I have 70w MH spotlight that has worked well 6hrs per day untill few days ago it started acting weird. The lamp lights okay but after 2-4hrs the fixture starts making loud electric noise. The lamp stays lit but from time to time it look likes the light is slightly vibrating like rectifying or something. The fixture has magnetic ballast and is quite compact and thus runs very hot.

Any ideas what could cause this? The MH lamp type is CDM-T

A rectifying lamp could do that.
Try to measure the DC current component: Hook a digital A-meter in series (could be the mains input, ac the capacitor wont influence the DC reading and it allows the fixture to be fully assembled in its normal positio - important, because the lamps use to be position and heat sensitive), fire up the lamp, set the meter to DC and observe the meter short after power up, after initial warmup and then once it starts acting.
If a significand DC appears (with 70Watter about 100..200mA or more), the lamp is definitely the culprit.
Sometimes swapping the lamp polarity may help (there may be some heat assymetry, the ignitor may be generating one polarity pulses higher so the lamp may prefer one polarity for ignition,...), youmy try torotate the lamp (so force the pool elsewhere in the tube).

Or if there is no DC component and yet the ballast is making wetrd noises, look what got loose there. The eventual movements of parts forming the magnetic circuit may modulate the inductance and so cause the lamp current so the light to flicker... If something loose, tighten or glue it with some epoxy (the material soaked between the winding and core plates may deteriorate due to heat and so get loose).

Other cause could be some contact getting loose and arcing. This may actually sometimes manifest itself by rectification as well..

Thanks. I have to see if I can measure the dc current as I only have normal multi meter that can't withstand currents very long. I have to disassembly the fixture in next week to see how the internals looks.  It probably wont hurt to try change the lamp as well although the current one hasn't give any indications of going eol.

There should be no problem to measure <0.5A (the 70W light with PFC) continuously with a "10A for 15sec" multimeter. The other problem is, the readings of cheap multimeters like DT830's in such applications are unreliable and meaningless

There should be no problem to measure <0.5A (the 70W light with PFC) continuously with a "10A for 15sec" multimeter. The other problem is, the readings of cheap multimeters like DT830's in such applications are unreliable and meaningless

The DC component is in fact the only type all meters inherently read very predictably and correctly, even with large AC superimposed on it. It is just because the DC all needed is a low pass filter.
The AC is the troublemaker if the shape is any different than a sinewave.

Here you need to make sure you do not overload the range, so the 10A range should be OK.

Forthe measurement I would start before touching the fixture itself, if it is something as a bad contact, just moving it may cause the problemto temporarily disappear and then good luck in finding the real cause when you have time to mess up with the instruments...

I measured it and during normal operation there is no dc current. When the problem appears dc current is 0.02A. At this stage the light still appears flicker free but the fixture makes really loud noise. From time to time the sound becomes even more louder and you can see the light starting to flicker and then the A meter shows 0.1A DC.

It also looks like the problem is getting worse.

And this was when the fixture was in it's normal operation position (base up +/- 20°) and I didnt change anything before measuring.

100mA is about 10% of the rms current rating, that could cause partial saturation so the noise. But still it should not cause any significant overheating yet.

What lamp type are you using? Is it possible to change its position? The problem seems to be thermally sensitive (need full thorough warmup to occur, but that could be just the warmer ballast core having lower saturation flux, so yielding stronger magnetostriction and/or magnetic leakage), so changing the lamp position may help (it is just on the border)

In any way, just to be sure, check all the connections within the fixture, if there isn't something loose, but I won't expect that. The 10% is quite consistent with your description (barely observable flicker, just the ballast buzzing).


Maybe some loose piece of something magnetic may be rattling there, but I guess you have already checked that.

I took the fixture apart and there wasn't any loose connections or other stuff loose. Nothing shows overheating and the Vossloh Schwabe NaHJ 70.158 ballast says it has thermal cut-out. Ignitor is VS Z 250 K A20.

Here is a photo of the lamp in use. It's Philips CDM-T 70W/830 with datecode 1L. The burning position is almost base up.


http://www.whitenightcape.com/kuv/CDM-T.jpg

So just try to reverse the lamp.
The 0.1A DC wont cause any overheating yet, just a weaksaturation so noise (the ballast is designed to handle that, just nobody claims it will remain silent).
With the warmup I did not mean any overheating or so, just warming up to the normal operating temperature, it uses to be in the 90..120degC range, so it takes a while for the bulk of iron and copper to warm up and show all eventual heat related effects...

I put it all together but decided to replace the lamp anyway since it was really easy at this point. It now has Osram HCI-T 70w/830 WDL. It's been working few days without any problems so I guess the lamp was to blame. Thank you for the advices Medved, my plant can continue enjoying lots of light.

Now the same fixture is malfunctioning again. Seems like changing the lamp didn't fix the problem.