# compact fluorescent lamp noise on digital signals

The answers to this question about CFLs (compact fluorescent lights) in an electronics lab indicate there is some concern about the presence of electronic ballasts and CFLs in connection to radio and audio electronics. It is my understanding that CFLs have a compact electronic ballast in their base. There is the implication that these answers have to do with the switched on steady state noise.

My question is different because I am interested in the possible effect on 3.3 volt digital wires in a prototype circult during the transition of the CFL from off to on and back to off. These CFLs are on 120 volts AC, 60 Hz and they get switched on and off by randomly behaving people. CFLs are at a distance of 3 meters or more from the prototype circuit. The prototype circuit is powered from a cellphone charger and a USB interface. Sometimes a computer is used instead of the cellphone charger.

I am indirectly observing (because I have no oscilloscope) some unexplained rising edges on Arduino inputs where the processor is programmed to pull up with an internal 20K to 50K ohm resistor. The specification states there is a wide range on the value of the resistor. Digital 1 is 3.3 volts. The problem is intermittent and may occur 1 or 2 times within the space of 20 hours. As a prototype the input pins are connected to breadboard jumpers that are 15 cm long and approximately 24 AWG copper. That gauge is 0.511 mm in diameter. The other end of the jumper is a rotary encoder. In the testing period there is no input activity on the encoder so the firmware can be considered to be in a steady state. A rising edge suggests that noise is enough to drive an input down to digital 0 and the subsequent pull-up brings it back to digital 1.

As a prototype the wires are longer than they will eventually need to be. As a prototype there was no electronics enclosure. I am currently running a long term (24 hour) test with the prototype circuit in an aluminum enclosure. Are CFLs in an electronics lab a known cause of flipped digital signals in a 3.3 volt circuit?

Low is defined as 0.2Vcc-0.1V which I interpret to mean 20% of 3.3 volts minus 0.1 volts, in other words, 0.56 volts.

• What happens when you add lower-valued pullup resistors to the pins in question? What happens if you use shorter wires? What happens when you remove the encoder and connect those wires to ground? Do the errors only happen when a light switch is flipped? – Elliot Alderson Jan 12 '19 at 20:07
• @ElliotAlderson The pull-ups are internal so I can't change them. I haven't tried short wires yet because it seems easier to get some foil pans and paper clip them around the prototype. The problem is not deterministic so flipping CFL lights is not guaranteed to have an effect. – H2ONaCl Jan 12 '19 at 20:10
• WHat happens when you choose Schmitt trigger inputs? or use twisted pairs? Is the system earth grounded? or floating? then all wires include DC adapter become a common mode antenna with an unbalanced input. Don't just focus on the wire inductance – Tony Stewart Sunnyskyguy EE75 Jan 12 '19 at 20:10
• @SunnyskyguyEE75 I have not shielded nor twisted nor shortened wires because 15 cm is not much longer than what a real product would have and this is digital so I will be surprised if it is necessary. I am still in a testing period so perhaps I will be surprised. I haven't investigated what is in the cellphone charger and the USB cable. – H2ONaCl Jan 12 '19 at 20:16
• Common mode noise affects all wires including power. Then your unbalanced input creates a differential voltage. Twisted pairs for all exposed wires, a CM choke and lower impedances reduce the inbalance as well as RF cap. to shunt > required BW. We used to get the same problem in the lab from a Weller iron going on and off. Of course earth grounding 0V shunts the CM impedance and attenuates the noise. SO there are many solutions. even if you have 30m pairs – Tony Stewart Sunnyskyguy EE75 Jan 12 '19 at 20:19

## 2 Answers

You might have magnetic field interference, or electric field interference.

With 50,000 ohm pullups, I'd suspect electric fields.

The capacitance between 2 wires is

E0 * Er * 2 * PI * Length

..................................

natural_LOG( separation^2 / (radius_wire1*radius_wire2))

For 10cm long wires, located 1meter apart, of radius= 1mm,

the capacitance is

9picoF/m * 6.3 * 0.1 meter

== == == == == == ==

natural_LOG(1,000,000)

or about

5pf

--- == 0.3 pF

14

How much current can you get thru that (Air Dielectric) capacitance?

If 20uA, into 50,000 ohm, then you have 1 volt upset.

Lets run some numbers: assume 200 volts, switching in 20 nanoseconds (some opening mechanical switch contacts). Remember --- 20 uA is a problem.

I = C * dV/dT = 0.3pF * 10 volts in 1 nanosecond.

I = 0.3e-12 * 10^+10 volts/second [I think there is a problem; I just don't know how big]

I = 0.3 e-2 = 30 milliAmps

I = 30,000 microAmps; this is 600X bigger than "you have a problem"

• While my wires are not a meter apart but I appreciate this as an example. In the end I had to supplement the internal pull-up resistors with some external pull-ups that "average" down to 4K ohm. This problem has been a lesson that internal pull-ups in the range of 20K to 50K ohm do not have to be useful for what I consider to be a typical breadboard scenario but perhaps they are useful elsewhere. – H2ONaCl Feb 10 '19 at 0:22

Are CFLs in an electronics lab a known cause of flipped digital signals in a 3.3 volt circuit?

CFLs are designed to cost little. There is little incentive to control the turn on surge.

As far as your circuit, there are some things that you can do to help in providing immunity. Stop relying on the internal pull up resistance to keep the line high. What you need is a lower impedance. Use external resistors for pull ups. I would use 1K.

The other thing to do is to clean up your wiring. Reduce any loop areas. What I mean by that is for you to consider the loop area of a circuit. If you have an external circuit, look at the complete current path. from the input, to the device, the return conductor, the connection through the board, the complete path current can take to get back to the starting point. Worst case is if the entire path was a circle. Best is two parallel adjoining wires connected to a low impedance source.

USB chargers are battery chargers and are not a quality power supply. They will have little noise immunity. A better supply can help, but will only reduce any line disruptions. It can be very likely the problem you are experiencing is coming in through the power supply.

Anyone with some basic knowledge can put electronics together and make it work. Making a reliable product can be more challenging.

• In parallel to the internal pull-up resistor I tried various external pull-up resistors and at about 4K ("averaging" the internal and external) the problem seems to be almost eliminated. Since battery operation is intended I don't want to go very low. – H2ONaCl Feb 10 '19 at 0:16