I'd like some clarification on the effects of lamp & ballast mismatches, since it is commonly done by accident (or sometimes, in the experimenter's case, on purpose).

Generally, what I have observed, that slightly underdriving a fluorescent or HID lamp (as in the case of a "residential"-quality ballast) actually extends lamp life when using the proper lamp.

I know that underdriving a discharge lamp to a large degree will cause electrode sputtering (blackening) and shorten life...and overdriving also shortens life by overheating the phosphor (if present) and arc tube (potentially weakening it)...but what is the worst scenario for ballast damage?

For example, I tried running a 4w T5 fluorescent on a 40w (rated 430mA) electronic ballast. The lamp was overdriven, but only measured to be about 13.5 watts. It didn't seem to massively overheat the lamp nor the ballast.

I am currently running 110w 1500mA PowerGroove lamps on an 800mA electronic ballast, also noting no abnormal effects.

I had been running a 125w mercury lamp on a 100w HX mercury ballast with no noticeable blackening or ballast overheating. What would happen if I ran an 80w lamp? Or a 250w lamp?

I frequently see 70w metal halide fixtures running 70w HPS lamps (which are 55v in the US) - they run dimly, but what is this doing to the ballast?

I tried running a 400w (100v nominal) HPS lamp on my 100w, 55v HPS ballast - it seemed to work fine, only the lamp only ran up about 1/2 way. Ballast didn't seem to be hot. Running a 35w, 55v lamp made the lamp run up too quickly and cycle, but 70w and 150w (55v) lamps seemed to work fine. Would either overheat the ballast over time?

Amazing !!     I was JUST thinking the same thing when I saw your post.  Other more informed will surely tell us the answers.

I'd like some clarification on the effects of lamp & ballast mismatches, since it is commonly done by accident (or sometimes, in the experimenter's case, on purpose).

Generally, what I have observed, that slightly underdriving a fluorescent or HID lamp (as in the case of a "residential"-quality ballast) actually extends lamp life when using the proper lamp.

Do not expect any lamp life extension wen operated on other then rated conditions. The reason is, then lamps components (electrodes, heat management,...) are optimized for the rated power, so the rated power is usually in the minimum of the wear rate.
In the best case scenario (the life does not depend on the change) the life would not change.
But what would deteriorate is the lamp performance (efficacy, color quality,...)

I know that underdriving a discharge lamp to a large degree will cause electrode sputtering (blackening) and shorten life...and overdriving also shortens life by overheating the phosphor (if present) and arc tube (potentially weakening it)...but what is the worst scenario for ballast damage?

Generally what the ballast "see" is the arc voltage (it is dictated by the discharge), so if the "wrong" lamp have the same arc voltage as the "good" lamp, there is no difference for the ballast at all.
If the arc voltage is different, it strongly depend on how the ballast is designed and how it respond to he different arc voltage.

For example, I tried running a 4w T5 fluorescent on a 40w (rated 430mA) electronic ballast. The lamp was overdriven, but only measured to be about 13.5 watts. It didn't seem to massively overheat the lamp nor the ballast.

I am currently running 110w 1500mA PowerGroove lamps on an 800mA electronic ballast, also noting no abnormal effects.

It for sure the F36T8 ballast overheat the F4T5 electrode filaments (by Joule losses, so current), so they will fail soon.
With electronic ballasts you can not judge based on it's external temperatures. The "thermal measurement spot" is there only to check the cooling when mounted inside a fixture, but can not be used as a base to determine, if it does not overheat when mislamped. For the surface emperature it may even happen, then some normally higher dissipation component run cooler, while other, normally cold, is overheating and so fail soon.
To really check, if the ballast does not have tendencies to overheating, you have to open it and compare temperatures of all components with the same ballast running the rated lamp.
As second you have to check critical component electrical loading, if it is not beyond the rating:
- Elecrolytic DC bus tank capacitor (with lower arc voltage lamps get stressed more for the HF ripple, with higher arc voltage lamps for the 120Hz ripple)
- Power semiconductors, as frequently they do not use heatsinks, as they run with very little power dissipation. But changing the ballast loas may cause the inverter to depart from soft switching mode, what could cause their power dissipation to rise even an order of magnitude (cheap, simple selfoscillating 36W ballasts dissipate no more then 0.5W on each power component, what by far does not require heatsink, but with different  or faulty lamp this power dissipation may rise to few watts, what then fry them)
- Ballasting inductor
- Peak voltage on the resonant capacitor (parallel to the lamp) vs it's rating. Count with at least factor of 2 margin for lamp aging.

I had been running a 125w mercury lamp on a 100w HX mercury ballast with no noticeable blackening or ballast overheating. What would happen if I ran an 80w lamp? Or a 250w lamp?

If the MV is able to warm up to the level where all mercury is evaporated, all have abut the same voltage (~95..110V).
So for the 100W HX MV ballast:
125W MV: Lamp slightly underdriven (so slightly lower efficacy), ballast OK. Within MV rating for dimming (50%), but without the required 15minutes warmup on full power. As the underdrive is only slight, I think no problem.
80W MV: The lamp 20% overdriven (obviously; so with all consequences), ballast fully OK
250W MV: May get stuck to cold, low power dissipation saturated vapor mode (not all mercury vaporize). That may overheat the ballast, as the arc voltage in such state is lower. MV ballasts are usually not rated for overcurrent, as MV lamps do not expose them to such conditions for longer time. Compare to CWA, HX boost current with lower arc voltage, so hey are more likely to transition the 250W lamp to the correct operating mode (all Hg vaporized), but as CWA does not rise the current, it does not overheat. I would say 250W lamp is too much for 100W ballast.

I frequently see 70w metal halide fixtures running 70w HPS lamps (which are 55v in the US) - they run dimly, but what is this doing to the ballast?

That definitely overheat the ballast.

I tried running a 400w (100v nominal) HPS lamp on my 100w, 55v HPS ballast - it seemed to work fine, only the lamp only ran up about 1/2 way. Ballast didn't seem to be hot. Running a 35w, 55v lamp made the lamp run up too quickly and cycle, but 70w and 150w (55v) lamps seemed to work fine. Would either overheat the ballast over time?

HPS are tricky, as their arc voltage depend on the arctube temperature (saturated vapor lamps).
So 400W/100V lamp on 100W/55V ballast: Check the arc voltage. If it is above ~50V, the ballast is OK.
35W lamp is overloaded, so it's arc voltage rise far above the 55V, so the series choke ballast is not able to maintain the arc. Ballast OK.
70W tube would be overheated, so have shorter life, but the ballast would have easier life (higher arc voltage on series choke => lower power dissiatin => lower temperature)
150W tube would be OK, but the ballast may run hotter (underdriven tube have lower arc voltage => higher ballast temperature)

Thank you, Medved, for the very detailed response!

  Quote from: arcblue on 19/12/2011, 21:19:00
  I frequently see 70w metal halide fixtures running 70w HPS lamps (which are 55v in the US) - they run dimly, but what is this doing to the ballast?

That definitely overheat the ballast.

Over here, MH and HPS lamps can use the same ballasts and ignitors, not sure if that's true for all types, but certainly the 70W MH and SON lamps I have can.

Some MH lamps (Philips HPI-T, Osram HQI-T/DV/DH, Tungsram HgMIF), operates on MV gear. And some MH lamps (Compact and CMH lamps, Osram HQI-T/D HQI-BT/D, Venture Whitelux, GE Arcstream), operates on HPS gear.
In the US, MH lamps operates on dedicated gear for either probe or pulse-start MH lamps, and much effort is needed to design them for operation on the American HPS gear.

In the US, MH lamps operates on dedicated gear for either probe or pulse-start MH lamps, and much effort is needed to design them for operation on the American HPS gear.

We don't seem to have probe start MH lamps here, at least I've never seen any.

Over here, MH and HPS lamps can use the same ballasts and ignitors, not sure if that's true for all types, but certainly the 70W MH and SON lamps I have can.

There is strong difference betwen European and American HPS design:
The American market lamps are designed to reach maximum efficacy for the given wattage, which means significantly different arc voltage for each wattage level.
The European market lamps were designed to get maximum efficacy, while still suffice with just a series reactor ballast on 220V mains. That means arc voltages between 70 till 100V for all wattages from 35W up to 700W.
The resulting arc and required ballast open circuit voltages then became the standard for HPS for the given market.
The reason for such difference in approach is the most frequent mains voltage. The 120V is way too low (exceptthe really low power levels) to becomethe ballast OCV, so the ballast has to be a transformer anyway. And if it has to bea transformer, then there is not much cost, nor efficiency penalty to design it for any OCV, so no obstacle to use such voltages to really get maximum efficacy.
For 220V mains, the 220V is sufficiently high, so when designing the lamps to work with just that as the OCV (so sacrificing some efficacy), the inherently lower losses of just a simple series reactor ballast gain the system efficacy back compare to the optimized lamp on a transformer style ballast.

The MHs started as probe start, to be operated on an MV gear. But it was quite soon clear, they would need a bit higher OCV than otherwise sufficient for MV (~220V). With American style ballast no problem - the extra OCV does not cost that much. But it was a show stopper for 220V world: The lamp won't suffice with just a series reactor anymore, so would need an autotransformer and/or an ignitor.
But when an ignitor became needed,it was feasible to make the MH completely without the otherwise problematic starting probe, so a pulse start lamp was born. And because that happened at the time the HPS were appearing, the pulse start MH were just designed to share the gear with the HPS (whe the MV compatibility had to begiven up). And because these probe less MH designs were able to reach better performance than the probed originals, they found their way to the US as well, of course there they needed again an autotransformer ballast. But these lamps retained the same specs world wide, so although compatible with HPS in Europe, in the US it mean incompatibility with any other lamp type.

But when an ignitor became needed,it was feasible to make the MH completely without the otherwise problematic starting probe, so a pulse start lamp was born. And because that happened at the time the HPS were appearing, the pulse start MH were just designed to share the gear with the HPS (whe the MV compatibility had to begiven up)

Any reason why 400W lamp does not follow this rule ? up to 250W the MH follows SON spec, then 400W MH uses the MBF spec + ignitor

Any reason why 400W lamp does not follow this rule ? up to 250W the MH follows SON spec, then 400W MH uses the MBF spec + ignitor

My guess: The higher wattages have inherently more stable arc (lower operating pressure, so longer ionization decay), co allow for lower reignition margin, so higher arc voltage while still remaining with just series choke. And the higher arc voltage means lower current and that means lower cathode losses, as well as ballast losses.
And other reason could be the fact the MV ballasts were more common than the HPS ones.
Anyway, the objective was not exactly the use of common complete ballast circuits, but mainly use the same components as used with already existing common systems (it is, because the European ballast systems are distributed as separate components like ballasting choke, ignitor, capacitor, so any combination of common components means available ballast system, even when such combination was never used before), so the need to combine a MV choke with some HV ignitor was no obstacle at all.

Then why werent the smaller wattage MH use the MBF spec as well ? The MBF have higher voltage / lower current than SON for virtually the entire range of Wattages

There are 400W MH lamps that operates on SON gear: Osram HQI-BT 400W/D, Venture Whitelux, GE Arcstream. Philips HPI-T 400W can also be operated on SON gear at the expense of shorter life.

Then why werent the smaller wattage MH use the MBF spec as well ? The MBF have higher voltage / lower current than SON for virtually the entire range of Wattages

Lower wattages need higher pressure to reach the full arc voltage with shorter arc, to allow sufficient arc loading even at lower current.
The higher pressure means shorter ionization decay at current zero cross, so a need for a higher reignition voltage. The US system designers responded for this by increasing the ballast OCV, but in Europe that is not possible. So if the OCV is fixed (the mains voltage), the only way to get higher margin for reignition is lowering the arc voltage. And that means higher current for a similar wattage.
The MVs do require higher pressure as well, but first still the used pressure is not that high, plus they can benefit rom the starting probe - it remains there, so it allows to generate quite a lot of charged ions even before the voltage across the lamp reaches the normal arc voltage, so that means way shorter time with ions just decaying without any generation.

Lower wattages need higher pressure to reach the full arc voltage with shorter arc, to allow sufficient arc loading even at lower current.
The higher pressure means shorter ionization decay at current zero cross, so a need for a higher reignition voltage. The US system designers responded for this by increasing the ballast OCV, but in Europe that is not possible. So if the OCV is fixed (the mains voltage), the only way to get higher margin for reignition is lowering the arc voltage. And that means higher current for a similar wattage.

Wouldn't this be solvable with an ignitor that actively helps reigniton at each zero crossing and not only starts the lamp ?

plus they can benefit rom the starting probe - it remains there, so it allows to generate quite a lot of charged ions even before the voltage across the lamp reaches the normal arc voltage, so that means way shorter time with ions just decaying without any generation.

Only in one polarity for most lamps...

Wouldn't this be solvable with an ignitor that actively helps reigniton at each zero crossing and not only starts the lamp ?

Only in one polarity for most lamps...

Ignitor: Theoretically yes, but it would be quite complex. The thing is, traditional ignitors wont have enough lifetime for such duty and will generate way too much RF disturbance. Of course at the end it is feasible to solve both things, but it would cost extra complexity and money, we are talking about something similarvto a fully electronic ballast...
 Designing the lamps with lower voltages so they wont need anything like that was way easier.

Probe: Of course, it is most efficient when on the hegative side, but it helps when on the positive too. It does not directly generate the free electrons, but it does generate UV and via photoemission on the main cathode it generates at least some. Plus it generates positive ions, which are accelerated towards the cathode and are able to free some electrons too. Again, not as efficient as directly the electrons when the probe is negative, stil it speeds up the plasma regeneration after the zero cross quite significantly.