What would the effects on the lamp be in terms of life, efficiency, etc if ran at let's say for example 20kHz or 100Khz?

What would the effects on the lamp be in terms of life, efficiency, etc if ran at let's say for example 20kHz or 100Khz?

Well for starters, electronic metal halide ballasts usually operate the lamp at a higher frequency. I don't know what it is but I think it's around 400hz. If you run HID lamps at too high frequency it can cause the arctube to explode since it can't handle the frequencies. I found this out the hard way with a metal halide bulb I tried running on an electronic Slimline ballast.

Most EHID ballasts run around 250 Hz

At too low frequencies (much below 50/60 Hz) you'll run into too long zero crossing periods, in which the plasma in the lamp will have sufficient time to extinguish, so the zero crossing of the AC cycle will effectively become a hot restrike. The arc won't be stable and will go out

At too high frequencies (order of KHz and up) the wavelength of sound at this frequency (and at the gas composition and pressure that's in the arctube) becomes on the same order of magnitude as the arctube length. If they "line up", you get a standing wave in the arctube - Uneven distribution of the gas pressure, which translates into uneven distribution of the heat dissipation in the arc. Think sort of a striation that's not moving but "frozen" in one position. From what i understand, the hot spots that form can destroy the arctube

Ontario used 25Hz power until 1949-1950 so there might have been very early MV installations that ran on 25Hz power. I know a lot of old preheat era fluorescents actually came with ballasts with a 25/60Hz selector switch. I suspect a lot of the old 25Hz only gear was long ago stripped out and scrapped.

Yeah I think 400Hz is pretty much the upper limit of most electronic HID ballasts.

Yikes, imagine how bad lights would flicker on 25hz!

No kidding. If I ever got my hands on some 25Hz fluorescent gear I'd be tempted to find a frequency converter to run them at 25Hz just to see how bad it was.

I think it might be possible to generate the 25Hz supply using a computer, audio generating software that can make an arbitrary sine wave, subwoofer amp, power resistor (to provide minimum impedance and prevent overloading the amp) and a step up transformer

For preheat with a simple choke (15..20W) using 2 standard 120V 60Hz chokes in series makes a ballast suitable for 120V 30Hz. At 25Hz this will overdrive the tubes somewhat and might also start to saturate, so when going down to 25Hz scale down the voltage to something like 100V

Would be interesting to also generate a frequency sweep to see how the flicker changes (for keeping the same current, the voltage must sweep accordingly at constant V/Hz)

Ontario used 25Hz power until 1949-1950 so there might have been very early MV installations that ran on 25Hz power. I know a lot of old preheat era fluorescents actually came with ballasts with a 25/60Hz selector switch. I suspect a lot of the old 25Hz only gear was long ago stripped out and scrapped.

Yeah I think 400Hz is pretty much the upper limit of most electronic HID ballasts.

Mercury Lamps cannot run on 25 cycle power. The time when the voltage is too low for the lamp to conduct is too long. As a result, the inductive kick can’t start the arc when the voltage reverses direction. Normal 50/60 cycle power has a short enough duration of time for the inductive kick to keep the arc going when the voltage switches. If you look at the voltage waveform of a mercury lamp on an oscilloscope, you will see what resembles a highly distorted square wave. There is a period near the zero point where the lamp doesn’t conduct, and then the lamp turns back on, and you can see a voltage spike at the leading edge of the wave.

A 1950 Westinghouse mercury lamp brochure gives the lowest frequency a mercury lamp will run at as 40 cycles. However, the lamps will run on direct current, provided the polarity is correct.

Look at the waveform in this video: https://m.youtube.com/watch?v=ZhSuFJ171u0

1950 Mercury Lamp Brochure:
http://www.lamptech.co.uk/Documents/Brochures/Westinghouse%20-%20Brochure%20-%20D%20Mercury%20-%201950%20US.pdf
Page 13 explains the frequency issue.

Running the lamp on DC is a problem too since you will be overheating one electrode. (You have to underpower the lamp somewhat, but then the performance is accordingly)

Running the lamp on DC is a problem too since you will be overheating one electrode. (You have to underpower the lamp somewhat, but then the performance is accordingly)

Plus you will be separating the fill by plasma electrolysis.

Of course, there are HIDs designed to operate on DC, but those can not run on AC then.

I agree that metal halide and sodium lamps must not be run on DC, but mercury lamps can be with some precautions. Polarity reversing switches were used.


http://www.lamptech.co.uk/Documents/Catalogues/GE%20-%20Catalogue%20-%201956%20US.pdf

Page 57, footnote #5

Mercury Lamps cannot run on 25 cycle power.

Actually it says right in here they have been tested down to 25Hz line frequency. Please see page 19 of this document.

Mercury lamps have been operated on frequencies down to 25 cycles with special ballasts.

In the same paragraph they also mentioned testing up to 2000Hz with no measurable increase in light output, however the main advantage of such a high frequency would be the lower cost of the ballast due to the lighter ballasts that could be used under higher frequency. 

Gavita makes 600W,750W and 1000W HPS grow lights where the lamp is run at 100Khz-High frequency ballasts.Supposed to be more efficient than 60Hz.Also Hortilux makes 600W HPS/CHPS/MH ballast that runs at 100Khz.Have two of these-run 600W HPS lamps very well.And the 600W CHPS.The 600W MH lamps have difficulty starting.Have some Growers Supply HF ballast fixtures that run 400W HPS and MH lamps-also can run 400WPSMH bulbs.These are HF as well.With some lamps can hear a ringing sound as the lamp powers up.Don't know what the frequency of these are-Digital HF ballast.

Mercury Lamps cannot run on 25 cycle power. The time when the voltage is too low for the lamp to conduct is too long. As a result, the inductive kick can’t start the arc when the voltage reverses direction. Normal 50/60 cycle power has a short enough duration of time for the inductive kick to keep the arc going when the voltage switches. If you look at the voltage waveform of a mercury lamp on an oscilloscope, you will see what resembles a highly distorted square wave. There is a period near the zero point where the lamp doesn’t conduct, and then the lamp turns back on, and you can see a voltage spike at the leading edge of the wave.

A 1950 Westinghouse mercury lamp brochure gives the lowest frequency a mercury lamp will run at as 40 cycles. However, the lamps will run on direct current, provided the polarity is correct.

Look at the waveform in this video: https://m.youtube.com/watch?v=ZhSuFJ171u0

1950 Mercury Lamp Brochure:
http://www.lamptech.co.uk/Documents/Brochures/Westinghouse%20-%20Brochure%20-%20D%20Mercury%20-%201950%20US.pdf
Page 13 explains the frequency issue.

If the phase shift is sufficient (assume inductive/HX lag ballast style), this problem wont exist, as at the momen of current zero cross the instant OCV is already above the arc voltahe, so continues to support the discharge for the opposite polarity. But because of rather high resistive losses at such low frequency (inductive components need more turns, so longer wires), it may need higher OCV to reach such phase shift condition.

There is big difference, if the lamp is operated at sinewave, peaking, or rectangular current.
Operating at rectangular current practically is operating at DC but swapping the polarity. Here the limits are the uneven electrode heating during single polarity  (so the temperature fluctuation) for the minimum frequency, then practically switching losses and related electromagnetic emission for themaximum frequency. Practically the ballasts use to operate somewhere between 100..400Hz (lower frequency for larger lamps, as the heavier electrodes have larger thermal inertia, but everything is physically larger, so has larger unwanted antannas to radiate, so has to use slower switching edges, which then need slower frequency to maintain the efficiency).
But because it is essentially a DC feed, there are no vibration in the megnetic forces (unless someone places there a permanent magnet, magnetic forces are proportional to the square of arc current, so because the current is just alternating polarity, the forces are constant,so no vibration), so there are no forces to exhibit the accoustic resonances so excite the standing waves and so hot spots.

The sinewave (inductive lag,...) or spiky (e.g. the CWA) current do exhibit the ripple ontheir square wave, so on the forces, so will excite the pressure waves. Therefore you should stay away from the resonant frequency range. With most high pressure lamps, the resonances are in the kHz till 100's kHz range, so the 400Hz is really a maximum for small lamps MV or MH. The narrow arctube of HPS makes the resonant peak very low, the higher modes even nonexistent, so it is possible to operate them using simple HF electronic ballast. But the lamp design must me made so the solid structures can not form a resonant peak either, as it may destroy them (like glassware shattering with loud sopranist voice in many older comedy movies).
Or you may operate the ballast way above the lamps resonance, but that is practical only with long arc lamps, mainly low pressure or some HPS.

Only the thick, long and mainly cold (so the ions are very slow, so cold) arc lamps (so the low pressure types) may benefit from the skin effect the high frequency feed bring, as the skin effect pushes the current density to the outer arc surface, so the emited radiation is less absorbed by the nonexcited atoms. From the era of thick T12's, gain about 10..15% of ectra light was observed already at 20kHz, for modern thinner lamps the effect is a less strong and need higher frequencies (but that is not a problem, given the newer semiconductors and soft switching ballast topologies).
The thin arc of high pressure lamps has already quite short distances and mainly large spectral line broadening (the high radiating atoms temperature means large Doppler shifts of the radiated photons, hence the broadening effect), so there is already very little selfabsorbtion (the selfabsorption happensmainly in the colder gas, so no Doppler shifts, so only narrow band matching the energetic transitions gets absorbed), so not much to gain.
So the frequency is more of a ballast simplicity and efficiency vs lamp robustness trade off (HF ballasts are way simpler, cheaper and less lossy, but their operation can excite the high pressure lamp resonances).

Only some MVs were operated using ~400Hz feed from a dedicated pressurized air turbines (so in a pressurized air powered lantern; no electricity went outside of the lantern body, which was pressurised by the feed air, so the electricity presence was possible only when lantern was well sealed and ventilated by the feed air at the same time), later these were replaced by using the the explosion proofed installation of rather standard mains.