I have a box of surplus CDM400/V/O/PS/4K/ALTO lamps. Several arc tubes have cracked and I'm trying to find out why.

I've been using different electronic ballasts, each rated for 400W, and all verified as drawing ~400W with a power meter in operation. My questions are as follows:

1) ANSI code on the lamp sleeve says M172/O. What does the O denote?
2) M128 M135 M155 M172 all appear to be 400W pulse start metal halide ballasts. WHY are there so many? What is the difference between them?
3) yesterday I discovered this datasheet https://www.1000bulbs.com/pdf/Philips%20MasterColor%20protected%20pulse%20start%20cmh%20lamps.pdf which states "Operates on metal halide pulse start ballasts (not suitable for electronic ballasts). This would seem to explain why mine are failing, but at least one of the electronic ballasts I've been using is labeled as a M135. Is the standard NOT a standard, regardless of topology?
4) This is the first time I've ever heard of a lamp as being "not suitable for electronic ballast". What is the mechanical/physical/electrical reason for this?
5) Tied to (2) above, if indeed these do require core/coil ballasts, M172 is rather rare. Can I substitute a M135 or M155?
6) perhaps a simpler explanation is I have a bad batch. How likely is this?

Ad 2)

I know there are few flavors of 400W pulse start MH, incompatible with each other:
- 3.15A rated, made along the MV specification (so to run on European 400W MV ballast plus an ignitor)
- 3.8A (or something around - don't remember the number exctly), designed around European 400W HPS specs

So when you operate the "3.8A lamp" on the 3.15A ballast, the lamp get actually 360W and will run underdriven, so its color may suffer.
There are even many lamp types rated for both ballasts, with different light properties depending on the ballast type it is used with.

When you operate the "3.15A lamp" on a 3.8A ballast, the lamp will be overdriven and fail soon.

Now what is strange on the "not suitable for electronic ballasts": These use to operate at really constant power, so a 400W ballast will feed 400W into the lamp regardless what lamp type it is.

There is only one explanation which came to my mind as I was typing the above:
These lamps are actually designed for lower power (300..350W instead of 400), by lowering their arc voltage and are designed to be fed with the given current. The magnetic ballast is nearly a constant current source, so its output power does drop when the load voltage goes down. But the electronic boosts the current so the power remains the programmed 400W, so in fact is overloading the lamps.

There is only one explanation which came to my mind as I was typing the above:
These lamps are actually designed for lower power (300..350W instead of 400), by lowering their arc voltage and are designed to be fed with the given current. The magnetic ballast is nearly a constant current source, so its output power does drop when the load voltage goes down. But the electronic boosts the current so the power remains the programmed 400W, so in fact is overloading the lamps.

Electronic ballasts boosts the current only during the run-up. At full output they operates at the correct current for the lamp.

Electronic ballasts boosts the current only during the run-up. At full output they operates at the correct current for the lamp.

No, standard electronic ballasts operate at constant power (aka measure the voltage and adopt the current so their product is constant).
This is the way they maintain the excellent color stability: Constant power means the positive thermal feedback present with standard ballasts (higher arc voltage yields higher power, that yields higher temperature, that leads to higher pressure and this increases the voltage further) is broken, so does not amplify the effects of gradually blackening arctubes.

Mainly for low power lamps (20W and some 35W), the ballasts go even further: Adjust the power so the voltage is kept constant. That means they stop the blackening (so absorbing more heat) from influencing the arctube temperature whatsoever by reducing the delivered power. But this functionality really requires three loops:
Most inner one regulates the current. This has very fast response time, in 100's of us.
Around this is other one, monitoring the voltage adjusts the commanded current for the inner one so the power corresponds to the required value. This has response time in 10's of ms. And it has limits on the current: Generally does not allow the current to exceed 150% of the rating and trips a "EOL fault" when the required current drops below about 50% of the current rating.
And around this is another, monitoring the arctube temperature (by measuring the arc voltage) and adjust the power so to keep the temperature (so voltage) constant. This one has to have time constant in 10's of seconds, maybe even a minute or so (way slower than the time constant of electrode temperature). Because of this long time constant, this one is possible to implement only in a digital way (analog implementation would be too drifty), so need microcontroller for that.

Standard electronic ballasts have only the two inner loops, current and power regulating ones.

Well, for short time deviation (till the power regulating one responds) indeed, the current is constant. But long time the system works on a constant power.


Automotive ballasts have other feature: They boost the power on cold lamp (to overcome the lower efficacy of pure Xe discharge when the halides haven't evaporated yet), this feature indeed isn't used on general lighting...

These lamps are actually designed for lower power (300..350W instead of 400)

I'm aware of 320W lamps designed for operation in a 400W magnetic ballasts. These have 320 in the order code, however.

Right in the datasheet, the "nominal watts" for this lamp is 400W. "CDM400/V/O/PS/4K/ALTO" The first number in the order code always corresponds with nominal lamp wattage in my experience.

I found a deal on ebay for a (magnetic) 400W pulse-start fixture at $50 shipped, so I'll give that a go. I've already killed 4, what's one more?

An update if anyone cares - a lamp ran in the core+coil M135/M155 for 40 minutes drawing 420W without flickering or cracking. For comparison, they all flickered and all failed in 3-10 minutes with electronic ballasts.

The only difference between the M135 electronic and the M135 magnetic is the operating frequency, so presumably the tube has some strange behavior with high frequency. I've had no problem running my 50, 70, 100, and 150W ceramics on electronic ballasts. Maybe the arc tube doesn't scale as well, which would explain why ceramic halides are rare above 150W, and even more rare above 400W. I dunno.

An update if anyone cares - a lamp ran in the core+coil M135/M155 for 40 minutes drawing 420W without flickering or cracking. For comparison, they all flickered and all failed in 3-10 minutes with electronic ballasts.

How did you measured the "420W"? It was the ballast input power? If so, assuming the 15% ballast losses would mean lamp is actually fed by 360W.

The only difference between the M135 electronic and the M135 magnetic is the operating frequency, so presumably the tube has some strange behavior with high frequency. I've had no problem running my 50, 70, 100, and 150W ceramics on electronic ballasts. Maybe the arc tube doesn't scale as well, which would explain why ceramic halides are rare above 150W, and even more rare above 400W. I dunno.

Electronic ballasts feed the lamps by a square wave, aka DC with polarity changed every few ms. So steady magnetic forces, so no vibration whatsoever, so no chance to excite any strong standing waves (as is the case with cheap HPS electronic HF ballasts; counting on the fact the HPS have quite a margin towards material limits along the tube itself and no components corrosive towards the alumina in the arctube mix). I very highly doubt this could be of any problem at all.

How did you measured the "420W"? It was the ballast input power? If so, assuming the 15% ballast losses would mean lamp is actually fed by 360W

Although a bit off, fair point. The ballast says 88% efficient on the side, so 370W presumably. With the included lamp, it draws a mere 408W, or 360W into the lamp.

But If you're suggesting putting 400W into a 400W lamp is overdriving it by 8%, that would mean a remarkably small margin for failure, and it would mean all of the specification sheets I've read are wrong: that "nominal lamp watts" should be read as "if you operate this lamp at nominal watts, it will explode". It also doesn't explain why all the other 400W lamps I have don't explode in a 400W electronic ballast.

Although a bit off, fair point. The ballast says 88% efficient on the side, so 370W presumably. With the included lamp, it draws a mere 408W, or 360W into the lamp.

But If you're suggesting putting 400W into a 400W lamp is overdriving it by 8%, that would mean a remarkably small margin for failure, and it would mean all of the specification sheets I've read are wrong: that "nominal lamp watts" should be read as "if you operate this lamp at nominal watts, it will explode". It also doesn't explain why all the other 400W lamps I have don't explode in a 400W electronic ballast.

The tolerance window for lamp operating at good performance (color, efficacy) vs reliability becomes very narrow when you are pushing the limits.
It depends how the lamp is designed: It could be designed to tolerate the 400W and get the 400W with both the magnetic, as well as electronic ballast. Assume this is the reference lamp for that system.
New lamp technologies yield higher efficacy.
One lamp could be designed with the same power but higher light output. This helps with new installation (less fixtures), but does not save power in existing installation.
Other lamp may be optimized to give the same ligt output as the reference so perform well in the existing installation, but with a reduced power. The only way for that on a CWA is to reduce the arc voltage (the CWA is a constant current, not constant power source). But on a real constant power ballast the power stays the same as the ballast is programmed for.
You may design this lamp so it still can tolerate the original power, but it cost you lower efficacy at the intended 360W.
Now if you want to reduce the power as much as possible, you can not have there the margin to tolerate the extra power anymore So such lamp gets ruined when fed with the "reference" power.

8% does not seem that much, but it actually is a LOT of difference, when materials are operated really very close to their limits. And that needs to be the case, if you want really to squeeze the maximum efficacy and color from the given lamp technology. Rule of thumb says 10degC increase in temperature halves the life when you are not yet approaching any phase change threshold (hen close to such threshold, the sensitivity becomes even way greater). Now if the lamp operates at 1500degC, 8% of that would be 100degC, so 1000x shorter life. In reality the temperature rise is not linear, because or the radiation formula has T^4 in it, so the temperature difference would be 25degC (in an optimistic case) instead of the 1000x, so life reduction about 6x when operating far away from any phase change threshold.

While I appreciate the refresher from undergrad physics, a presumed 6x reduction in life from 24000hours is 4000 hours, not 3 minutes. I also still don't have any answers to my original questions.

One you've been dodging repeatedly: is a nominal rated X-watt lamp (according to the spec sheet) rated for X watts of power input or not?

While I appreciate the refresher from undergrad physics, a presumed 6x reduction in life from 24000hours is 4000 hours, not 3 minutes. I also still don't have any answers to my original questions.

That was an assumption there is no phase change nearby, plus all the stress remains the same. If you start to melt something (e.g. an electrode or its seal), it means immediate destruction. The same if you overpressure it. E.g. by triggering a thermal runaway..

One you've been dodging repeatedly: is a nominal rated X-watt lamp (according to the spec sheet) rated for X watts of power input or not?

Not guaranteed. Some publish the real power, but some publish the power rating of the ballast it should be used with. Example are European "400W" rated MHs rated for both MV, as well as HPS ballast. Clearly the real power can not be the same on both (one is 3.25A, other 3.8A), but it is still reffered to as "400W".
Or practically all HPS: They increase their power input as they age by nearly 40% (from new, till the EOL). So a 70W could be anywhere between about 70..100W. Or another maker 65..90W. The common thing these have is the ballast characteristic they need to be run on.

Out of curiosity, what is your (others welcome as well) prediction if I were to overvolt the magnetic ballast just enough to force a full 400W into the lamp? explosion in 3 minutes like with the electronic 400W, or something else?

My money is on nothing unusual happening.

I am thinking I might try to see if a M135/M155 psmh or CMH lamp will be able to run on a H33 ballast using a European 4-5kV superimposed ignitor in the ballast circuit.

If a lamp cant handle 400W it won't handle them from either ballast type. My guess is it will either destruct as with the electronic, or cycle. (The longer zero crossings with magnetic ballast may give it enough opportunity to extinguish once it is overheating, but whether this happens depends on the lamp's pressure and saturated/unsaturated type)

I wonder if the Nominal arc drop for a 400w M172 pulse start metal halide lamp is around 170v. I got that arc voltage rating after running one of those lamps on a 400w H33 CWA mercury vapor ballast and superimposed ignitor. In addition, I noticed that the ballast seemed to run pretty cool.