No thallium appears when I see the spectrum of this lamp run-up. Also, the yellow doublet and blue line of mercury are also strong.
Also: The lamp would be greenish if thallium would be present, and it isn't the case.

How do you know this? Tl 535nm is rather close to Hg 546nm, hard to tell on uncalibrated spectrum. One way to tell is to look at run-up. At start, only four Hg lines present, as the burner heats, nearby Tl green line appears, and later usually over-dominates green Hg line.

Or something stuck around the impeller 🤔

Actually, some displacement pump like rotary vane may be in use! Why? Turbines struggle to get really good vacuum and do this with poor efficiency, especially small ones. Look at a regular vacuum cleaner motor-turbine. Rather complicated mechanics here, including multiple turbine wheels in series. And it eats up to 2kW to get the job done satisfactory, modern 'green' models suck literally))

There is no thallium in the spectrum. Only mercury and dysprosium. The green line is of mercury.

I forgot to say XMH lamps can work on 35W CMH electronic ballast.
Or I am confused and the opposite is true that 35W CMH lamps can work on XMH ballasts?

So you wanted a 6500K lamp and Chinese bro's supplied you with Dy-Tl lamp accordingly, okay. ) Tl line is always very prominent in these, as it is essentially monochromatic even at high arc loadings. It may look bright on the spectrum, but overall energy is limited as the width is still very narrow.


If checked with a calibrated spectrometer the lamp will show the color on CIE1931 XY space goes too much high to the green, we may speculate the lamp is improperly dosed or runs too cold (due to a leak to outer bulb, just badly made etc).

Well, generic 35(39)W MH lamp is specified assuming sinewave current at line frequency. It is a little known that MH lamps need some ~10% less current at the same power when running on squarewave electronic ballast due to power factor now essentially becoming 1 instead of ~0.9 when on line frequency sinewave.

Automotive lamp was for sure never meant to be run on the ballast choke, so may be Dor was not that much off the truth... )

So, I don’t know enough about the lamp part of your question, but for the Ballast part, maybe I could help soon. I have a 250 watt Hps CWA ballast with a cap. Maybe I could measure the voltages from the ballast.

Trust me, that is not what is happening. Why wouldn't you use pistons? Because those are low volume, high vacuum pumps that also cause a ton of vibration. Also way way too expensive to put into a 50 to 500 dollar device.

It is more likely that it has worn out bearings making it noisy, or have a bit of debris stuck to the fan throwing it off balance. Other things that can make it sound differently are BLDC motors instead of series wound universal motors, the former have a more constant motor speed, and the absence of commutator whine in a BLDC motor.