Emi/rfi blocking problem, shielding

[rxdeath]

my scenario:
so we are working on a problem in a building and are struggling with making the top floor (cable and other assorted digital equipment) playing nice with the bottom floor warehouse (digital ballasts on the ceiling).

my question:
where does the interference 'come from'? light? ballast box? connecting cords? is there a (reasonably cost effective) way to take care of this? what meter can/should i buy to detect this problem? i read that most digital ballasts operate at 20-30 MHz but i assume that a practical solution would take care of the entire VH band, i haven't found any that that operate above that band.

what we've tried:
approx 10 feet of chicken wire as well as aluminum screen wrapped around the ballast
getting the wires for digital equipment and other devices with high impedence connection
as far as 15' from the ballasts.
sheilded cat5e solid core cable replaced all UTP cat5e cable

the only real difference I've noticed on anything is with the aluminum screen wrapped around the ballast and grounded the screen to nearby electrical conduit, the interference seemed to get better, but was still obviously present

thanks for what help anyone can provide, we've been struggling for about 3 weeks

 

[yungman]

What is your problem with the equipment, give some sort of description first why you think there is a problem...like what symptom?

The question first is to determine whether the interference are radiated from the air or travel in the power line. I don't know what is a ballast, but I worked quite a bit on EM suppression in CE testing.

Without looking more deeply into the environment and knowing more about the electrical circuit. I am just going to blind guess.

You mention CAT5 cables. Does the noise of the device disappeared when you disconnect the cable? If so we use ferrite clamps to clamp onto the cable to break the loop to avoid EM radiation:

http://www.digikey.com/product-search/en/filters/ferrite-cores-for-cables-and-wiring/3408554

If you have CAT5 cable running inside the wall, try clamping this on and see whether it improve or not.

As for aluminum foil wrapping the ballast, can you ground the foil close to the ground wire of the ballast?

Is it possible for you to draw the wiring diagram of your floor electrical wiring? If possible, the drawing of the ballast circuit also. It is really blind shooting if I don't know the circuit. I only suggested the ferrite clamp because it is very common practice, you'll see it comes on a lot of the USB and other inter connect cables. If somehow the grounding form a loop and acts like a loop antenna on the old tv, the ferrite clamp should break the loop.

Answer back.

 

[rxdeath]

i know there is a problem because the display on some digital equipment (temp/humidity monitors) tied into the building goes insane when the lights are on, as well as occasional complaints about problems with various things like wireless phones, wireless internet.

i am not sure if it travels through the air or the lines either.

the ballasts are lumatek 1000w ballasts, that run at 20MHz (high intensity lights for warehouse)

the cables all connect a system of devices to monitor the area, if i unplug them, i can't get a reading so it doesn't help to do that, but everything works fine when the lights are off, once they turn on humidty in the areas with the ballasts goes up to 100% then will go off the charts.

ferrite beads we thought about, but wasn't sure it would work. we would just get them for the cat5 cable? not for the power cords or lamp cords ? are you saying the cat5 is acting like an antennae and if i put some of these ferrite beads it will break the loop? what size and strength of bead do you recommend?

also are their meters that i can use to test for this that arent' too expensive?

 

[yungman]

i know there is a problem because the display on some digital equipment (temp/humidity monitors) tied into the building goes insane when the lights are on, as well as occasional complaints about problems with various things like wireless phones, wireless internet.
By insane, you mean the reading jumps? Disconnect the sensor connections one by one and see is there any particular one that affect the most. I know you lost reading when disconnect, but it should still jump if it is not from the connection.
i am not sure if it travels through the air or the lines either.

the ballasts are lumatek 1000w ballasts, that run at 20MHz (high intensity lights for warehouse)

the cables all connect a system of devices to monitor the area, if i unplug them, i can't get a reading so it doesn't help to do that, but everything works fine when the lights are off, once they turn on humidty in the areas with the ballasts goes up to 100% then will go off the charts.
As I said before, disconnect the sensor connection one by one and find the sore spot.
ferrite beads we thought about, but wasn't sure it would work. we would just get them for the cat5 cable? not for the power cords or lamp cords ? are you saying the cat5 is acting like an antennae and if i put some of these ferrite beads it will break the loop? what size and strength of bead do you recommend?
Ferrite works magic, you just have to put in the right place. I don't know your wiring circuit, so I only talked about the most obvious and easiest. If I can confirm that the noise is travel through the power lines, you bet I will suggest to put onto the power line.
also are their meters that i can use to test for this that arent' too expensive?

I don't know of any meters for this kind of work. We use an oscilloscope with a coax cable, at the end of the coax, we connect the inner wire to the outer shield to form a small loop about 1" diameter and use it as a loop antenna, then we use that to snoop around the wires to try to look for noise. Not very reliable though. If the noise is from the air, you'll see something on the scope. But my bet at this point is still on the power line.

How about the wiring diagram. If you have an accurate wiring diagram, that will help trouble shoot. It is very important to show the exact way you connect the ground. I want to find any accidental loop you create and break it. Ground has to be "star" out from one point if possible to avoid ground loop.

The sensor or inter equipment coax connection create a ground loop as you connect the two grounds together through the cable. That's where the ferrite clips work magic. You definitely need to buy a set of those clips. Try to loop a turn or two of the CAT5 cable into the clip to increase the effect of the ferrite clip.

 

[gnurf]

rxdeath,

Check out this not-so-glowing review of your ballasts (or very similar): http://www.w0qe.com/RF_Interference/grow_light_electronic_ballasts.html

From the link:

The amount of noise produced by these devices is larger than anything I have ever measured.

The authors measured a common-mode current of 7.8mA (!) on a Lumatek LK-1000. If your ballasts are similar then every power cord connected to them will effectively act as antennas and your whole floor is probably one large radiating mess.

Their solution is power line filters with common-mode chokes on in- and output. Unless you're a HAM or really want to, you don't have to build these yourself. Have a look at their spec and do a parametric search at digikey.com for 'power line filters' or something similar. Good luck.

 

[Carl Pugh]

This may be an insane idea, but
Try incandescent lights.
At high power, there may not be that much difference in operating cost.

 

[rxdeath]

thanks for all the replies

incandcants lights are not an option

gnurf:
so you're saying that with the information you learned about the ballast its not (exclusively) an over-the-air problem? I'm broadcasting from the ballast itself as well as all the power lines going in/out of it?

the ferrite beads/torroids on both of ends of every power cord as well as cat5 cords would probably be enough to kill the problem?

do you know of any meter i can use to test progress in fixing the problem?

thanks again

 

[yungman]

One thing I learn from working with EMI suppression is don't asume. If you don't want to draw out the wiring, try the ferrite on the CAT5 first.

 

[the_emi_guy]

do you know of any meter i can use to test progress in fixing the problem?

You will need a spectrum analyzer and a preamp. The old HP 8591 spectrum analyzer and 8447 preamp were standard equipment in EMC labs for many years. (I can tell from the screenshots that is what w0qe was using). I just checked ebay and an 8591 in working condition is still going for $2000. You will be measuring common mode current on your mains cables, high freq cable between ballast and lamp, and on your data cables. For 10MHz up to 500MHz you can make a common mode probe easily. Below 10MHz you may have to buy (or build) a LISN. Unfortunately, there really is not an inexpensive way to actually view this stuff. You can shop for an inexpensive spectrum analyzer, there are some that work with your PC, but you will need to check their dynamic range and DANL.

BTW, these digital ballasts run in the 10s of KHz (not MHz). But, depending on their design, create lots of harmonic energy well into the MHz.

Couple of other things (you may have already considered):
If you were not the lighting installer, you may want to download literature from Lumatek and make sure they were installed properly and are using the correct bulbs. In particular make sure the output cable between ballast and bulb is shielded, and the shield properly bonded at each end.

Shielded RJ45's will not be effective unless the cable shield bonds to chassis gnd at each end of the link. Not all RJ54 receptacles provide chassis gnd connection.

Good luck.

 

[yungman]

I don't have the heart to suggest the spectrum analyzer! That's so expensive. But yes, this is what is needed.

 

[rxdeath]

emi_guy

well played i double checked and it is 20,000 Hz, my bad. so without buying thosands of dollars of equipment what would you do to try and take care of the probem in a step by step fashion? sounds like i need some sort cage as well as the line input filter because the whole wiring system is a giant antennae... also I'm not sure how to tell if its sheilded lamp cable or not? can i use some cu tape?

sorry for all the random questions but i have no idea what I'm up against here and even if i could get all this fancy equipment i wouldnt' know what to do once i found out whatever it is that tells me. I'm seriously thinking of getting into amateur radio after this because i should know this stuff as a semi-dedicated nerd.

 

[gnurf]

so you're saying that with the information you learned about the ballast its not (exclusively) an over-the-air problem? I'm broadcasting from the ballast itself as well as all the power lines going in/out of it?

The radiation would be primarily from the cables, and not from the ballast itself. If you have a conducted noise problem as well is anyones guess and even if you provided a wiring diagram I'm not certain you could diagnose the problem with the help of a forum like this (not with my help anyway).

sounds like i need some sort cage as well as the line input filter because the whole wiring system is a giant antennae... also I'm not sure how to tell if its sheilded lamp cable or not? can i use some cu tape?

If you design (or buy) your power-line filters correctly, you'll reduce the troubling common-mode currents and thus the radiation. I think it's easier to tackle the noise at the source, instead of a "band-aid" approach with shielding. You might have to do both but do them in the right order!