon transformer. Many of my projects involve neon (think retro signage) and I'd like to add a "failing" flicker effect. One company sells a unit specifically for this, but I'd love to see if I can tackle this myself for a bit less cash. (See this site for the effect I'm looking for.)
On the original board, I use triacs to switch the mains voltage for each channel of incandescents, along with a cap and inductor to eliminate noise. I know that switching an inductive load, however, requires different considerations. I've Googled the problem quite a bit, but most discussions seem to fall into a rabbit hole of theory and never produce a practical solution. So, I'm hoping the folks here can help me figure out the simplest and most reliable way of creating the desired effect that I can be sure won't destroy any transformers.
The archetypal flickering is on the high voltage side. While it is possible to switch 15KV, it's going to require relatively expensive components and some good knowledge of HV.
You've obviously decided to switch the low voltage (mains voltage) side, which is at least a much more common operation so there are inexpensive parts available. I don't know if that will give the effect you want. Something like a PowerSwitchTail could be used as a test, to see if an Arduino-simulated flickering might look OK.
Switching a switchmode power supply like this will destroy it very quickly. Like under a minute. A transformer will be a little more robust. It's certainly going to work for a stage effect or a show, but I would not expect it to work for more than a month if it's left on 24 hours a day. You might be lucky. Those neon transformers are built to some pretty conservative specifications.
Does it have to be HV neon? Can you simulate neon with EL wire or LEDs?
If you're really brave, do the switching on the HV side with a stepper or servo pushing on a wire with a really long insulated rod. That will make the vital sound effects and secondary sparks too.
Thanks for the suggestions so far. However, I want to make something clear up front: I'm not interested in EL wire or LEDs. They are not a substitute for actual glass tubing. My only goal is to make a unit to create an effect with a neon transformer, nothing else. (I've been working with neon in my commissions for years now, so this aspect isn't something new to me.)
Also, switching the high side is an incredibly bad idea. Sparks are not a desired outcome. The effect I'm looking for is clearly possible switching the low side, as the unit I linked to does just that. I could buy one and reverse-engineer it, but I prefer not to work that way. I'd rather develop my own circuit if at all possible.
As for the non-zero-crossing SSRs, that's a good start, but I think I'll need a little more guidance. I've read this paper a number of times, but I'll admit I can follow it only to a point. As I understand it, I'll need to switch the SSR on only at the 90 degree point of the AC waveform to avoid a high inrush current.
I detect the zero crossing in my original circuit and incorporate some timing in my programming to switch the incandescents on at different points to control brightness and to create fading effects, but it's nothing as accurate as I think I'd need to safely switch a transformer. How might I go about making sure I switch the SSR at a specific point in the waveform? Are there SSRs that are designed to do just that?
WesleyTx:
I detect the zero crossing in my original circuit and incorporate some timing in my programming to switch the incandescents on at different points to control brightness and to create fading effects, but it's nothing as accurate as I think I'd need to safely switch a transformer. How might I go about making sure I switch the SSR at a specific point in the waveform?
.
Detect zero crossing wait for 90 deg (5 milliseconds in europe then fire)
Wait at least 30 cycles then switch off.
Downside is that ac mains frequency can vary +- 0.5% , im not sure that would make a significant difference though.
I wonder if the more recent neon sign inverters would be easier to use, need to check manufacturers data for that though.