Metal Halide CWA EMC issues

[stolzie]

Hi Everyone,

I am looking for some advice on a problem that I am currently having with some metal Halide Constant Wattage Auto transformers (CWA) 240VAC (1000W) and the EMC issues that I seem to be having when I turn them on with relay. The system implements 4 metal halide lamps that are run from a generator. There is an LCD to indicate the Voltage being supplied from the generator and when the lamps are fired and such. This board is run from the battery (12V) which starts the generator.

The problem that I am getting is that the LCD will get scrambled when the CWA are switched on. The scrambling does not seem to happen all the time but it does happen fairly often. I have checked all the wiring to the CWA and they are all consistent. I can isolate the problem by switching the problem Circuit breakers off on the particular CWA and not power the unit.

The relay is being controlled via an MCU which is also running an LCD. The relay contacts switches the Active wire (240V) to CWA.

I believe I need to implement some sort of snubber across the relay contacts in order to resolve the issue but lack the understanding of how it all works and how they would be implemented.

Any feedback would be appreciated.

Thanks,
Jason

 

[ScottyUK]

Check whether there is a sag in your supply voltage as the MH lamp starts up. My guess is you have a supply problem. 4kW on 12V is a lot of current, so managing wiring resistance is critical to controlling volt-drop.

 

[itsmoked]

Scotty I believe only the control is done on the 12V stuff not the lamps. They're fed 240Vac.

stolzie; Welcome to Eng-tips! It's nice to have clear questions asked, as you've done.

Typically the contactor/relay coil kick-back is what tilts the microcontrollers or displays. That's when the coil current is interrupted and there is no kick-back "free wheeling" diode installed or it's failed. But that is typically on contactor release.

If it is the ballast and only occurs on initial turn-on then a snubber might work fine for you.
[URL unfurl="true"]http://www.alliedelec.com/red-lion-controls-snub0000/70030263/[/url]

Keith Cress
kcress -

http://www.flaminsystems.com

 

[stolzie]

Scotty I believe only the control is done on the 12V stuff not the lamps. They're fed 240Vac.

stolzie; Welcome to Eng-tips! It's nice to have clear questions asked, as you've done.

Typically the contactor/relay coil kick-back is what tilts the microcontrollers or displays. That's when the coil current is interrupted and there is no kick-back "free wheeling" diode installed or it's failed. But that is typically on contactor release.

If it is the ballast and only occurs on initial turn-on then a snubber might work fine for you.

http://www.alliedelec.com/red-lion-controls-snub0000/70030263/

Thanks for the replies and Thanks for the welcome Keith! The only way to get the answers required is to ask nice clear questions or else we would be wasting every ones time.

Yes the control is only handled by the 12V and the ballasts/Lamps are fed by 240VAC generated from the generator. As I mentioned it only seems to occurs on initital fire up of the ballast and it does not seem to happen all the time. I was suspecting some sort of wiring issue in the ballast itself but that seems to have checked out OK. And not all ballast perform the same I have other units that can be stopped and started and it never glitches.

I am not use to handle this 240V stuff, and it sort of scares me in a way so I have been doing a lot of reading to get things done right. I gather that putting the RC Snubber across the load is the preferred option rather than across the relay contacts as I have described?

Thanks,
Jason

 

[itsmoked]

If only everyone provided solid info with their questions..

Sometimes the display should be grounded and it's not or shouldn't be and it is, making it more susceptible to noise upset.

Yes, across the load(ballast) not across the contacts. Snubbers always pass current. If you put them across the contacts you will always have power supplied to the ballast. Not enough to operate it but enough to light someone up not expecting it and enough to cause some ballasts to act strangely, like blinking or making noise.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[VEBill]

If the controller is powered from a battery, then the PS voltage should be stable except on generator start-up. Lamp start up shouldn't affect the battery voltage.

Is there any low voltage (control or sensing) wiring running parallel to the load wiring?

 

[MacGyverS2000]

I'm going to take a stab and say you have some serious noise on the LCD board control/power lines... scope them and see what they look like.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[stolzie]

Yes, across the load(ballast) not across the contacts. Snubbers always pass current. If you put them across the contacts you will always have power supplied to the ballast. Not enough to operate it but enough to light someone up not expecting it and enough to cause some ballasts to act strangely, like blinking or making noise.

Ok sweet, will look at grabbing some of these and placing them across the loads(ballast) and see what they achieve.

If the controller is powered from a battery, then the PS voltage should be stable except on generator start-up. Lamp start up shouldn't affect the battery voltage.

Is there any low voltage (control or sensing) wiring running parallel to the load wiring?

The wiring of the unit is pretty much looking like spaghetti. I have been trying to tidy up and isolate the 240V and low voltage wires to overcome any noise that can be introduced to the low voltage side. There are pretty much wires running all over the place which is not a good start.

I'm going to take a stab and say you have some serious noise on the LCD board control/power lines... scope them and see what they look like.

Yes you are right it is a pretty noisy environment. I have implemented a number of different solutions to overcome the noisy environment so far and I figured the snubbers may need to be added as well hence the question, I have already implemented things like sheildeding IDC Cables, 1k pullups on I2C lines, increasing Cap values to smooth out the Power rails.

Thanks for everyone's feedback it is greatly appreciated.

Thanks,
Jason

 

[Comcokid]

When the lamps power up, their cold resistance will be 1/5 to 1/10 their hot resistance, and the CWA may drop into saturation for a moment. This could be creating a huge current pulse which may be coupling across control wires/cabling harness (if they are bundled together, or even directly into your control board. If so a snubber will not fix the problem.

A noise pulse like this may be common on all, or most of the wires going to the control board, and may be disrupting it's operation. Try a clamp-on ferrite noise suppressor, and clamp it around the I/O wires going to the control board. If there are a couple bundles of wires going to the board, then use a ferrite on each bundle. Don't clamp it around both AC and control wires together, or you will just couple AC noise directly into the control wires.

I suggest this because usually it is quick and easy. You can get clamp-on ferrites at Radio-Shack (if still open), or some stereo/audio/computer shops.

 

[stolzie]

When the lamps power up, their cold resistance will be 1/5 to 1/10 their hot resistance, and the CWA may drop into saturation for a moment. This could be creating a huge current pulse which may be coupling across control wires/cabling harness (if they are bundled together, or even directly into your control board. If so a snubber will not fix the problem.

A noise pulse like this may be common on all, or most of the wires going to the control board, and may be disrupting it's operation. Try a clamp-on ferrite noise suppressor, and clamp it around the I/O wires going to the control board. If there are a couple bundles of wires going to the board, then use a ferrite on each bundle. Don't clamp it around both AC and control wires together, or you will just couple AC noise directly into the control wires.

I suggest this because usually it is quick and easy. You can get clamp-on ferrites at Radio-Shack (if still open), or some stereo/audio/computer shops.

Thanks for the input. It does seem as though there is a huge current pulse or something like that happening. I will give the ferrite noise suppressors a go as you have mentioned they are a quick and easy to install.

Thanks Again,
Jason