I would like to know what the best Fulham Workhorse ballast is for each SOX and SOX-E lamp. I only know that using a Workhorse 3 with both red wires will power an 18w SOX lamp, but I also heard that it could overdrive such a lamp. So, what is the best Fulham Workhorse ballast and the number of red wires for each SOX and SOX-E lamp?

"but I also heard that it could overdrive such a lamp."

That's all of course old wive's tales propagating around lamp lovers circle...

Someone is in need to step in and finally measure that damn current and see how it matches the lamp's prescribed value of 0.35A!

I know measuring RMS at ~30KHz is kinda tricky, as it is out of range of cheap multimeters. So some specialized tools or a bit of soldering is required.
The best and safe way is to use a precision current probe and an oscilloscope. I have ones, but no Workhorse 3 ballast. I know @FrontSideBus have them too (that nice Micsig probe). Anyone else?

Also one can stuff 1 ohm precision resistor in a *cold* wire of ballast output, needs to be checked with oscilloscope, I assume this has to be *not* a red wire. And then carefully measure RMS voltage across the resistor with an *isolated* ocilloscope, or run this experiment on an isolation transformer.

Or one can stuff a diode bridge of reasonably fast diodes like 1N5817 Shottkys in a cold output wire and measure average DC current at the bridge DC output with a regular multimeter or just an ampermeter gauge. The value measured will be average rectified which is 1.11 times lower than RMS and also may be somewhat off if the current is not exactly sinewave.

Could you clarify a bit about the diode method? I'd rather buy cheap diodes than an expensive multimeter. Also, what do you mean by "a bit of solder"?

Quite easy to make, yet very accurate sensor for high frequencies is using a ring core, wind about 33 turns on it, connect it to a 3.3 Ohm 1% resistor, then connect that to the oscilloscope input (you may switch it to 50 Ohm if you really want to go till 100's MHz, with 11mm diameter core).
Then you just pass the wire with the current to measure through the center of the core.
It froms a 33:1 current transformer, so with the 3.3Ohm shunt it will have sensitivity 0.1V/A and work fromfew kHz up tp higher 10's of MHz (the upper end depends how exactly you form all the connections, but in real life the inductance of the primary will be stronger limiter for use of this contraption for high frequencies/fast edges; used this to debug less than 1ns artefacts on 10ns current edges in a 2MHz SMPS, so bandwidth in 150+MHz).
The drawback is the low frequency limit: A typical Al=100nH core yields minimum usable frequency around 10kHz, when the low 3dB corner is 3kHz.
But for a 25kHz HF ballast output measurement it is good enough, if you don't mind messing with wires when going to different measurement point.
The big advantage is, it offers galvanic insulation from the oscilloscope, so you don't need to worry about shorts via ground leads...

@Medved - And you don't even mention ferrite ring magnetic permeability...

Someone will take a HF ring or even a powder core with mu=50 and then will be surprised why the results are so off.

Too much problems when you can't verify / calibrate your DIY current transformer against something known to work - so not a good solution for laypeople.

A problem of chicken and the egg...

Also, what do you mean by "a bit of solder"?

A little bit of soldering iron work.

This is the circuit I propose.

Make sure everything is making contact and meter is set to DC amps before turning power on. Otherwise, the ballast will certainly try to apply its high open circuit voltage and will fry the diodes and potentially your meter.

Thanks for drawing it out; it makes it a lot easier to understand. Should I use a SOX lamp in that case, or could I use a normal fluorescent lamp?

The fluorescent is good enough, if it has very close nominal arc voltage, it would be perfect...

But if the intent is to run SOX lamp in spec, the current should be finally verified with said lamp fully run up, sure.
Though, 18W SOX is electrically close to 18/20W 600mm fluorescent.


 

Also, it may be good to add an electrolytic capacitor to the circuit (any practical voltage you have on hand). Normally, a meter should suppress this, but some cheap ones may go mad about a little remaining HF in the DC current measured, why take a chance?

I was wondering what would happen if you gave a multimeter in the DC a super bumpy DC. The cap will definitely be helpful.

Also, it may be good to add an electrolytic capacitor to the circuit (any practical voltage you have on hand). Normally, a meter should suppress this, but some cheap ones may go mad about a little remaining HF in the DC current measured, why take a chance?

Would using this full bridge rectifier work? It's a Schottky-based type, so the recovery time should be really low.https://www.digikey.com/en/products/detail/bourns-inc/CD-HD201/6561443

Yes, this bridge will work, though you should be careful with modern SMD parts as they get their cooling through the board mounting and when used in volume mount 'rat nest' circuit the cooling may be not adequate. Here the load is about 1/7 of max specified for the part, so you decide...

I was wondering what would happen if you gave a multimeter in the DC a super bumpy DC. The cap will definitely be helpful.

Multimeters use double slope integration AD conversion with relatively long times so are insensitive to the AC component on DC ranges. Some modern ones are using a sigma-delta AD with a digital decimation afterwards, which is generally a modern equivalent of good old integrating ADC.

But you never know how many corners are cut in a cheap design)

Again, do not leave the ampermeter load in this circuit disconnected even momentarily as this will fry the diodes and blow the capacitor. You may add a safety bypass of some kind (like ~5 pieces of 1N4007 diodes in series) if you can't be sure.

Thanks for letting me know not to disconnect the multimeter while measuring the lamp current. As for cooling, I'll limit the setup to short runs. I will also use this rectifier because of the higher voltage rating. See here: https://www.digikey.com/en/products/detail/smc-diode-solutions/KMB220S/7898347. It is also based on Schottky diodes.