Ferrite rings

[dexters_lab]

If i was to install a ferrite ring on a two core cable (figure 8 style cable, 12v DC) from a plug-in wall wart power supply i can just loop the two core wire around the ring but would it be better / make any difference to separate the gnd and +ve wires and wrap them around separate rings?

Whenever my nixie clock is on it seems to murder one of my wifi APs, which is nearby so wanted to try and see if the ferrites helped on the power cable which runs right next to my router.

[Andy Watson]

Yes. Just wrap the fig-8 through the ring - assuming it's common-mode rejection you're after.

[skipjackrc4]

Splitting the wires between two different ferrites would be counter-productive, as it would no longer attenuate the common mode current. 

[dexters_lab]

thanks, good to get some clarification on it

[T3sl4co1l]

Not quite correct: most often, the wires are wound in the same direction, but on opposite sides of a single core, maximizing leakage inductance while still getting current cancellation.  Using two independent cores, however, will probably saturate them, because the currents are unbalanced (instead of ferrite, powdered iron cores would be required, and the inductance will be much lower).  The same is true if you screw up the direction of one wire -- the currents add rather than cancel, the core saturates, and you get little effect.

I'm curious how one could "screw up" a nixie clock such that it trashes the 2.45GHz band!  I wonder, does it matter how close in proximity they are, or if they are plugged into the same outlets?  If it is interference, I would be surprised if it's going up the power line; it's probably radiating from something internal, which means a careful inspection of the circuit layout and what kinds of things could be causing it.

Tim

[dexters_lab]

I'm curious how one could "screw up" a nixie clock such that it trashes the 2.45GHz band!  I wonder, does it matter how close in proximity they are, or if they are plugged into the same outlets?  If it is interference, I would be surprised if it's going up the power line; it's probably radiating from something internal, which means a careful inspection of the circuit layout and what kinds of things could be causing it.

i did wonder if the interference was just emanating from the device itself, but i didn't design it. The build guide eludes to it being somewhat noisy because it defaults to turn off the HV when it's reading time from the MSF radio time module which plugs in the back of it.

The ring is a simple cheap thing to try, it seems to have done the trick though as wifi seems fine now.

[TerraHertz]

i did wonder if the interference was just emanating from the device itself, but i didn't design it. The build guide eludes to it being somewhat noisy because it defaults to turn off the HV when it's reading time from the MSF radio time module which plugs in the back of it.

The ring is a simple cheap thing to try, it seems to have done the trick though as wifi seems fine now.

Oh good grief, is the nixie HV supply a plain 'transistor driving small ferrite transformer' switch mode? AKA unintentional wideband noise transmitter. Perhaps with the 12V DC supply wires acting as transmit antenna.
And they modulate this wideband noise source with a square wave by turning it on and off repeatedly. Great.

[T3sl4co1l]

What gets me is, is it really 2.45GHz?  Because it's one thing to have switching noise, even RF noise from glow/arc discharge instability (you can use nixies as oscillators... probably not recommended operation ), but to get it over 100MHz, you have to really work at it.  The fastest switching circuit I've made has very little signal above 100MHz.  That oscillator (or whatever it's using) would have to be using an RF transistor to even be capable of instability at that kind of frequency.

It's also worth noting that generic ferrite beads don't do much beyond a few hundred MHz; special materials are required for attenuating GHz, and even the most special-purpose material doesn't act like a conventional magnet at those frequencies.  It's the kind of thing that... yeah it's worth a try, but adding that ferrite bead could just as well make things worse, who knows?

It's just as well possible it IS a lower frequency (below 30MHz is the range most critical to conducted emissions -- anything that finds itself on cables), and the router is just absolutely terrible (if it fails just from nearby noisy wires, I doubt it would actually pass a real FCC Part 15 test).  Yeech!

Tim

[mrkev]

What gets me is, is it really 2.45GHz?  Because it's one thing to have switching noise, even RF noise from glow/arc discharge instability (you can use nixies as oscillators... probably not recommended operation ), but to get it over 100MHz, you have to really work at it.  The fastest switching circuit I've made has very little signal above 100MHz.  That oscillator (or whatever it's using) would have to be using an RF transistor to even be capable of instability at that kind of frequency.

Couldn't it be noise at freq. of intermediate mixer, assuming there is one in wi-fi?

[T3sl4co1l]

Uhmm... maybe?  I don't know too much about the internal design of those things, but they're always a bunch of chips buried under RF cans... so, that part should probably be pretty robust?

I guess the other things I might suspect would be poorly controlled signal lines to and from the processor and other chips (resulting in crappy logic signals, erratic values, extra transitions, possibly even extra pin change interrupts or something, etc.), or just a really bad DC-DC converter or something (assuming there's circuitry inside to generate 3.3V, 5V, etc., whatever is needed).  That sounds silly, an SMPS that can't take noise -- it has to tolerate its own noise at least -- but who knows.

Tim

[mrkev]

I think they mix the input with another two-point-something GHz to get it at lower freq so they can easily demodulate it. Not sure as I was never in this kind of thing... Anyway even good shielding is not bulletproof when you have source of wide bandwidth noise few cms close.

[dexters_lab]

blimey, you guys have gotten all excited about this!

i wasn't going to get any more involved but the clock will be on my workbench tonight hopefully if your interested. I have a different wifi router in that room too so i can see if it might affect it too.

the clock is located within a meter of the router (Linksys running DD-WRT), the clock and the linksys are powered from a smps wall wart that plugs in just behind the router so the clock power cable is within a few cm of the router.

The clock uses pwm from the micro to drive the 12v to hv circuit, which is then multiplexed through a 20 channel driver (driven by serial data from the micro) so this drives the nixie's in 3 pairs for the 6 tubes. It's probably pwm'ed here as well as they have fading digits. There is also a 7805 off the 12v to run the micro.

The Schema is at the back of the build doc:
http://www.pvelectronics.co.uk/kits/qtc/qtc_v8a.pdf

The wifi issues were seen as poor connectivity, the two laptops that were using this access point (both in the same room) would not actually drop out but would simply struggle to load pages. If i removed the power from the clock, the problem went away.