Hey, I haven’t forgotten about this, just couldn’t get my setup going. I will have the results for you in a bit. Sorry for the wait!

@James
Ahh I see, so it has to be pretty close to the arc tube for it to work. Maybe a device that incorporates the ignitor into the fixture's socket would be better. But that seems unnecessarily complicated.

I wonder why Iwasaki didn't have their ignitors all in the lamp cap. Of course the lamps look cooler and are more identifiable when the components are visible in the envelope, but it seems like a lot of work just to have it in the envelope.

Not easily possible with a standard lamp. For the Iwasaki FEC’s they only work in conjunction with a resistor (a tungsten coil in the outer bulb) and a sidac (hidden in the base to help keep it cooler).  The stems of these lamps have a third leadwire to build the necessary circuit.  Without that it will not work.

The Sylvania design lamps do manage with a 2-wire stem and they comprise both the FEC and Sidac in the lamp cap, in parallel with the arc tube.  The voltage spikes are quickly lost with increasing distance, maybe it could work if very close to the lampholder.

Correct, as far as I know all “35W” metal halide lamps except the automotive type are actually 39W.  Same as the 70W are actually usually somewhere closer to 75W, and the Philips 20W PGJ5 capped types are actually 22W.  The reason is in part because they were developed to run on HPS standard ballasts, but due to the very different power factor of the metal halide discharge their actual power dissipation is rather higher, despite having tried to limit that by reducing the arc voltage vs HPS.  Electronic ballasts will however often run the metal halide lamps at lower RMS power, due to the power factor of the discharge approaching unity at higher frequencies.

that is not correct. IEC 61167 rates these lamps at 35W as well.
The german DIN standard is the same as the IEC for these lamps as well as most other european national standards. If you look at datasheets, from example from Osram, these are also rated 39W

FEC ignitors that are built into many self-starting HPS lamps only need two leads to work, and use very few components. Could these be adapted into a standalone two lead lamp ignitor? I know that the integral ignition FEC HPS lamps have ignition antennas, is that strictly necessary or could it work with a normal antenna-less lamp?

Obviously to do this you would have to break open a self-starting lamp and assemble it's innards into a little box. Maybe someone could do that with an already dead FEC lamp.

Like superimposed ignitors, this would enable the use of pulse start lamps in probe start ballasts, but unlike superimposeds they would present their high voltage to the ballast windings, which is probably not so good for long term use with a probe start ballast.

Is this possible?

For those of you that like to browse my pics, just to let you all know that you’ll find them much easier to find now. I’ve created albums for each lamp type and wattage, and tube size and wattage. I’ve spent the last week moving photos. Theres much less photos per album now.

@James: What is the pressure of xenon inside XMH lamps, that they provides instant light during starting?
As far as I know, only in the US, 35W MH lamps are rated 39W, not at the EU and other territories.

Most important here is the power difference, and operating frequency. 

The XMH lamps have rod electrodes often with balled ends, whereas CMH usually have a heat radiator coil on the tip.  If you run an automotive lamp at low frequency, the cyclic current variation causes overheating of the electrode tip with irreversible damage during life - and on average too much heat loss via the rod to the seal.  These electrodes can only survive on the long term at higher frequencies : the design was optimised for use at 400Hz.  You can usually hear this while they are operating on the correct ballast.  Use at 50/60Hz will destroy them rather quickly.

Moreover, general lighting lamps are actually rated 39W, not 35W.  Even if you use a higher frequency CMH ballasts the RMS power is higher, and likely to accelerate not only electrode damage but also quartz corrosion.

By the way, the actual ignition voltage of regular automotive XMH lamps is rather low - its only for hot restrike that very high voltages are required.

Very good question!

Actually this goes back to the history of fluorescent lamp standardisation.  Many decades ago it was desired by many manufacturers to have FL lamps in discrete colour temperature steps eg 2700, 3000, 3500, 4000, 4500, 6500K.  Each manufacturer had their own unique phosphor composition to arrive at those values, which meant that the spectrum of one manufacturer’s lamp could be quite different than another even though their CCT’s were identical.  As such, their chromaticity appearances and colour rendering indices could also differ.  This was especially troublesome in large installations when replacing failed lamps of one brand by another : the differences were visually undesirable.

The lampmakers then began to standardise on the same phosphor recipes for each lamp colour, and the situation was greatly improved.  Later still, it was discovered that actually it was not efficient to target numerically precise colour temperatures.  For instance, by shifting from 4000K to 4100K it was possible to achieve a considerable cost saving in the raw phosphors, along with higher luminous efficacy, often a better colour rendering, and sometimes also better life.  There were large-scale collaborative efforts among the world’s principal lampmakers, co-ordinated via the standardising committees of the IEC and ANSI, to work out a set of optimal chromaticity points for FL lamps that would allow manufacturers to make better and lower cost lamps with improved interchangeability between brands.  These target chromaticities are specified in the relevant international standards and while they are not compulsory, it is highly recommended that manufacturers adhere to these.

When LEDs began to become popular for general lighting, both the ANSI and IEC performance standards encouraged nanufacturers to follow the same chromaticities and CCT’s as had been long established for fluorescent lamps.  This position was strongly supported by the traditional lampmakers and fixture manufacturers, so as to make the transition from traditional to LED lighting as seamless as possible.  However, the LED emitter manufacturers did not agree, for the very good reason that just like the performance and cost of FL lamps can be improved by targetting certain colour points, exactly the same is true for the available LED phosphors.  As such, led by the USA LED manufacturers which were formerly the leaders of that industry, ANSI standardised a different set of colour points.  Those have in the mean time been adopted almost worldwide.  The IEC LED performance standard continues to state that it is desirable to make LED products with same chromaticities as the old standardised F-series colour points, which is a bit silly because I think nobody actually does that any more.  Indeed, LED phosphors are also now developing at such terrific rate that even the ANSI LED colour points of the early 2000s are no longer always being followed.