I live in the US, so all of the ignitors I have ever seen are semiparallel. I understand that there are other types of ignitors that aren't used here, but I don't know how they work. Specifically, does anyone have general schematics of how these different ignitors work?

---

1) Superimposed:
I know there are only 3 components in a semiparallel ignitor, which makes them cheap, but all of the superimposed ignitors that I have seen on Ebay are very expensive. What is in there?

2) Parallel:
I believe these are only used for things that need less than 1kV ignition voltage, but I still don't know how they work. They must me simpler because they only have 2 wires, right?

3) Internal:
I hear that many Japanese high pressure sodium lamps have internal ignitors. How do they work? They appear to be pretty simple.

4) Other types:
Are there other types of ignitors other than parallel, semiparallel, and superimposed? If so how do they work?

---

Thank you so much!

1. Superimposed ignitors are similar to semi-parallel ignitors, but have their own pulse transformer. Also, electronic HID ballasts have it in their circuit. Also: Your semi-parallel ignitors are different than the European ones.
2. Indeed.
3. GEC used vacuum thermal starter in their internal ignitor HPS and MH lamps, which is similar to fluorescent air thermal starters, but in the vacuum atmosphere of the lamp outer jacket. It mains disadvantage is it is requires 15-20 mins to cool before it can close again.
Most HPS and MH lamps with internal starters have a glow bottle starters, similar to a fluorescent lamp starter. But these have a bi-metal in series to disconnect the starter when the lamp is hot.
Iwasaki made HPS lamps with vacuum thermal starter that have tungsten filaments rather than nichrome heater. Now they use an internal electronic ignitor consists of a Ferro Electric Capacitor + SIDAC in their HPS and MH lamps. Its advantage is there are no moving parts, but it also have a bi-metal to disconnect the ignitor when the lamp is hot.
4. The only electronic ignitors exists are parallel, semi-parallel and superimposed.

In what way are European semiparallel ignitors different from US ones?

How do parallel ignitors work?

I just spent about a half hour looking up what a ferroelectric capacitor is, and I think I understand now, but how are these used in a circuit to ignite a lamp?

Parallel ignitors generally work by two priciples:

1. A small storage capacitor is charged from device terminals and then is discharged by a SIDAC or SCR to a primary of a step-up transformer. Secondary of the step-up is again connected to device terminals, and so, parallel to the lamp.
2. Low-voltage parallel ignitor. SCR or SIDAC with relatvely high holding current is connected across ignitor's terminals. So when input current drops below the holding current, semiconductor switch abruptly closes and generates inductive kick at the ballast.

Ferroelectric capacitors work closely to semiconductor parallel ignitors variant (2) - when FEC voltage reaches certain threshold, it abruptly changes its capaciance, so againg causing abrupt current drop and inductive kick. Mostly loved by Iwasaki.

By the way it is a bit of stretch to say (as @dor123 did) that all electronic ballasts contain superimposed ignitors. While some (like Osram or Philips) may contain a dedicated pulse transformer to generate ignition pulses, its switch is not self-acting as in the case of external ignitor, but is timed by ballast controller. And others like Tridonic and Venture likely too, even do not have a specific ignition transformer, but drive one of their ballast coils into series resonance to generate ignition voltage, by mosfet switches also used for ballasting operation. The same coil is dormant in normal run, but works as a ballast during run-up.

   

Does anyone know of a schematic of FEC ignitors or of parallel ignitors?

Learn how to do proper web searches! Like google.com/patents. Fed with proper keywords, Iwasaki FEC ignitor, it shows EP0565113A1 for example. EP0565113A1 is full of ideas of how it's done, from a basic single FEC across the lamp all the way to intricate helper circuitry.

LG site is also already has some of parallel ignitor teardowns and also some reversed schematics!

In what way are European semiparallel ignitors different from US ones?

Do not know. Probably not very much. The idea is very simple with a storage capacitor, a semiconductor switch and a charging resistor, three components at minimum. Semiconductor switch discharges the capacitor to a partial winding of the ballast coil, working in autotransformer mode to step up capacitor voltage to >1.5kV ignition pulse.

Ask @WorldwideHIDCollectorUSA he seems to have invented that phrase...

FEC ignitor is just a special capacitor connected plain parallel to the lamp, maybe via a thermal switch (to reduce the dielectric stress of the capacitor during normal operation, as well as to remove the rather significant capacitance parallel to the discharge, as that makes the discharge less stable; also many ceramics tend to turn conductive at high temperatures, which could lead to local thermal runaway when triggered by some temperature spike, disconnecting it once the lamp ignites and warms up could be a way to prevent this runaway from happening, so allow the FEC to be placed tighter with the arctube, so in a more compact assembly; but such switch is used only when deemed necessary for a particular lamp model).

There is no circuit complexity, it is really plain parallel to the lamp. All the "magic" is within the dielectric material used to form that capacitor and its properties:
At low electrical field (so low voltage across the terminals) it exhibits very large capacitance (few 100's nF till uF, depends on the sizing), so creates quite significant current. But once the field reaches the saturation, in other words the capacitor is charged to a certain voltage, the material saturates, so the capacitance suddenly drops to way lower value (10x or even more). That means the charging current becomes way smaller than it was just before reaching that critical field. Because the charging current has flown through the ballast inductance, this sudden current decrease creates high voltage spike on it. And this spike is what ignites the lamp, what all the thing is after.


Parallel ignitors use to work by really broad range of principles.
Some are just switches (special SCRs, or even a thermal switch) temprarily shorting lamp, then breaking the circuit and so creating an inductive kick. This is used by fluorescent starters, glow bottle starters or resistor heated contacts built into selfstarting HPS anbd so on.
Or it could be a capacitor charged to 200V or so, then something like SIDAC discharging it to HV pulse transformer primary, the secondary is then in series with the terminals of said ignitor, so it passes the HV pulse across them.
Another type are high frequency oscillators, taking the low frequency (50/60Hz) power abd generating few 100's V of few kHz voltage. This is commonly used with SOX, as these contraption are able to deliver quite substantial power (10W or so) at that higher frequency so able to support the extra drop of copld cathode mode discharge and so warm up the cathodes, mainly in low pressure lamps where the anode column has rather normal drop even when the lamp is cold.
And another type I've seen was a Delon voltage multiplier, charging a capacitor to high voltage, until a spark gap ignites and discharges that capacitor to the terminals, so to ignite the lamp.
And I'm pretty sure there will also be many other principles for various uses, all qualifying as "parallel ignitors".

@Medved do you have any practical example of a parallel ignitor working as HF generator? Sounds snake oil. Outside of a standard circuit employed also in superimposed ignitors, where a SCR or SIDAC with high enough holding current is capable of doing a tight series of capacitor discharges over a single powerline half-period?

All the Philips SOX ignitors and I would guess also many of others.
Inside is a triac, capacitor, maybe an inductor and few components around, essentially forming a "negative-R" type oscillator (sometimes called "spark gap" oscillators, as the working principle is the same as the real spark gap oscilators worked in the Tesla's and Marconi's time) from either the ballast inductance (some H) and the series capacitor (some 10's till 100 nF, yielding higher 100's till few kHz) or the internal LC (like the Philips ones). The 50Hz mains essentially acts as a DC-like supply for the circuit, the triac with its control is acting as the negative dynamic resistance (operating like a bit glorified spark gap, using solid state components and an enable circuit which activates it only when the drop over the SOX is too large).

Iknow, these frequencies are correctly not called "high" but "medium", but the point was it is a generator of higher frequency than the mains...

I wonder if these FEC ignitors could be packaged externally and used with standard lamps. It seems super duper cheap but sort of reliable. Although if it needs to be used in a vacuum that would be annoying.

I don't think FEC needs any vacuum at all, it is just a ceramic disc capacitor. If the disc is big enough so there is long surface distance from one electrode to other, it should work.
Problem would be the lifetime: FEC involves rather high electric fields, so the material will suffer from TDDB (Time Dependent Dielectric Breakdown).  Energy traps are forming within the dielectric over time at strong fields, gradually forming "bridges" that then turn the material into a conductor, so a breakdown. It needs certain time to build up, it is accelerated by temperature. So such ignitor will fail rather soon, most likely when the first lamp starts to cycle (so raking a lot of hours of the HV stress over rather short time). So it would need to be replaced with each lamp anyway, so it is then easier to just build it into the lamp itself. The FEC by itself is very cheap part, we are really talking about few cents, so even some sort of socket mechanism (even the part related to the FEC itself) would even cost way more than the main FEC device itself.
If you take some high-K ceramic capacitor material like Z5U, X7R or so, hone the manufacture quality so it has minimum defects so can handle the high electrical field and you have something very likely usable as FEC ignitor.
In fact similar component is often used in miniature AC LED lamps (like the halogen capsule retrofits) - there the fact that the charge per cycle is practically independent on the voltage is exploited to get a constant average current ballast for the LEDs. And again it has a form of some high K material ceramic capacitor, driven high enough with voltage so it saturates. Although with the LEDs, the dielectric is stressed way less than with the ignitor role, so it is able to last at least as the LEDs there.


 And by the way the TDDB is the main mechanism behind power film capacitors losing capacitance over time: The dielectric breaks down at some spot, creates a short circuit on a small section, whivh then blows the fuse ling to that section, so disconnecting the damaged section out. But as that section gets disconnected, it does not contribute to the capacitance anymore, hence the capacitance loss observed externally.

In some cases, in regards to HID ignitors in Japanese HID lighting systems, I have seen catalogs mentioning that although many Japanese HID lamps use internal ignitors themselves for retrofitting existing mercury vapor fixtures, I have seen that some ballasts have the ignitor integrated into the casing and have seen that some remotely ballasted fixtures have the ignitor integrated into the fixture itself.