I don't know if they ship to Finland. All of these are with offers. Last one is very nicely priced. You will need the Philips SX76 ignitor.
https://www.ebay.co.uk/itm/326535770415
https://www.ebay.co.uk/itm/315185634080
https://www.ebay.co.uk/itm/396669150993

Very interesting, I wonder where the barium comes from. Good to know it isn't just tungsten from the cathodes

I did a bit of research, and it turns out it is not a getter; it is excess Barium that evaporates at some stage in the arc tube filling process. Everything else I said is correct though.

There is a getter in the arc tube?!?! This is news to me! What is it made of?

Here's what I have heard. In the factories for SOX lamps, they used a sensor of sorts in order to ensure that the getter applied inside the arc tube was even on each side of arc-tube. In later years, this sensor ended up breaking, the result was uneven coating of the getter inside the arc tube, often with one electrode getting an absolute sock while the other only got a small amount.

A glow bottle fluorescent starter won't strike a 90w, believe me. Yes they do work for 10-18w (18w is a bit hit and miss). I haven't personally tried on a 35w but can confirm it doesn't work for a 55w. Therefore a 90w definitely won't.

Just buy a 90w ignitor, they're on eBay. They're a Philips SX76. Don't risk such expensive lamps on the wrong equipment.

@Ash
Yes of course, absolutely I can use the big equation with ballast impedance to determine PF, that is what I did to determine the average PF values across technologies. These averages are good enough to get ballast characteristics for most more common lamps.

But for other rarer lamps (short arc LPS, linear neon, super high pressure lamps), these averaged technology values are probably far from what the particular lamp actually is. I am sure it would not be wise of me to assume that my NA-1 has the same PF as an average SOX or SOX-E lamp, even though both are under the LPS category they surely are very different. And I am sure linear neon has a PF that is different from fluorescent even though they are physically similar. I don't have ballast specs for these less common types of lamps.

So naturally the need arises to get the power factor of a lamp which you don't have the ballast specs for, so you can use that PF to calculate the ballast specs. Let's use a 160W NE/H for example. I know that a 160W NE/H runs at 157V at 1.33A, but cannot find ballast specs for it. I normally should be able to calculate PF with just power, voltage, and current (PF=W/VA), but we have just proven that for some mysterious reason that method doesn't work accurately for discharge lamps (that calculated power factor does have applications that I have found, just not relevant here). Dead end.

So I can't get PF because I don't have ballast characteristics, and I can't get ballast characteristics because I don't have PF. Is this a standstill? Is there anything we can do for situations like the NA-1 or NE/H to get either PF or ballast characteristics from jus tthe running specifications so we can solve the puzzle?

The lamp power factor can be calculated by the same formula you used in reverse, ie when you know I and find out PF. It consists of a little displacement and a lot of distortion, so cannot be calculated just from phase angles like a purely displacement PF

It can be numerically measured if you have both Ilamp and Vlamp on a scope or power analyzer with some sampling and processing

HID ballasts are fairly efficient, so R is not significant compared to X. However, Vdrop(R) is almost directly in phase with Varc. To include it, change "Varc" to "Varc + Ilamp * Rbal" in the right places

Something im thinking about now..

In my recent visit to Dor - https://www.lighting-gallery.net/gallery/displayimage.php?album=9118&pos=0&pid=264621

We lit (for the first time since i found it) the M4A3 as is with the lamp and gear that came in it

Lamp is IIRC Philips Belgium SON-T Master 250W. Ballast is ELT Spain standard 250W ballast. Ignitor is ELT ballast dependent ignitor (connected to the ballast tap, not adding resistance in series with the lamp in run mode)

The nominal current for it is 3.0A

In our run up (measured with a fairly basic Uni-T AC+DC clamp meter, hall effect based), the current went from 3.15A for starting (seems legit, tho i expected more) to under 2.6A (quite a bit below the 3A)

Line voltage was around 220V, gear for 230V

@Ash

The actual lamp power may be different from the rating. See if you can find the real lamp power, or if it lines up better if you assume something like 39W instead of 35W for the lamp. (This still have to be verified, but lets see if this would explain anything)

The formula used is correct for Xbal, not for Zbal. They are not too much far away from each other, this alone would not likely cause as high differences as you see. Unless you have a formula that can account for Xbal and Rbal separately, use Xbal and disregard Rbal

On page 44 there is a ballast curve for this lamp, and it does use wattage as the y-axis. Maybe this is just relative wattage, but at 52V is does say 35W.

Is there a way to calculate the actual usable power factor without having every single spec of this lamp?

I was thinking the same thing with the X and R. I could make a formula that uses both, but I don't know how to get the X and R of a ballast when all that is given to me in the standards is Z. So whatever, Z probably good enough.

The actual lamp power may be different from the rating. See if you can find the real lamp power, or if it lines up better if you assume something like 39W instead of 35W for the lamp. (This still have to be verified, but lets see if this would explain anything)

The formula used is correct for Xbal, not for Zbal. They are not too much far away from each other, this alone would not likely cause as high differences as you see. Unless you have a formula that can account for Xbal and Rbal separately, use Xbal and disregard Rbal