Does anyone know, or have any tricks for measuring output voltage coming from a HF electronic ballast? Is there a correction factor or equation to get the approximate voltage from a reading on a meter that's made to measure sine waveforms (most basic mutimeters)? I'm assuming the output is approximately a square wave, meaning the RMS approximation in the meter will give a lower than actual voltage, and my measurements read lower than expected. I've got the Amprobe AM-60.

The HF resonant (so virtually all mains electronic) ballast output / lamp arc voltage is something between triangle and sinewave, not the rectangle anymore. The reason is, then the ionization level (so plasma conductivity) can not follow the so fast changing current.
From this point of view it would look like there would be no problem of using any multimeter.

But the problem is the high operating frequency vs multimeter operating frequency range. All multimeters are designed to operate mainly at 50 or 60Hz, with a bit derated error at 400Hz, while their frequency bandwidth end few kHz (where the error already reach about 30%).
As you never know the exact frequency response of the multimeter you have, it is not possible to extract any "correction factor".
Moreover such high frequency very frequently cause the multimeter front-end to stop working correctly, so you easily get "out of ranges" warning with 4V signal at 400V multimeter range...

The only option to measure at HF is using scope with HV probe (so ~1:100 for most newer scopes) and let the math in the scope (or transfer the curve into Excell and do the math there (function stdev(wave_points) get you the rms value; assume no DC component) calculate what do you need.
But be careful, for these you need an insulation transformer to supply the ballast (the best way, as it protect even you) and/or fully insulated scope (e.g. USB scope operated using battery powered NB) - unlike mains powered multimeters, most scopes have their shields grounded to the socket PE terminal, so connecting them into the ballast circuit you create an effective mains short.

Cool, thanks! That's sort of the answer I expected, but it's interesting with the different waveform.  The other thing the scope should answer is what the current crest factor of these fulham ballasts is that I've been using. Them not listing the number in the spec sheets is never a good sign. Now to get access to a scope... Maybe the electrical engineering department at school would be kind enough.
I know my meter is able to read frequency up to something like 2 Mhz, but the voltage/current measurement response is probably not nearly high enough as you said.  It measures a 45 Khz signal from the workhorse 2 ballast driving my 18W SOX lamp and around 80 Khz frequency from the workhorse 5 driving a 55W SOX lamp. It also gets a reading from these even if just one probe is touching the insulated wire. High frequency voltage is weird...

But be careful with the grounding issue.
If you would use the scope on the notebook, don't forget the notebook would electrically "sit" on mains live potential (of the net you will use to connect the input shield), so:
- Disconnect all peripherals from it and don't touch the touchpad (you are on the ground potential, so the 100's V could damage it).
- Place it on some well insulating surface
- Disconnect the mains adapter
- Don't touch any of it's metal/conductive part, you may get shocked
- Use only "regular" plastic (without any metalized part) and/or (better) wireless mouse for control

@Medved : Can you rectify it to DC with rectifier bridge + small capacitor, measure DC, divide by sqrt(2) ?

Measuring with a multimeter will be hard, and give misleading readings. The only multimeter that will work are the Fluke Scopemeter series, and the clones of it. If using a scope, then you will be best off be to use a isolated probe set, or a differential probe with an isolating transformer with low leakage ( a medical isolation transformer would be the lowest leakage transformer). Otherwise you should use the industrial Tektronix units, which are specifically designed to be used on industrial inverter outputs, and which have a totally isolated input that is rated to 1kV isolation, and which have some quite good on board maths designed for industrial automation. Good but pricy.

Most true RMS meters will provide a correct reading, even on a badly distorted waveform, but many TRMS meters have a limited frequency range, and may roll off at lowish audio frequencies. They are generally only specced to give a true reading to around 440Hz.

@Ash:
Shape error: Triangle (lower arc voltage lamp) would give ~20% higher, "circular" (like sine, but with lowered peaks; present mainly with higher voltage lamps) about 20% lower reading then real rms when using the peak-detector with the sqrt(2) correction.

Rectifier error: Diodes have drop, the rectifier may cut down the peaks. Both causing error in ~10% range

@SeanB: Regular insulating transformer is enough. In any case you should be aware, what node is "cold" (to connect the grounding wire) and what "hot" (to connect the probe tip). For current measurement you have to use transformer, e.g. wind ~10..30 turns of wire around some ferrite ring core, connect 10..30Ohm resistor (1Ohm per turn) and connect this resistor to the scope input (or make 50 turns and use it with 50Ohm scope setting), pass the wire with the to-be-measured current trough the core and you get nice 1A->1V current probe usable from ~10kHz (most ballasts operate at 30..70kHz) with the isolated input. Moreover you may measure the real arc current without the error caused by the heating current (in current mode heating lamp configuration - pass both wires from the single lamp end trough the core, the filament current would compensate out).

And as insulated scope for ballast measurement I would prefer something like this, it provide isolation for each channel and it is way cheaper then the Tek (however more expensive then non-isolated USB scopes)

Does anyone know, or have any tricks for measuring output voltage coming from a HF electronic ballast? Is there a correction factor or equation to get the approximate voltage from a reading on a meter that's made to measure sine waveforms (most basic mutimeters)? I'm assuming the output is approximately a square wave, meaning the RMS approximation in the meter will give a lower than actual voltage, and my measurements read lower than expected. I've got the Amprobe AM-60.

You'll need a special meter. You'll need to be very careful if you're measuring OCV as you might fry the meter. I believe the Fluke 289 goes up to about 100KHz, but it may have a V*Hz limit. Many Fluke meters are limited to one or ten million.
So, at 100KHz, a 10 million V*Hz meter should not be used to measure over 100v.  (100 * 100K = 10M). Current flow through the meter's capacitive component increases with frequency.

Yokogawa, Agilent, Hioki, Voltech and such make power meters specifically for such applications, but they cost about as much as a cheap new car. 

Quote from: SOX55W on August 25, 2011, 06:43:11 PM

Does anyone know, or have any tricks for measuring output voltage coming from a HF electronic ballast? Is there a correction factor or equation to get the approximate voltage from a reading on a meter that's made to measure sine waveforms (most basic mutimeters)? I'm assuming the output is approximately a square wave, meaning the RMS approximation in the meter will give a lower than actual voltage, and my measurements read lower than expected. I've got the Amprobe AM-60.

You'll need a special meter. You'll need to be very careful if you're measuring OCV as you might fry the meter. I believe the Fluke 289 goes up to about 100KHz, but it may have a V*Hz limit. Many Fluke meters are limited to one or ten million.
So, at 100KHz, a 10 million V*Hz meter should not be used to measure over 100v.  (100 * 100K = 10M). Current flow through the meter's capacitive component increases with frequency.

Yokogawa, Agilent, Hioki, Voltech and such make power meters specifically for such applications, but they cost about as much as a cheap new car. 

These meters all are VERY expensive, way more then some 100MHz digital scope. As even portable computers are common as dirt, I see no reason to not use screen-less scope for this. All offer the extraction of the basic parameters (frequency, rms values,...), so you have quick-readout as well. Majority of them is able to export data, so in the computer you may easily do even quite complex math, including relation between multiple signals (what is in some cases very useful)
For way cheaper cost (even if you add one HV and one current probe) then these specialized meters you get the ability to measure up to 100MHz (this is a minimum for power electronic troubleshooting) and you get an universal equipment usable for troubleshooting the ballast electronic (e.g. find out, why the transistors are suddenly dissipating so much,...)

Such deeper analysis (more complex then work - don't work) on ballasts you would do on the bench anyway, so the larger dimension of the scope and PC are not as big problem...
Moreover many hand-held computers (phones,...) allow to connect the scope to them and corresponding SW is available, what make from it in the field usable instrument as well...

Quote from: Luminaire on September 08, 2011, 07:31:07 PM

Quote from: SOX55W on August 25, 2011, 06:43:11 PM

Does anyone know, or have any tricks for measuring output voltage coming from a HF electronic ballast? Is there a correction factor or equation to get the approximate voltage from a reading on a meter that's made to measure sine waveforms (most basic mutimeters)? I'm assuming the output is approximately a square wave, meaning the RMS approximation in the meter will give a lower than actual voltage, and my measurements read lower than expected. I've got the Amprobe AM-60.

You'll need a special meter. You'll need to be very careful if you're measuring OCV as you might fry the meter. I believe the Fluke 289 goes up to about 100KHz, but it may have a V*Hz limit. Many Fluke meters are limited to one or ten million.
So, at 100KHz, a 10 million V*Hz meter should not be used to measure over 100v.  (100 * 100K = 10M). Current flow through the meter's capacitive component increases with frequency.

Yokogawa, Agilent, Hioki, Voltech and such make power meters specifically for such applications, but they cost about as much as a cheap new car. 

These meters all are VERY expensive, way more then some 100MHz digital scope. As even portable computers are common as dirt, I see no reason to not use screen-less scope for this. All offer the extraction of the basic parameters (frequency, rms values,...), so you have quick-readout as well. Majority of them is able to export data, so in the computer you may easily do even quite complex math, including relation between multiple signals (what is in some cases very useful)
For way cheaper cost (even if you add one HV and one current probe) then these specialized meters you get the ability to measure up to 100MHz (this is a minimum for power electronic troubleshooting) and you get an universal equipment usable for troubleshooting the ballast electronic (e.g. find out, why the transistors are suddenly dissipating so much,...)

Such deeper analysis (more complex then work - don't work) on ballasts you would do on the bench anyway, so the larger dimension of the scope and PC are not as big problem...
Moreover many hand-held computers (phones,...) allow to connect the scope to them and corresponding SW is available, what make from it in the field usable instrument as well...

Many of these meters have the proper accuracy for Energy Star testing and such and offers on-the-fly simultaneous display of various parameters without having to record, then calculate later on computer.  Many are six channels so multiple measurements can be made (i.e. to make measurement inclusive of filament power or measuring efficiency of a three phase device.

Many of these meters have the proper accuracy for Energy Star testing and such and offers on-the-fly simultaneous display of various parameters without having to record, then calculate later on computer.  Many are six channels so multiple measurements can be made (i.e. to make measurement inclusive of filament power or measuring efficiency of a three phase device.

But their cost limit their economical use...
You may write scripts doing the same calculation with the scope.

And as with any high frequency, the parasitic coupling affect the measured result, what I see as the biggest obstacle in the format of a meter with two separate leads.
So without checking the shape you would easily end up with wrong reading, so for reliable and accurate measurement you need to use the scope anyway - to ensure you do not have any significant coupling into the measured signal, spoiling the result...
So if you do not need a certificate for your measurement, the scope + math combination is way more affordable solution...

Was looking for this subject and found this topic so I won't start a new one.
So how about measuring current?
Will those readings be unreliable too? (Either with probes or a clamp meter)
I'm currently testing the following setup:

Generic electronic ballast rated
Lamp power: 55 - 95W
Lamp current: 0.8 - 0.85A

SOX 90W specsheet:
Nominal 91W 112V 0.82A

This is what I measure with my multimeter after about 30 minutes warm up:
180V
0.45A (measured through multimeter and clamp meter)
The ballast is connected to a wattmeter displaying 83W

Can the readings for voltage and current be this far off because of the high frequency?
The ballast clearly states 0.8 - 0.85A lamp current.
The power reading is correct though. So what if the voltage is indeed this high and the current this low? Will it damage the lamp?

There are a number of gallery threads where member Globe Collector gave details on how to construct a rectifying circuit (a current transformer?) for measuring HF current going through lamps using an ordinary 50hz/60hz meter.

He had a specific diagram in his gallery showing the number of wire turns on a ferrite ring and the necessary components (a resistor + capacitor from what I remember), plus the simple math to convert the results to meaningful mAmps. But he deleted this image + info.

Discussions remain, however, in a few of my gallery posts, typically in the mid to bottom of the replies:

https://www.lighting-gallery.net/gallery/displayimage.php?pos=-116708
https://www.lighting-gallery.net/gallery/displayimage.php?pos=-116711
https://www.lighting-gallery.net/gallery/displayimage.php?pos=-163738

Thanks, that was an interesting read but some things way over my head.
But if I understand correctly even if a power meter says you're at the right wattage, if the lamp voltage is higher and current is lower than the given specs of the lamp it's still underdriven? I only see people talking about current, never voltage.

The current of a discharge lamp is the critical factor. Voltages can vary within certain parameters, but it's the current that is important. That's no doubt over-simplifying things rather a lot, but the current is what matters.