Spark generator - How to remove high voltage/frequency noise

[bdsmith21]

Full disclaimer: I do not suggest building anything that is discussed below. I am aware of the danger inherent in anything that creates flames. Please don't try this at home, unless you really understand what you are doing, and then please take proper safety precautions. I am currently testing my system using only compressed air. I will do this until I am confident in the electrical portion of the design. I assume all responsibility for my personal safety and those around me. You should do the same.

I have built a flame effect device that uses a Teensy 4.1 to control up to eight gas valve solenoids and 16 spark modules. I am having trouble with what I think is high frequency/high voltage noise generated by the spark modules. I am looking for suggestions on how to reduce this noise.

I am using the Teensy 4.1 to communicate through serial to three daisy-chained 74HC595 shift register boards. The shift register boards use mosfets on the outputs. The first two boards control the 16 spark modules. The third board controls 8 solid state relays that then control the 8 24V gas solenoids. Maybe this isn't the best way to do this, I am no electrical engineer.

The system uses 48V input from a Mean Well 120v power supply (I plan to eventually replace this with a 48v battery). That is reduced to 24v by a second Mean Well power supply. Then there are three voltage regulators; a 3.3v regulator sends power to the three shift registers VCC, a 5v regulator powers the Teensy 4.1, and a 5V 9A regulator sends power to all of the mosfets.

I am using the serial monitor to verify that the button presses and output to shift registers is correct. The serial monitor always prints the correct button press and 8bit variable, so I don't think my problem is in my programming or serial output.

The shift register boards have led indicators on all of the 74HC595 outputs. The 74HC595s have output enable pins. I have these connected to buttons that either ground or connect this pin to 3.3v (VCC on shift register board). When I only activate the gas solenoids, everything works properly, but if I activate any one spark module, many of the 74HC595 output leds will flicker and the corresponding spark modules and gas solenoids will sputter/spark. I have 6' long 20 gauge Igus style multi conductor cables running individually to each spark module. Putting distance between these cables seemed to help but only had maybe a 50% improvement in perceived noise.

I am not an electrical engineer, so I am a bit unsure of what to try here. I don't know what frequency or voltage the spark modules output. Should I try adding ferrite beads somewhere? Would adding a capacitor between the inputs to each spark generator help? Do I need shielding for my cables? I'm just not sure where to start.

Code attached. Links below.


3.3V 2.6A Regulator (up to 36V input): https://www.pololu.com/product/2857
5V 2.5A Regulator (up to 36V input): https://www.pololu.com/product/2858
5V 9A Regulator (input 5-38V): https://www.pololu.com/product/2866

Shift registor/Mosfet boards:https://www.summet.com/blog/2012/08/05/serial-shift-register-mosfet-driver-version-1-1/
Spark Generator: https://www.ebay.com/sch/i.html?_from=R40&_trksid=p2553889.m570.l1313&_nkw=1PCS+DC+3v-6v+to+400kV+400000V+Boost+Step-up+Power+Module+High-voltage+Generator&_sacat=0
Solid State Relay: https://www.digikey.com/en/products/detail/sensata-crydom/DC60S3/221844
ASCO 8210g076 Fuel Gas Solenoid Valve 24V 10.1Watts :https://www.emerson.com/documents/automation/catalog-gas-shutoff-series-210-asco-en-4290880.pdf





View attachment Fire_Effect_Shift_Registers.ino

 

[MarkT]

Shielding for everything in a high-noise environment is a given - in fact there's rarely an excuse not
to shield enclosures and cables...

Good decoupling for supply rails can prevent transients travelling along them too.

Keep signals low-impedance to reduce the effect of capacitive interference - so pull-up resistors
of a few k, not 100k, potentiometers 5k rather than 100k, etc etc.

Yes capacitors on inputs can be good - feed-through caps are the best as they don't compromise
shielding - pretty standard to use 1nF to 10nF ceramic caps on DC rails (or low bandwidth signals).

Often careful attention to how things are laid out can avoid the worse problems.

 

[bdsmith21]

Hi Mark. Thanks for the advice. I would be glad to try these things, but I don't exactly understand what I need to do. I have uploaded some labeled pictures which will help everyone see what I am working with.


Decoupling supply rails: What exactly does that mean.
Keep signals low impedance: How can I do this when my signals are going out to premade devices (coils). I can't lower their impedance can I?
Caps on inputs: What size/volt spec and where exactly do they need to go?
Layout tips: Please see the attached photos.

 

[MarkT]

First step is get a metal enclosure or wrap you existing enclosure in something conductive, and ensure your wiring is all shielded.

Decoupling supply rails: What exactly does that mean.

http://www.thebox.myzen.co.uk/Tutorial/De-coupling.html

Keep signals low impedance: How can I do this when my signals are going out to premade devices (coils). I can't lower their impedance can I?

outputs are by nature low impedance, its sensor inputs that might be high impedance, and thus pick up lots of noise.

Caps on inputs: What size/volt spec and where exactly do they need to go?

Depends on the signal bandwidth and impedance - for slow signals 100nF ceramic is a good option (voltage same as you circuit or
greater).

Layout tips: Please see the attached photos.

Keep the noisy wiring (spark gaps) well away from your sensors and circuit as far as is possible.
Ensure each solenoid has a free-wheeling diode across it. If you can shield the high voltage cables to the spark
gaps this may help enormously (if they aren't already shielded).

 

[jonr]

What happens if you leave all the output enable pins connected to 0V? Tri-state outputs combined with weak pull up/down resistors aren't noise resistant.

 

[bdsmith21]

I felt silly, but wrapped the box in tinfoil. It didn't seem to have any effect.

I have all three output enable pins connected to switches. These switches connect the OE pins to ground or 3.3v+ depending on their position. Shift registers 1 and 2 (spark generators) share the same switch. Shift register 3 (gas solenoids) has its own switch. The fourth tiny shift register's OE (fire button leds) is always grounded.

 

[mlu]

Perhaps optocouplers between shift registers and spark modules