Hello,
In the process of building several HID lamp test boards i have notice the electronic ballast have some advantages, compared to magnetic ballast. AN electronic ballast can be seen as a constant current source. for example Sylvania 20W and 15W MH lamps share the same electronic ballast as they share the same lamp current. however, in my opinion the ignition pulse of those ballast are very annoying.
There are balalst without build in ignitors. That is generally done for application were there is a rather long lead length between ballast and lamp. For example HMI studio lamps have an electronic ballast and the ignitor separated from the ballast in the laminar itself. I quit like that concept.
So I started to think, is it possible to disable the the ignitor in the elEctronic ballast and use a 4 wir ignitor. That would make for me the building of my test setup easier as i do not have to use long high voltage cable ( I have a lots of it from oil furnaces but it still is a pain in the a***) and also i don't have to isolate it ( a 3 pole relay is used to switch electronic hid ballast, Line, Lamp 1 and lamp 2 -- problem the 5kV pulses).
So is tarted to look at ballast pictures online. It seems like most simple ballast seem to use a circuit ver similar to the on found in three wir superimposed ignitors. That may be rather easy to disable by removing the diac.
However it seems like my higher quality ballasts seem to have another approach. My other thought was to maybe just short out the puls transformer. I will try to experiment with the ballast i have right now.

Anyone already thought about that or has anybody already done something similar? Does anybody maybe has shematics of Philips primavision, Vossloh Schwabe and Sylvania electronic ballast from MH lamps from 20-150w?

best regards,

Alex

Only automotive electronic ballasts for xenon metal halide lamps, are available without an internal ignitor.
But why the built in ignitor inside the electronic ballast annoy you?

I would not short out the transformer, it maz create excessive currents in the primary circuit of the ignitor part.

There are two ways how I would see the ignition may be done:
1)
One halfbridge is controlling the "DC" current (in each halfcycle of the output) by a 25+kHz PWM, with a filter on its output, calling it a "hot" side.
The second halfbridge would then switching the polarity at 400Hz. Lets call this the "cold" side.
The cold side output will be then connected to a HV pulse transformer, a tap of the HV transformer is then connected via a capacitor (call it an "ignitor capacitor") to the negative DC bus.
Assume the ballast is in a phase when the cold side is positive and it is about to change the polarity to negative.
If the lamp arc is not present, at the moment the top transistor switches off, the ignitor capacitor remains charged. Once the bottom transistor switches ON, it discharges the Ignitor Capacitor into the primary of the Ignition Transformer, creating an ignition pulse.
But if the lamp is ignited, once the top transistor switches off, the lamp arc current (regulated, so not that high) will cause that capacitor to discharge down to the negative bus side by the rather low arc current, only after that with some delay the bottom transistor is switched ON. so no spike generated.
This means just the HV transformer and the Ignition Capacitor needs to be added to form an ignitor, the rest is shared with the components responsible for normal function. It even would not need any special operation for the ignition, the circuit will do it automatically based on what load the arc presents.
With such topology, it would be enough to just disconnect the Ignitor Capacitor.

2) For ignition, the filter choke and capacitor are excited at HF into resonance and generate the high voltage this way. Once the arc gets detected, the circuit then transitions into the normal PWM regulated LFAC mode.
With this concept I'm afraid, would be very difficult to eliminate the ignition, unless you know how the circuit detects various states (lamp ignited or not yet) so are able to "fake" the signals.

Hello Madved,
thank you for answer. I think the topography of my ballast is the first one, as they have a seperate ignition transformer. I will draw out the exact ignition system of my ballast and report from there.
Disconnecting the ignition cap is indeed the easiest. I shortly thought about the loading issue when shorting the puls transformer but didn't think longer about it.

Dor123, your firs statement is wrong. Regarding your second statement, read the initial post again.

best regards,

Alex

Generally, there seems to be three different topologies for ignition transformer in an electronic MH ballast.

1. Ignition transformer with a DIAC/SIDAC/SIDACTOR/whatever it is called) as an oscillating element and MOSFET/IGBT switch to enable it, usually by passing B+ 400V from DC rail.
2. Ignition transformer where the storage capacitor is directly discharged in the series of short pulses by a MOSFET/IGBT to the primary winding, no DIAC. Typically fed from B+ rail  too.
3. An asymmetrical full bridge circuit where one half is doing a current regulating PWM and the second is used to form AC squarewave and briefly used to drive an ignition transformer, as Medved said. Mostly a Tridonic thing.

For all 3 topologies there should be an easy hack to disable the ignitor circuit. For (1) and (2) I would just remove the driving transistor. It is typically a ~600V MOSFET or IGBT located near the ignition transformer. Remove it or have its collector/drain disconnected. For (3) the easiest way is *probably* to disconnect a mid-point of the ignition transformer, not 100% sure here.

Hello,
I successfully eliminated the HV-puls transformer one two electronic ballasts made by VS and Philips. I removed the coupling capacitor that couples the drive signal onto the ignition coil of the ballast. It is worth noting, that now the electronic ballast make a rather annoying sound. If thats due to the missing case or something else I don't know yet. I haven't had time yersterday to completley finish the project. I definitely plan to find the Fet that drives the ignition coil and disable it too. I hope I find today some time to realize that.

There do not have to be any separate ignitor MOSFET, nor any other power semiconductor device.
And the capacitor used for generating ignition voltage could be the same capacitor that is normally filtering the lamp current, used to get moreless clean square wave, without the HF AC ripple from the PWM activity (that HF component could excite the accoustic resonances within the arctube, so cause heat uniformity issues).

Plus the current regulation loop relies on the capacitor to be present, so it may become unstable once you remove it.
The unstable and so oscillating regulation loop could, in fact, be the cause of the noise you are describing - the oscillation frequency of such instability I would be expecting in the few kHz ballpark, so right in the middle of the frequency band to be audible.
The sound won't be that much of a problem, but the oscillating regulation loop is definitely a big problem, as it prevents the ballast from maintaining the correct output and on top of that it also can also fall into a resonance of the arctube. So before keeping it like that, better verify with an oscilloscope the shape of the arc current. It should be rather clean 400Hz (or so) rectangle, maybe with smoothened edges (it has to pass a filter) or some overshoot (as the regulation loop is settling after polarity swap) and a little of the 30..60kHz "PWM ripple" (the filter is never ideal), but no oscillating in the kHz range or so, that would be clear sign of the regulator being unstable.

As sensing element, hypothetically ideal would be a clamp-on current probe, but I would guess that to be out of your reach. Then the best is a current transformer with its secondary connected to a resistor and the oscilloscope, to isolate the ballast output from the grounded oscilloscope.
You may make shift one using some mains-to-low voltage transformer with the secondary current rating at least the same as the lamp current rating.
You connect low voltage "secondary" in series with the lamp, then to the "primary" a resistor and the oscilloscope. The transformer will reduce the current in the same ratio as it transforms the voltage (so e.g. 1A through a "12V" winding will cause about 50mA in the "230V" winding), so you choose the resistor accordingly (to get about 1V signal for the oscilloscope, so about 22 Ohm for the "230/12V" example).
Even though normally the transformer is designed to operate at 50(60) Hz and may exhibit a lot of parasitics (leakage inductance,...) at the kHz range, it still works well as a current transformer till 100's kHz, so you do not have to worry at all.

Hello Madved,
thank you for your comment.

Regarding the Philips ballast, I doubt that there is a problem with the feedback loop as it indeed is of a topograpy that use a small power semiconductor to switch the ignition winding. The Exact topology of that ballast has been expleined here:
https://www.lighting-gallery.net/gallery/displayimage.php?album=2363&pos=76&pid=72025

regarding the VS ballast im not so shure... Here ther is no isoltion between the lamp part of the ignition transformer and the ignition line. there might come some problems.

Regarding the sound i heared is more going towards 400Hz not several KHz. It may very well be the result of just running the bare PCBs....

Best regards,

Alex

Anyway it is better to check the waveform (compare before vs after the modification).
But if the sound did not change by that mod, it is indeed likely you are right.

The thing is, what may sound as "400Hz" could well be 4kHz bursts of damped ringing (the poor stability may not lead directly to undamped oscillations, it may yield just damped ringing, but excessive ringing is not less a problem), coming at the 400 bursts per second rate. The ear and mainly brain is not equipped to distinguish that, because with the 400Hz rectangle what you hear is not the 400Hz itself, but the multi-kHz mechanical ringing which is being just triggered by the 400Hz edges.

So at the end it is ringing vs ringing, so very similar sound, you won't tell which is which without at least side-by-side compare.