Hello, I have some more questions regarding CWA ballasts for HID lighting:

---

1) HPS Differences:
How come HPS CWA ballasts have the capacitor connected between the primary and secondary when MV and MH ballasts have the capacitor in between the secondary and the lamp? Either way it is in series, so it shouldn't make a difference, but for some reason the ballasts are configured differently. Why?

2) Penning-Start Retrofit Lamp Compatibility:
Why is it not compatible? This really infuriates me because penning-start retrofit lamps seem to be a lot easier to find, but they can't be used with CWA, which is by far the most common ballast type for the larger wattages.

3) HX History:
Every single large-wattage ballast I have ever seen in my life has been CWA. Now they say with these penning-start retrofit lamps to only use choke or HX, but I have never ever seen an HX ballast that was 400W or larger. Anyone seen one?

4) Capacitor Omission:
What would theoretically happen if you just bypassed the capacitor? Would the lamp current go up or down? I would assume that it would go down, because a capacitor in series with an inductor makes less impedance, but I don't actually know and I don't have the means to test it right now.

---

Thanks so much!

My response to question 3:

Yes, I do own a rare North American specification Jefferson Electric high power factor 400W H33 mercury vapor high power factor HX ballast. It has the part number “233-1331”. Usually, most North American 400W H33 mercury vapor HX ballasts that I have read about are low power factor ballasts due to the absence of a capacitor wired in parallel with the mains. Examples of those ballasts can be found in Westinghouse’s OV400 security light, which has the exact same fixture body as a Westinghouse OV15 street light and a few versions of GE’s 400W PowerBracket security light fixtures.

Usually, lag type ballasts for larger mercury vapor lamps in North America are either 240V 60Hz reactor ballasts for 1000W H34 mercury vapor lamps or 480V 60Hz reactor ballasts for 1000W H36 mercury vapor lamps. On an interesting note, I have also found some evidence that North American mercury vapor HX ballasts were even available in 1000W H36 mercury vapor versions under the following part numbers:

GE 9T65Y2250 (1 lamp) 120V 60Hz tap available

Jefferson 232-1366 (1 lamp) 120V 60Hz tap available

Jefferson 232-2366 (2 lamps) 120V 60Hz tap available

Jefferson 233-1366 (1 lamp) 120V 60Hz tap available

I have found out about them after reading a GE brochure about its “I-Line” metal halide retrofit lamps designed for operation on mercury vapor ballasts. There might be more 1000W H36 mercury vapor HX ballasts out there under different part numbers, but they seem to be extremely rare.

See here:

https://www.lighting-gallery.net/gallery/displayimage.php?album=search&cat=0&pos=3&pid=210282

I also read that there might also be evidence of 1000W H34 mercury vapor HX ballasts out there after reading a 1961 Westinghouse mercury vapor lamp brochure had data listed for 120V 60Hz high power factor 1000W H34 mercury vapor HX ballasts:

See here:

https://www.scribd.com/document/394833080/Westinghouse-Mercury-Lamp-Brochure-1961

Info is found on page 12. Those ballasts seem to be extremely rare nowadays.

On an interesting note, 100V input mercury vapor HX ballasts for 400W mercury vapor lamps, low voltage 700W mercury vapor lamps (130V 5.9A), and low voltage 1000W mercury vapor lamps (130V 8.3A) are pretty common in Japan and much more common than 100V input mercury vapor CWA ballasts in Japan, plus that country in particular has a very wide plethora of high pressure sodium, quartz metal halide, and even ceramic metal halide retrofit lamps specifically designed as retrofit lamps for use on existing mercury vapor ballasts.

However, all Japanese quartz metal halide and ceramic metal halide retrofit lamps for mercury vapor ballasts that I have seen all seem to only be compatible with HX and reactor ballasts and explicitly warn users to not use them on CWA ballasts.

My partial correction for question 1:

In addition, when I recall looking at some specific information about 2 coil autotransformer HID ballasts, I often see that one coil serves as an autotransformer component to step up or step down the incoming line voltage and the other component serves as a “reactor” or choke component to limit the current. I have read that a transformer that consists of one single coil all by itself is known as an “autotransformer”. However, there were some other members even those with specialized knowledge claim that autotransformer ballasts have no real reactor component.

According to the 1961 Westinghouse, here is the description about how HX ballasts work:

“  THE HIGH REACTANCE BALLAST (Fig. 13) is used with circuits that lack sufficient line voltage to start the mercury lamp directly. This type of ballast consists of an auto-transformer section in series with a reactor element. ln power factor corrected high reactance ballasts, a few extra turns of wire are usually added to the auto-transformer section and a capacitor is then connected in parallel with the auto-transformer. The function of the auto-transformer is to step up the voltage to the value required to start the lamp. The reactor section limits the current as required by specifications.”

You can find this engineering bulletin that talks about this situation here:

https://www.scribd.com/document/394833080/Westinghouse-Mercury-Lamp-Brochure-1961

The information is on page 12 of the bulletin.

My response to question 2:

As far as I understand, penning start high pressure sodium retrofit lamps specifically designed for use on mercury vapor ballasts are said to only be compatible with HX and reactor ballasts instead of CWA ballasts because these high pressure sodium lamps and standard high pressure sodium lamps have a saturated sodium arc tube chemistry. The thing about high pressure sodium lamps with this arc tube chemistry is that they exhibit an increasing operating voltage during their lifetimes. This is said to be detrimental to mercury vapor CWA ballasts and the saturated vapor high pressure sodium lamps that operate on them because the constant current nature of these ballasts causes the lamps to draw excessive amounts of power whenever the operating voltage increases. This could potentially overload the lamp and possibly overload the autotransformer component of CWA ballasts. However, with HX ballasts and reactor ballasts, the lamp is able to maintain its power draw as its operating voltage increases because the operating current decreases as the operating voltage increases.

However, North American high pressure sodium CWA ballasts are able to safely function with standard high pressure sodium lamps because those ballasts usually have a much lower OCV compared to mercury vapor CWA ballasts. This ensures that the lamp cycles before its operating voltage increases to an excessive level.

Please note that I may be wrong in some of my responses.

Hello, I have some more questions regarding CWA ballasts for HID lighting:

---

1) HPS Differences:
How come HPS CWA ballasts have the capacitor connected between the primary and secondary when MV and MH ballasts have the capacitor in between the secondary and the lamp? Either way it is in series, so it shouldn't make a difference, but for some reason the ballasts are configured differently. Why?

If something is connected in series, for the circuit current it does not matter what is the exact order. So the circuit will work the same, whether the order is secondary-capacitor-lamp or capacitor-secondary-lamp. When nothing else is involved (with MV and probe-MH), it is easier to make one connection inside of the transformer component, so these use to be attanged winding-capacitor-lamp.
But once an ignitor function gets involved, a few complications arise:
First the ignitor needs to have a way to directly "observe" the lamp voltage, so two of its terminals need to be directly across the lamp. Connecting the series chain be capacitor-secondary-lamp makes that condition met with just 3 wires for ignitor pulser and a single tap on the secondary for the pulser.
Plus it generates high voltage spikes on the lamp terminal. Having the capacitor on the lamp side would mean there would be the full ignition voltage between the capacitor internals and its case, a stress that can be easily avoided when the circuit is arranged capacitor-secondary-lamp.
Because MV and probe-Mh do not use ignitors, the capacitor on the lamp side is offering simpler so cheaper and mainly less wiring error prone arrangement.
The HPS does need an ignitor, so the capacitor is placed between the primary and secondary for less stress and a simpler ignitor.

Hello, I have some more questions regarding CWA ballasts for HID lighting:

---

4) Capacitor Omission:
What would theoretically happen if you just bypassed the capacitor? Would the lamp current go up or down? I would assume that it would go down, because a capacitor in series with an inductor makes less impedance, but I don't actually know and I don't have the means to test it right now.

---

Thanks so much!

In CWA the capacitor is the dominant impedance component (at 60Hz), so it is not "capacitor subtracted from the inductor reactance", but rathet "inductor subtracted from the capacitive reactance". So shorting out the capacitor means you have eliminated the main ballasting impedance from the circuit.
Now what happens depends on how the two components are designed, what is the inductive component. If only small, the capacitor short would lead to severe increase of the current.
If the inductive reactance is large enough, it will lead to decrease of the current.
Noe each way has its pros and cons.
The lower inductance way leads to less turns on secondary (when placing the primary tap higher in voltage),so less losses and lower voltage rating requirements of the capacitor. On the other hand the capacitance needs to be higher, plus a short circuit capacitor fault leads to overcurrents, so ballast and lamp overheating. Plus the lower inductance means the higher harmonics of the current are less supressed, so the current gets more spikey.

Using higher inductance has tha benefit of fail safe (lower currents) response when the capacitor fails short circuit, but it needs longer secondary so higher losses and higher voltage rating of the capacitor.

If something is connected in series, for the circuit current it does not matter what is the exact order. So the circuit will work the same, whether the order is secondary-capacitor-lamp or capacitor-secondary-lamp. When nothing else is involved (with MV and probe-MH), it is easier to make one connection inside of the transformer component, so these use to be attanged winding-capacitor-lamp.
But once an ignitor function gets involved, a few complications arise:
First the ignitor needs to have a way to directly "observe" the lamp voltage, so two of its terminals need to be directly across the lamp. Connecting the series chain be capacitor-secondary-lamp makes that condition met with just 3 wires for ignitor pulser and a single tap on the secondary for the pulser.
Plus it generates high voltage spikes on the lamp terminal. Having the capacitor on the lamp side would mean there would be the full ignition voltage between the capacitor internals and its case, a stress that can be easily avoided when the circuit is arranged capacitor-secondary-lamp.
Because MV and probe-Mh do not use ignitors, the capacitor on the lamp side is offering simpler so cheaper and mainly less wiring error prone arrangement.
The HPS does need an ignitor, so the capacitor is placed between the primary and secondary for less stress and a simpler ignitor.


In CWA the capacitor is the dominant impedance component (at 60Hz), so it is not "capacitor subtracted from the inductor reactance", but rathet "inductor subtracted from the capacitive reactance". So shorting out the capacitor means you have eliminated the main ballasting impedance from the circuit.
Now what happens depends on how the two components are designed, what is the inductive component. If only small, the capacitor short would lead to severe increase of the current.
If the inductive reactance is large enough, it will lead to decrease of the current.
Noe each way has its pros and cons.
The lower inductance way leads to less turns on secondary (when placing the primary tap higher in voltage),so less losses and lower voltage rating requirements of the capacitor. On the other hand the capacitance needs to be higher, plus a short circuit capacitor fault leads to overcurrents, so ballast and lamp overheating. Plus the lower inductance means the higher harmonics of the current are less supressed, so the current gets more spikey.

Using higher inductance has tha benefit of fail safe (lower currents) response when the capacitor fails short circuit, but it needs longer secondary so higher losses and higher voltage rating of the capacitor.

interesting! because one thing I have always noticed about US CWA High pressure sodium ballasts, as well as the aforementioned difference of connecting the capacitor between the autotransformer and the choke part of the ballast, I also noticed that for a given wattage, the capacitance value of a HPS ballast has always been *much* higher then that for a equivalent metal halide or mercury lamp of the same sort of current range, and I have always wondered why that was exactly

for example a 250W 100V 3A CWA HPS ballast has a capacitance value of 35uF



but a 400W 135V 3.25A CWA Metal halide ballast is 24uf



or if you compare wattage to wattages a 400W 100V 4.6A CWA HPS ballast is 55uF



I note however that pulse-start metal halide ballasts, although they like HPS ballasts have the capacitor connected before the choke, their capacitance value seems to be the same as that of a regular probe start ballast



so obviously HPS CWA ballasts have a bit of a different design philosophy so to speak? (click on the images to make em grow larger )

100V HPS lamps require higher current then equivalent wattage mercury (mostly universal all over the world) and American MH (practically equivalent to the same wattages of mercury). So HPS at the same wattage generally requires lower ballast impedance and so larger capacitor value than both mercury and MH CWA ballast.

Another concern is a specific load curve required to work with HPS and avoid voltage/power overrun.

@WorldwideHIDCollectorUSA

Very interesting, I didn't know HX ballasts of these wattages existed.

Also, "saturated" means that there is both vapor and liquid, right? So, some may say there is an "excess" of sodium in typical arc tubes?

---

@Medved

That is what I thought, that it wouldn't make any difference. I never even looked at the schematic of a pulse-start metal halide CWA ballast, so it definitely makes sense to move the capacitor in any type of circuit with an ignitor involved.

I might test out my 400W CWA ballast (on a Variac) to see whether or not the short-circuit current is higher with or without the capacitor in series.

---

@LightBulbFun

Thanks for the explanatory pictures, I don't own any HPS ballasts so I never knew about the drastic capacitance difference between MV/MH and HPS.

---

@RRK

That definitely makes sense, higher current = more capacitance. I also hear that HPS is more prone to runaway situations, so that may explain the ballasting differences.

---

Thanks so much!

If anyone else wants to chime in about the HPS specifics feel free.

@Medved

I just tried it out, it looks like the secondary current goes down with the omission of the capacitor, though my testing was not super ideal.

Because of the severe power factor loss from removing the capacitor (I would assume), I had to use that same capacitor on the primary side (on the 277V winding just like they do with HX ballasts) to test it out, but even that wasn't enough to bring the primary current below the 5 amps necessary for my Variac. So instead of cranking up to full 120V, I could only go up to around 60V so I didn't blow my Variac fuse again.

With the measurements that I took of the secondary current at 60V input, the secondary short circuit current is significantly lower without the capacitor in series (with the output just shorted out) than it was when the capacitor was in series.

Does this mean that (with enough power factor correction) I could use this CWA ballast as an HX ballast for a lower wattage MV / MH lamp?

I think there is misunderstanding here. Yes you can change the capacitor from primary to secondary in theory. However said capcitors capacitance needs to be transformed from secondary to primary level.

No. The magnetic shunt in a HX ballast is designed to not saturate, but in CWA it is explicitely designed to saturate atthe desired current.
So with shorted (bypassed) capacitor, the CWA will always end up either at way higher than rated curren, or way lower. The "inbetween" is then rather unstable.
Plus the HX have the main primary magnetic circuit (the path for the magnetic flux when the secondary is open circuit) designed without any air gap, to minimize the inductive reactive current the primary sees, as it is anyway exposed to rather high inductive reactive current from the secondary load.
On the other hand the CWA secondary has the capacitance as the main ballasting reactance, which without any air gap in the primary magnetic loop would lead to significant capacitive reactive current. So the CWA have an air gap in the primary circuit, so to co pe sate thecapacitive load even beflre it hitsthe primary winding. So the primary is then handlingjust the real power, so has less losses.
Now once you remove the capacitor, you turn everything into inductive (so current phase lagging behind voltage phase), so you get double reactive loading on the primary, so overloading it. Adding a PFC to the primary reduces the mains input current, but has no effect on the primary, it will remain overloaded the same way.
Short term and when the ballast can handle that as a matter of tolerating a fault before it gets fixed (when the lamp power so heat in the fixture is reduced), but long term, like operating it that way normally, it would yield significant shortening of the ballast life. Plus the contraption becomes very inefficient (due to the extra primary losses).

Darn it, I though I was on to something. Once again the actually smart people prove me wrong!