Modulating HIDs

[GTBecker]

I have an application that requires identification of a visible illuminating source. Has anyone done any recent work with modulating HID lamps?

Contemporary automotive HID "ballasts" appear to strike an arc with ~15-25kV and then run the lamp with an ~85V 450Hz square wave. I can probably change the drive frequencies of several lamps to uniquely identify them, but I'm also considering the possibility of digitally encoding the drive, perhaps with a NRZ stream.

Any experiences?

[GTBecker]

FWIW, an HID update:

I've learned that automotive HID headlamp ballasts are of two general types, expensive and cheap, which each operate the arc very differently.

- The cheap "replacement" ballast produces ~85VDC @ ~400mA from ~30kHz switching to sustain the arc, once it is ignited with a short burst of ~15kV pulses. Since the drive current is DC the arc is continuous, so FM modulation is not possible.
- It might be possible to AM the light by injecting a control voltage at an opportune point in the ballast's constant-arc-current control hardware; these ballasts use analog design.
- The lamp starts at maybe 30% brightness and slowly brightens to 35W, perhaps over 30 seconds. The maximum input current is said to never exceed that of the 65W halogen bulb the HID would replace (~5Amps) but normally it draws ~3A@13.6V. If the arc fails once ignited, the ballast must be power-cycled to reignite the lamp.
- Poor internal construction; although sold as "waterproof", the cheap ballasts have large voids in the dense foam rubber potting material which exposes the circuit board; the same material makes hacking impractical.
- DC lamp drive erodes the arc electrodes differently, eventually cracking the quartz arc tube, which can explode; HID lamp manufacturers advise against DC drive.
- I got two different types of cheap ballasts; although one is labeled "Advanced German Technology" and has a few other German words on it, it is 100% Chinese inside and has no relation to Bosch, I am certain. That one cost $49+ship, the others (actually a pair of ballasts, with bulbs) cost $40 shipped).

- The expensive ballast (Philips/Hella) produces ~5VAC @~6A semi-squarewave @400.0Hz from 250kHz-switching; the design is processor-based. Unless the processor clock can be steered, the lamp drive probably can't be FM'd since code surely fixes frequency.
- Since the arc can extinguish briefly with each zero crossing, a ~10uS-wide >100V pulse on the leading edge of each 400Hz half cycle strikes the warm arc path.
- If required, a series of 23kV pulses will automatically restrike the arc.
- The bulb gets very bright very quickly. The lamp is initially overdriven by >300% to quickly bring the arc tube to temperature; applied power drops off as the bulb warms, settling at ~35W in a few seconds. The ballast briefly needs up to 20Amps during startup and settles at ~3A@13.6V.
- Sealed with dual o-rings and no potting, the circuit board is accessible.
- Far superior internal construction, compartmentalized, more power devices, well heatsunk, high-g shock protection.
- Cost $129 shipped.

Perfectly contrary to one another, one design uses high-voltage, low-current DC to drive the arc. The other uses low-voltage, high-current AC. Oddly, both methods put 35W into the arc, and at the same efficiency. The Philips ballast is clearly superior in every other regard but cost.

Now to see if it can be modulated.