I have been thinking of making chandeliers that use HID lamps such as MH, MV, HPS, and LPS. I am wondering what parts I can use to make a chandelier that consists of enclosed ballasts for each lamp and what I can do to firmly secure such chandeliers to ceilings to prevent them from falling down?

I would try to go for remote ballasts: In that way the chandelier does not have to support it, nor it will form unsighty boxes or blobs. Although you would have to install a dedicated "control cabinet" hosting the ballasts somewhere around.
But make sure the wiring length is within the limits of lamp to ballast cable lenght the ballast is rated for.
The problem uses to be with lamps requiring ignitors.

For certain lamps (pulse MH, S56 and maybe the 250W HPS) you may disconnect the original ignitor from the ballast and use an European superimposed ignitor placed close to the lamp, but these are limited to ballasts with OCV around the 230V and arc voltages below about 130V. Plus not sure how that installation will be viewed from the US electrical code perspective.

For certain lamps (pulse MH, S56 and maybe the 250W HPS) you may disconnect the original ignitor from the ballast and use an European superimposed ignitor placed close to the lamp, but these are limited to ballasts with OCV around the 230V and arc voltages below about 130V. Plus not sure how that installation will be viewed from the US electrical code perspective.

I have done this with a M98 HX ballast and remote superimposed ignitor with reliable success. I have also tried with a 50W M110 ballast and lamp but the lamp would intermittently flicker and sometimes the ignitor would time out on hot restrike. In fact, the 70W setup is in daily use in my kitchen.

With the one I converted, I removed all the E12 sockets and just hung a (remote-ballasted) E27 socket in the middle.

I think I have a picture somewhere... yeah, from a decade ago. Here it is with a silly 20,000K aquarium lamp in it. Ballast is housed in a table below, and is interchangeable.

@Xytrell : I remember that picture, males a very nice effect with all the glass parts.

A neat idea would be to use either the 20w or 39w CMH lamps since they are so compact . 

I’ve thought about it but I wouldn’t use one of those without a protective glass globe. I suppose you could make it work with a vapour proof fixture, though.

They would work in a fixture with some sort of globe or chimney over the lamp. Definitely not a good idea with bare lamps though.

I’ve thought about a chimney as well but they are relatively thin glass and they are open at the top. Otherwise, I would have used some already. Maybe if I could find one large enough to accommodate a vapour proof globe, however.

Lol, I had one in my living room when I first moved in, it was brass with glass beads and housed a remote ballasted 100watt Osram HQI-E lamp!

The remote ballast for one lamp can sometimes be wired "upstream" in the circuit, using the existing house wiring. In Israel in many houses there are junction boxes in the walls (high near the ceiling), which enables to install a cabinet on the wall over the box and install the ballast in it. When junctions are done in the device boxes or in the ceiling box above the luminaire this option is not as readily applicable

With a very small HPS lamp you might get away with the space in the "cup" that covers the ceiling connection of the chandelier..

Thats about it really



The wiring from the ballast to the lamp is a safety concern

Short can happen in the building's wiring for any number of reasons - production defects, damage during installation (hidden in the walls), damage from drilling holes in walls after installation, water leaks in the floor above, mice, etc..

As it is on the ballast output, if a short happens in this wiring, the current will be limited by the ballast - so there could be arcing going for quite long without tripping any breakers - not even many AFCI breakers (that require a minimum current to detect the arc, which may be higher than the output current of a ballast)

The most reliable option to protect from this is to install the circuit on a GFCI, and use a cable with bare Earth wire. If a breakdown situation happens, the current to Earth will trip the GFCI

Installing an ignitor remotely is additional danger, as the ignitor voltage can start isolation failures which otherwise would not happen. (In applications where the ignitor is in the luminaire, the section of wire that could be damaged is contained inside the luminaire. In outdoor applications of remote ballasts, the risk is considered insignificant). So i don recommend to install high voltage ignitors remotely at all

I also might consider also making a chandelier consisting of MH, MV, or HPS high bay fixtures. I am wondering how I can secure such a chandelier to a ceiling and prevent it from falling down.

In order to prevent it from pulling out of the ceiling , A retrofit ceiling fan box kit will work great.
https://litestoreusa.com/index.php?id_product=3251&id_product_attribute=0&rewrite=arlington-fbrs4200r-fanfixture-box-with-adjustable-bracket&controller=product&gclid=CjwKCAjwrKr8BRB_EiwA7eFapvEGyVUh-VKeEIvQa3d_AdSjsxVfOduzBlScsWj9PAoG4Y-PUyjVMBoCSUcQAvD_BwE

 Sorry for the painfully long link !   I've used these before and had no problems with them . This particular one is rated for a 90 lb fixture !

Installing an ignitor remotely is additional danger, as the ignitor voltage can start isolation failures which otherwise would not happen. (In applications where the ignitor is in the luminaire, the section of wire that could be damaged is contained inside the luminaire. In outdoor applications of remote ballasts, the risk is considered insignificant). So i don recommend to install high voltage ignitors remotely at all

I'm not sure what you're saying here. I always thought having the ignitor close to the lamp meant that the high voltage pulse would be limited to the (short)wiring between the ignitor and the lamp. Here in North America, remote ballasts are relatively common for probe start lamps, and I cannot imagine that a remote ballast with remote ignitor is much different than that. I'm not criticizing, I'm trying to understand what you're inferring.

I'm not sure what you're saying here. I always thought having the ignitor close to the lamp meant that the high voltage pulse would be limited to the (short)wiring between the ignitor and the lamp. Here in North America, remote ballasts are relatively common for probe start lamps, and I cannot imagine that a remote ballast with remote ignitor is much different than that. I'm not criticizing, I'm trying to understand what you're inferring.

He is refering to long range ignitors, which allow the cable between the ignitor and ballast to be very long (limit being in the 30..100m range; compare to max 2..3m limit with standard ignitors). These use to be of the parallel or a semiparallel connection and so stress by the high voltage the whole wiring between the lamp and ballast.
In theory the wiring should be able to handle it, but mainly outdoors the isolation get damaged over time, so while it may be still able to hold the normal operating voltages, it may arc over with the high voltage pulse. I'm talking about wiring in a state where it should not be used anymore, but we know the reality: "Dont fix it if it works", so it is operated with users oblivious of its state. There the extra HV pulse can trigger the disaster.