Really neat! How far off the ballast specs are you running the lamp?

It is an ongoing experiment but this arrangement seems my best do far. The lamp current is close to correct with 0.81A versus 0.83A with the ballast but the voltage sits about 12v lower than specification. The SON 50W ballast is the next closest and runs cooler but isn’t as close a match. It looks like the fluorescent ballast has a higher resistive component than the SON ballast so maybe less efficient. So, the hunt for the ideal ballast goes on. As it isn’t used much, other than for demonstration purposes it likely to be fine other than subjecting the lamp to short cycles.

We run ours from a 50W SON ballast, which seems pretty much on spec. I keep meaning to run it up again and take some measurements.

How do you start the lamp exactly, Ria? Do you use an electronic ignitor, like the way you would start a 50 W HPS lamp, of do you use a starter or a push switch connected in series with the SLI's electrodes? Surely GEC's and Thorn's original starters for their SLI lamps are no longer available?

Tuopeek - Your drive current is pretty close to the rated one for the ballast, so I'm don't see why the latter would overheat. Ballast losses depend solely on current and internal resistance.

Yeah, it is pretty close to the rated current Max, but I think the lamp has a lower resistance than the 65w fluorescent so the voltage falls across the lamp and current has increased. My interpretation is the wattage is about right but the choke has a larger volt drop across it. The inductive component is fine but the resistive part may mean the windings dissipate more heat. That was my thinking. Fluorescent chokes seem to run a lot hotter than HID ballast in general so maybe I'm jumping to conclusions thinking the choke seems hotter than I'd like.
I use a suitably rated fluorescent glow-switch starter for this circuit, think that was the normal practice for the 60W SLI/H

Normally the relation between voltage and current should be fixed in a given choke, determined by its overall impedance, i.e. R+jLω. So, I'm not sure how you'd get a higher voltage for a given (close to nominal) current passing through the choke... unless SLIs generates a lot more harmonics than fluorescent tubes, but I really doubt that.

Maybe I need more experiments and thinking on this one yet

@Max - I believe the lamp was designed for switch-start as per a fluorescent tube, so that's what I do. I usually use a pushbutton (must put up a picture of it) but a normal fluorescent starter seems to work ok.