Hi all,

I recently got a HPS yard light from one of my close friends. We tried all we could to make it work once again in the field, to no avail.

I brought it home with me (for the permanent collection) and am now beginning to disassemble it. I measured a few places with a multimeter. Across the taps of the ignitor (which is just a bare electronics board superimposed onto a sheet of "waterproof" paper) I had 120V. Additionally, checking the lamp socket, I also had 120V. This being a 55V lamp, is this normal? And if not, what component would cause the voltage to rise to that of mains? Additionally, if the behavior exerted is not normal, then is it possible that the lamp fried along with whatever other components? Additionally, there is no ballast hum whatsoever when turned on. Only silence.

The fitting is an ITT DR-70H.

that to me sounds like the ignitor has failed,



and to be honest thats a good thing, because if you had probed the socket otherwise, with the ignitor still working, it would of likely fried your multimeter! (since the ignitor outputs multiple kilovolts to strike the lamp)


(with no lamp but a good ignitor normally you would hear some buzzing from the ignitor)


120V is normal to measure with no ignitor, what you are measuring there is whats known as the open circuit voltage, of the ballast, when there is no load on a ballast the voltage will rise significantly, in the case of what is likely a Choke/series inductor type ballast in your case


that will be whatever the mains voltage is




I would fit a replacement ignitor and see if things work



as a getto fix, you could wire an FS2 starter in series with a 100W incandescent lightbulb, and put that across the E39 lamp holder terminals, this will at least let you try and verify if the rest of the system is working

The voltages are normal. The ballast is just an inductor in series with the lamp. The "55V" of the lamp is not what the power source of the lamp should exhibit, but the voltage that the lamp imposes into the circuit, regardless what the current feed is. Because of that, you need some means to control the current fed into trhe lamp and that is, what the ballast is doing: It imposes a point in the circuit where the current gets restricted to the value the lamp is designed for (1.5A for a 70W lamp), letting the lamp "dictate" its own voltage.

In reality the arc voltage is dependent on the exact arctube temperature, so responds to the real power fed to the lamp, but only after a delay, so the ballast is required anyway.

If the lamp is not there or is not ignited, there is no current passing through the inductor, so no voltage drop across it, so you see the full mains voltage across the lamp socket.

With HPS it is a bit more complicated that the lamp needs some higher voltage in order to breakdown the gasses in the arctube, about 1kV or so, until an arc gets formed in the tube. That is the task of the ignitor, in conjunction with the ballast coil. The ignitor senses there is 100V+ across the lamp, so generates 100V pulses between the "ballast tap" and the lamp end of the coil. The ballast coil is then acting as a step up pulse transformer, boosting the 100V pulses from the ignitor to the 1.5kV or so available at the lamp socket. Once the lamp ignites, the voltage across the lamp gets clamped to the 55V (well it is way lower when the lamp is cold, but rises to the 50..60V as the lamp warms up), so des not reach the 100V threshold of the ignitor, so the ignitor stops generating the pulses and becomes passive, so lets the arc burn in the lamp

Now what could be wrong with your fixture:
1) First look for obvious signs of severe damage, indicating the ballast is gone, like charred insulation on the ballast, awful burning smell or so.
2) It could be a bad lamp (very frequent are broken welds after the lamp is shaken during transport when even partially worn out). If you happen to have a known good lamp, just try it.
3) The ballast winding wire could be broken. But your test has shown the mains is passing, so it seems the ballast is conductive, so this issue is unlikely in your case.
4) The winding may have an internal short circuit, so beside being unable to limit the current, it is unable to step up the ignition pulses. A test for this would be to insert about 100..150W incandescent into the socket. If it glows significantly less than when directly on mains, there is a good chance the choke is not shorted. If the lamp lights nearly full brightness, there is an internal short circuit in the coil. It is better to disconnect the ignitor (disconnecting its Neutral wire is enough) for this test, to prevent igniting an arc within the incandescent.
5) The tap connection going to the ignitor may be broken. To test it, just take a voltage reading across the whole ballast coil (LineIn vs LampOut) and compare to the voltage between the IgnitorTap vs LampOut. The later shoudl be about 5..10%.
6. Check the ignitor, most often the internal resistor/HF choke break open and/or the capacitor loses its capacitance. Connecting a "capacitance range" of a common multimeter between neutral-Tap or Neutral-Lamp (only one of them, you need to try) should give a reading in the 50..500nF ballpark range (broken resistor or RF choke will yield <100pF reading, degraded capacitor <<100 nF reading), but none should be DC conductive (resistance reading >1 MOhm; about 1..10 kOhm means the SIDAC is shorted). Either of these, the ignitor is bad.
Dunno how accessible are its components (some were easy to disassemble), it could be fixed, there are just 4 components: A capacitor (usually in the 100..470 nF ballpark, rated at 250V at least), a few kOhm resistor (in the 1..10kOhm ballpark, rated at 2..5W), a RF choke (some mH, a spool of a thin wire with either open ferrite core or with no core at all; these tend to be extremely sensitive to breaks due to corrosion, because of the very thin wire used; the breakage could sometimes be located and fixed, but sometimes the coil needs to be replaced), then a sidac (a diode-like looking semiconductor device, acting as a voltage triggered switch; it could be tested separately by connecting in series with an 20..60W incandescent, it should act like a dimmer yielding slightly dimmed operation but without any flicker). The sidac very rarely fails on itself, but a humidity may corrode out its leads - that would be quite visible even without any electrical test.
Generally broken ignitor uses to be fixed by just replacing the ignitor (any ignitor designed for a "55V" HPS will work here, the exact wattage does not have to match).

Thank you Dez and Medved! I'll have a look tonight (I'll unscrew the ballast from its mount and have a look that way.)

The ballast looks to be in perfect shape honestly. Coils look pristine and there is no corrosion anywhere. Also, upon disconnecting the ignitor hot/neutral leads, what I do with those wires? Do I just put electrical tape on them for insulation or do I reconnect them to another place on the circuit? It seems to me that the wires going out of the fitting and through the wall go to the ignitor first, and then only later the ballast. Though I will have to confirm this when I disassemble it a tad later. Thank you again!! 

Also, lucky for me Dez, my multimeter has replaceable fuses. So even if it were exposed to such high voltage, I would likely only need to change the fuse (I speak from experience  .)

Thank you Dez and Medved! I'll have a look tonight (I'll unscrew the ballast from its mount and have a look that way.)

The ballast looks to be in perfect shape honestly. Coils look pristine and there is no corrosion anywhere. Also, upon disconnecting the ignitor hot/neutral leads, what I do with those wires? Do I just put electrical tape on them for insulation or do I reconnect them to another place on the circuit? It seems to me that the wires going out of the fitting and through the wall go to the ignitor first, and then only later the ballast. Though I will have to confirm this when I disassemble it a tad later. Thank you again!! 

Ignitor has 3 leads: One goes to Neutral (internally connected to a resistor inseries with the HF choke), one to the lamp (the one with the pulse capacitor) and one to the tap on the ballast winding (the one with the sidac). There is nothing on the ignitor that should go anywhere else.
In a fixture could be two other components though: A power factor correction capacitor connected parallel to the mains input and sometimes the photocell (one lead to Neutral, one to the input power and one towards the ballast Line input terminal). For a basic ignitor functionality the Neutral wire needs to be on its place, the other two (the Lamp and Tap wires) could be swapoed and the ignitor will still work, just a bit less reliably (a bit lower peak voltage, less favorable pulse polarity, but still should be able to ignite a good lamp at least when cold)
From your description it looks like someone had messed up the wiring...


The capacitor has no effect on the lamp itself, its function is only to reduce the reactive power the lamp circuit imposes to the mains wiring (reactive power does not transfer any real power, it is just a chunk of energy bouncing between the ballast coil and the mains and back, just unnecessarily straining the distribution network in the process; but a single fixture is of no problem, real problem would be many uncompensated fixtures doing that together). So you may leave it out, I would even recommend that until you make the fixture working.

And regarding the meter and ignitor: Indeed, the HV pulses are extremely dangerous for a meter. And even no fuse may protect against it: There are no high currents that may trip the fuse, just high voltages causing a dielectric breakdown. And moreover the HV pulses could be the thing that may also keep an already blown fuse still conductive (by igniting an arc in it), so even when some excessive currents would be involved, the ignitor action would prevent the fuse doing its job.

Medved tend to demonize the ignitor circuit it certainly will not strike an uncontrollable arc through a fuse, energy/voltage is too low. BUT it can flash over some poorly made input circuits and kill the multimeter. That's real, fuse won't help.

Anyway, if the ignitor is open circuit board, it is rather trivial to verify the components. If it is potted type, the easiest way is to check it is by replacement.

You  may try to start the lamp the old school way, by connecting a low voltage (single tube) fluorescent starter in series with say, 100W incandescent bulb, parallel to HPS lamp. Incandescent lamp works for current limiting, so a short circuit current of the choke (about 2A) won't stuck the starter.
 

Disassembled the luminaire (removed the ballast for easier access.) I find it pretty interesting how this ballast only has two connection points, those being for the photocontrol socket. The white wire off of the photocontrol plugs onto the ignitor board, as do the line-in leads (both Com and Hot) as well as the wires from the socket. So, how do I know which plug into the in-built capacitor, and which plug into the ignitor? I have provided a photo for reference.

Thanks.


EDIT: so I'm thinking the leads coming from the lamp sockets are in the ignitor circuit (for ignition pulses) whilst the hot/neutral wires coming from mains run through the capacitor for power factor correction. Though I'm lost as to why a third cable is needed for the photocontrol socket, and why it is plugged into the ignitor/capacitor board in the first place.

May be the wiring is messed up. Please do us a favor, draw a schematic in a normal way. It is super simple here, but we can't decide it from single photo shot from a single angle, where black wires are going into darkness...

 

Okay, will do. Thanks for bumping me in the right direction.

Maybe if you may also take more detailed photos of the ignitor board (the topside showing the components, also the bottom side so the connection could be traced), it seems to be something else than the usual semiparallel type...

Here you go, RRK and Medved. Please excuse the crappy schematic, I have yet to take physics / electronics class. 

Well, nothing crappy with this schematic. Connections look legit. Now you can lift ballast cold lead and lamp hot lead connection (black wires) from the ignitor board, disabling the ignitor, and probe the circuit safely. I would just use ~100W incandescent with two wires to probe, as multimeters can be sometimes misleading having too high input impedance and picking capacitive interference instead of a true voltage.

First, connect the incandescent lamp across the lampholder. It should light up at about half brightness. If it does, okay, ballast and photocontrol is good. If it does not at all - there is a circuit break somewhere. If it lights up at full - the ballast is shorted.

Next, I would try the trick with an incandescent and a fluorescent starter as a makeshift ignitor across the sodium lamp.

There may be tough cases as the choke partially shorted, hard to diagnose without some specialized tools. You can try to connect the choke right across 120V line directly and measure short circuit current. Should be around 2-2.5A AC I assume. Be careful as if the choke is really shorted, you can expect some fireworks ) Use some current-liminting in the circuit, at least 10A breaker, better something like 1KW heater or large lightbulb. Better use something short circuit resistant as a current clamp, as you can blow multimeter 10A input that way.

There is nothing crappy about the wiring diagram at all, I would say on the contrary, it is very clear and depicts how the thing is wired, which is the point of it in the first place...

How to test the ballast for an internal short came to my mind:
It is based on quality factor (aka losses) measurement, but allowing you to use just basic instruments you likely have.
First measure the DC resistance of the coil. That would determine what losses are to be expected as "normal".

Then connect the ballast in series with an incandescent lamp (disconnect the ignitor from the ballast and socket, connect the socket to the ballast directly or via an Ameter when measuring the current, put the incandescent into the socket).
Measure:
- The circuit current
- The mains voltage
- The voltage across the incandescent
- The voltage across the ballast coil.
All three voltages must be measured as close in time as possible and with the same meter on the same range, so they depict as accurately as possible the ratios at the same time.
Try to pose the results here, then we (myself or RRK) may do the math, I guess that would be too much for you if you still did not go through any electrical course yet...

Now the theory behind:
Now the 2'nd Kirchoff law says the sum of the lamp and ballast voltages must equal the mains voltage. Because we are on an AC circuit, that sum must be a vector one, taking into account the phase relations. But the meter only shows the absolute values, the sum of just those won't match.
But still we may calculate the phase relations (vs the circuit current) out of them. Plus we know the incandescen is a resistor, so we know its phase shift is zero.
Then calculate the phase shift in the ballast itself, with the current reading its exact impedance, subtract the anticipated lossesfrom the wire resistance and then look at what losses remain.
If the remaining losses is something smaller than the wire resistance, the ballast is very likely good.
But if the calculated remaining losses are way higher, there is very likely an internal short circuit within the ballast, as a shorted turn forms in fact a secondary winding (that turn) with the turn's wire resistance as its "load".

This is just a typical 2-wire ignitor that ITT/American-Electric used in the 80's...finding another will be difficult...may get lucky on eBay, or another member may have one.