A Real Odd Issue

[J'adore electrics]

Interesting read thou! Good for future works to keep this potential issue in back of mind.

 

[Phoenix]

Just asking!
If the lights were on different phases would this affect the frequency and possibly amplify the emc?

Probably not, it sounds like the noise is comming from the later stages of the switch mode power supply cirucits, after its already been converted to DC, but as the lights will not be in sync, you could find that the effcetive frequency is dependant on number of units in close enough vicinity with the peaks comming from different units

Switch mode units can also result in harmonic distoration of the mains though when units spit across phases have a current that comes in a burst at the voltage peak, and instead of balancing against one enough as in a sinusisol current draw, the harmoic part of the RMS currents actually add in the neutral rather than cancel! as you have three sets of current peaks occuring at separate times and you start to get 150hz components from 3x50 (as well as higher multiples of 3x50 as well).

AC theory is something that can get a lot more complicated than the couple of afternoons in college that most of us probably struggle to remember after a couple of decades on the tools (and most of the current crop probably sleep through it, if it hasn't really been dumbed down already ) Maybe thats a tad unfair - I'm just semi-amazed that that back in the later 1800s they managed to figure out so much of what we are still using today, its only in quite recent times we have started to get DC interconnects between countries and inverter drives for motor control (they don't have the mesmerising beauty of mercury arc recifiers though....)

 

[Geoff1946]

Just asking!
If the lights were on different phases would this affect the frequency and possibly amplify the emc?

I'm not sure about radiated interference, but it is certainly an issue for conducted third harmonics, which add together in the neutral to the extent that the neutral cable can be overloaded and lighting can be visually affected. In one instance I was aware of fluorescent lighting displayed a very noticeable "swirling" effect. The cause was numerous computers in the room, powered from the same panel as the lights.

 

[johnb2713]

Probably not, it sounds like the noise is comming from the later stages of the switch mode power supply cirucits, after its already been converted to DC, but as the lights will not be in sync, you could find that the effcetive frequency is dependant on number of units in close enough vicinity with the peaks comming from different units

Switch mode units can also result in harmonic distoration of the mains though when units spit across phases have a current that comes in a burst at the voltage peak, and instead of balancing against one enough as in a sinusisol current draw, the harmoic part of the RMS currents actually add in the neutral rather than cancel! as you have three sets of current peaks occuring at separate times and you start to get 150hz components from 3x50 (as well as higher multiples of 3x50 as well).

AC theory is something that can get a lot more complicated than the couple of afternoons in college that most of us probably struggle to remember after a couple of decades on the tools (and most of the current crop probably sleep through it, if it hasn't really been dumbed down already ) Maybe thats a tad unfair - I'm just semi-amazed that that back in the later 1800s they managed to figure out so much of what we are still using today, its only in quite recent times we have started to get DC interconnects between countries and inverter drives for motor control (they don't have the mesmerising beauty of mercury arc recifiers though....)

There's a blast from my past Mercury Arc Rectifiers. During my days in the steelworks. I remember the little arm moving down to strike an arc in the pool of mercury and then drawing the arc to get it running. Real Frankenstein stuff.

 

[Dyslexicbloke]

You might try some bypass capacitors, close to the drivers, which you could also screen with foil.
It is hard to say what values you may want but pretty small to avoid inrush on the DC side whilst dumping what sounds like RF eipple over the DC. Perhaps also try a small choke. You are trying to flatten the DC so not comon mode.
You should put the scope on the DC to th LED's... You will likely find loads of HF noise, dump that and all should be well.
I wouldnt recomend a standard EMC filter though... That will likely dump the HF to ground which will play havock with an already poor Buck circuit, at least I am assuming that is the issue.
Try something between 0.1 and 0.47 uF ceramic in paralell with a modest aditional electrolytic, perhaps 330 or 470 uF.
That shouldnt stress the driver but will get rid of a load of HF. Honestly just play, and or simulte.

If you are directly measuring do it in isoltion, just the one driver under test, or the noise from the others will make life hard.
Any long cablen runs, AC or DC, try to twist. Rember the DC is probably the bigest issue but noise on the AC, due to none synchronious switching, can also be a an issue. Bypass caps on the AC will work but watch the currents, hrmonics can do odd stuff.
.
Poor AC/DC converters are a pain... Just ask any PLC IO card!

 

[mort2376]

You might try some bypass capacitors, close to the drivers, which you could also screen with foil.
It is hard to say what values you may want but pretty small to avoid inrush on the DC side whilst dumping what sounds like RF eipple over the DC. Perhaps also try a small choke. You are trying to flatten the DC so not comon mode.
You should put the scope on the DC to th LED's... You will likely find loads of HF noise, dump that and all should be well.
I wouldnt recomend a standard EMC filter though... That will likely dump the HF to ground which will play havock with an already poor Buck circuit, at least I am assuming that is the issue.
Try something between 0.1 and 0.47 uF ceramic in paralell with a modest aditional electrolytic, perhaps 330 or 470 uF.
That shouldnt stress the driver but will get rid of a load of HF. Honestly just play, and or simulte.

If you are directly measuring do it in isoltion, just the one driver under test, or the noise from the others will make life hard.
Any long cablen runs, AC or DC, try to twist. Rember the DC is probably the bigest issue but noise on the AC, due to none synchronious switching, can also be a an issue. Bypass caps on the AC will work but watch the currents, hrmonics can do odd stuff.
.
Poor AC/DC converters are a pain... Just ask any PLC IO card!

I now feel like it's Monday morning and I'm back at college on day release.

I'm sure what your saying makes perfect sense and would solve the problem, but my brain can't cope with that much information this early.

I now need to lie down in a dark room with lots of coffee.

 

[apprentice87]

Sounds like the usual carp light drivers with a CE mark but which if ever tested, would no doubt fail the EMC requrements.

I get the impression good design and testing has gone, in favour of just put a sticker on it, nobody will know.

Years ago i used to work for a firm that made ranges for chip shops. They had to be CE marked. All the firm did, was to have the one model tested, then stuck the sticker on everything they made, and these were items that back 30 years ago cost about £20,000..

As you say, fake stickers just stuck on..

john..

 

[UNG]

Have you checked for any mains harmonics

 

[johnb2713]

You might try some bypass capacitors, close to the drivers, which you could also screen with foil.
It is hard to say what values you may want but pretty small to avoid inrush on the DC side whilst dumping what sounds like RF eipple over the DC. Perhaps also try a small choke. You are trying to flatten the DC so not comon mode.
You should put the scope on the DC to th LED's... You will likely find loads of HF noise, dump that and all should be well.
I wouldnt recomend a standard EMC filter though... That will likely dump the HF to ground which will play havock with an already poor Buck circuit, at least I am assuming that is the issue.
Try something between 0.1 and 0.47 uF ceramic in paralell with a modest aditional electrolytic, perhaps 330 or 470 uF.
That shouldnt stress the driver but will get rid of a load of HF. Honestly just play, and or simulte.

If you are directly measuring do it in isoltion, just the one driver under test, or the noise from the others will make life hard.
Any long cablen runs, AC or DC, try to twist. Rember the DC is probably the bigest issue but noise on the AC, due to none synchronious switching, can also be a an issue. Bypass caps on the AC will work but watch the currents, hrmonics can do odd stuff.
.
Poor AC/DC converters are a pain... Just ask any PLC IO card!

Thanks for the info, I'll try and have a look but theres 45 lights in this ceiling, I will need to make a plug in filter for them.

J

 

[johnb2713]

Have you checked for any mains harmonics

Ive scoped the mains and it all looks ok including with the inverters for aircon and laser cutter running.