Hi all, I've recently run some tests and noticed an interesting phenomenon with my mercury vapor lamps.

When I originally built my 50 watt mercury vapor fixture, I ballasted it with an Advance L-140F-TP, which fed the lamp only 58% of its rated power. Despite this, the lamp always ran up all the way (90V), and it ran well. I later fitted it with a 50 watt metal halide ballast, which, after a current reduction modification I did, runs the lamp at about 92% of rated power (and it reaches a slightly higher 105V).

I've found that now that the lamp has become "used to" running at full power, it will no longer run up at reduced power, even on the same ballast that it spent the first many hours of its life on! On the Advance L-140F-TP, it now only runs up to 70V, and stays there, even when left for an hour or more. Same if I try it on my Advance R-140-1-TP, which feeds it 72% of rated power, which is actually very near specification, since it's rated to run at both 40 and 50 watts. I also tried my spare lamp, and despite only running a total of 15 minutes on the full-power ballast, it will no longer run up all the way at reduced power, even though it did so fine before being introduced to the full-power ballast. If I throw either lamp back on the full-power ballast after these tests, they have no problem running up all the way.

I've also found that when running on the full-power ballast, if I reduce power "on-the-fly", they will remain fully run-up and at peak arc voltage.

Anyone know the science behind why my lamps have become "spoiled", and will no longer run up on reduced power after being introduced to a full-power ballast? 

When operated in the saturated vapor mode, the setup is rather thermally unstable.
The thing is, the arc voltage increases there very steeply with the mercury pressure, or said in the opposite way, lowering the temperature dramatically reduces the arc voltage drop.
Now the ballast is a constant current source, so the output current is rather independent on the actual arc voltage.
That mean the power delivered to the arc is directly proportional to the actual arc voltage.
That means when you supply the lamp with low current, the power created at the rather low arc voltage is insufficient to vaporize more mercury, so the lamp get "locked" in this state.
But when you alter the balance only slightly, so e.g. increase the current, decrease the thermal losses or increase the cathode drop a bit, the thermally induced feedback may become positive, so the setup starts to "runaway": The elevated power rise the temperature, this evaporates a bit more mercury, increases the pressure, hence the arc voltage, that increases the power  further,... This runaway process is then stopped at the point, when all the mercury gets evaporated, so when the increase in temperature can not lead to pressure increase anymore. That is the point, when the lamp reaches the unsaturated vapor state.

Well, all the above you probably already know and at first glance, by your judgement, it does not explain, why the lamp did warmed up correctly before, but now it doesn't.
My guess, you are assuming the lamp parameters do not change over time signioficantly.

Well, they do not by itself, but you should not forget the extreme sensitivity of the "runaway" thermal feedback there, making it extremely sensitive to anything. So then even the tiny changes could have large consequences.
Here few mechanisms could play a role:
I would expect the electrodes went up through the complete burn-in, causing the cathode drop to become lower, so redsuce the dissipation there.
Or the end mirror coating had degraded a bit, loosing more heat.
Generally with HID's, the first 100..200hours is a kind of "burn-in" time, so in fact a period, when still the manufacturing process get finished and only when the lamp reaches its rated "initial" parameters (lumen output, arc voltage,...)

Generally don't forget with the 40 or 50W MV's, the manufacturers already struggle a lot to make sure it "tips over" to the unsaturated vapor state reliably all the lifetime even on the rated ballast, so I'm not surprised to see even the slightly lower power levels become so unreliable.

Once I tried to run 50W MV lamps using a 40W T12 fluorescent magnetic ballast.
I tried two identical lamps. Both lamps ran well and gave the expected light output, only during the first hour after every start there was a sort of occasional "pulsating" (flickering in the 25Hz (?) frequency) - one of them did it more visible.
I tested those two identical lamps with the same ballast (the same piece) and one lamp had slightly lower than suggested arc voltage - should be: 81V to 109V, was: about 76V, while the other lamp surprisingly had higher than suggested arc voltage: up to 118V! That surprised me a lot as I expected both lamps to be underdriven.

So it's probably as Medved said, it was a "burn-in" period when the lamp parameters aren't stabilised yet.

@merc: Two lamps are never exactly the same. In your case once the lamp "tipped over" to the unsaturated vapor state and then run at the 118V, the other case it didn't.

The 118V (instead of the rated 105V) was there due to two reasons: Thinner (lower current) arc mean the ionization had decayed way more during the mains zero cross, so it resulted into higher reignition spikes. Second effect increasing the arc voltage is the lower current providing less heat to the electrodes. That means they run a bit colder, so with higher cathode fall drop (to recover the power dissipation there). So this extra cathode drop then manifest itself as an elevated arc voltage (mainly in the moments, when the current is on it's peak level).
Either way, this elevated voltage at lower curtrents is pretty normal for nearly all discharges, where the heat for the cathode emission is extracted from the cathode fall (so most hot cathode lamps)

Thanks for the info; I hadn't considered changes due to the burn-in period.

Thanks for the info; I hadn't considered changes due to the burn-in period.

Not only burn-in, even normal aging does shiftparameters over time...

I've been playing around with this again, and I found something interesting. I again hooked the lamp up to my Advance R-140-1-TP, and to my surprise, the lamp ran all the way up to 110V this time. This is with the lamp oriented horizontally, and I found that if I tip the lamp into a base-down position while it's running, the lamp visibly dims, and arc voltage drops to about 75V, in the span of about 2 minutes. If I tilt it back to horizontal, it quickly brightens again, and arc voltage raises back to normal. If I tilt it base-up, arc voltage holds a bit better, only dropping to 90V.

Do MV lamps simply not like running base-down? I'm going to do some more testing and see how the arc voltage holds when running at full power base-down.

Edit: On a full-power ballast, the lamp runs up to 90V base-down, but it takes a VERY long time to do so.

The coldest spot are the seals, while as the lead wires carry the heat away and the cap side has the starting probe connected, it is that part, which run the coldest. On top of this difference the botom side of anything hot (so include the arctube), would be cooled the most.

So when the end with the starting probe is the lowest point, it will become so cold, the mercury starts to condense there, so the pressure goes down.
With the base up position the seal will run a bit hotter, because the auxiliary probe either isn't there at all, or is not connected (the connection does carry not only the electroity, but the heat as well), so although it will be most lilkely still the coldest point on the arctube, it will be a bit warmer than the starting end in the previous case.
In horizontal position the seals will be still the coldest points, but because they are not the bottom one, their temperature will be higher than the cold spot in the previous cases. And on the top of this, when the mercury condenses there, it will flow down, that means towards the center of the arctube, but there the quartz is hotter, so it will evaporate again. And that will keep the (saturated vapor) pressure the highest from all of the three positions.

Once the lamp is able to easily keep all the mercury evaporated (e.g. on a full power ballast), it isn't position sensitive anymore, as the pressure then depends mostly on the temperature in the majority of the arctube volume (it is just a gas like each other).


The lamp is on purpose designed so, these differences are not too large, but if you operate the lamp "on the border" with the feed power, it starts to play quite significant role, given the rather unstable manner of the saturated vapor operation.


By the way the base down position is dangerous from another aspect: When switched OFF, the condensed mercury could flow there. And if it does form too large droplet, it may short out the main and auxiliary probes and by that way prevent the ignittion. Therefore some makers either install starting probes on both ends (Iwasaki,...), or exclude the base down position.
The starting probe is on the base side for two reasons:
First it is technically easier, as connections of both termnals are available there.
And second, most fixtures operate the lamps either base up, or horizontally  (so then the starting probe can not be flooded)

By the way the base down position is dangerous from another aspect: When switched OFF, the condensed mercury could flow there. And if it does form too large droplet, it may short out the main and auxiliary probes and by that way prevent the ignittion. Therefore some makers either install starting probes on both ends (Iwasaki,...), or exclude the base down position.
The starting probe is on the base side for two reasons:
First it is technically easier, as connections of both termnals are available there.
And second, most fixtures operate the lamps either base up, or horizontally  (so then the starting probe can not be flooded)

Many years ago I've had exactly this problem with a 125W Osram HQL deluxe. When the lamp was operated base down, after a cool down it doesn't restrike anymore. To restart the lamp I had to shake it. After many attempts I've found that the problem was only base up, horizontal and base up didn't had any problem. With an another lamp(philips HPL comfort) there wasn't any problem.

Thank you Medved, that makes sense. I've relayed this info to TheUniversalDave1, since he's measured only ~75V across the lamps on his home made fixture, which runs them base-down.