How to measure voltage output of AC flyback transformer

[MRMILSTAR]

My next project is going to be a large Cockroft-Walton voltage multiplier. I want to use an AC flyback transformer as the HV power supply. Does anyone know of a method for measuring the voltage output? I would also be interested in knowing how to measure the output of a DC flyback transformer.

[davekni]

Before buying a used HV probe, for AC, I'd use a home-made capacitive divider.  Make a small linear HV capacitor, such as foil on the inside and outside bottoms of a small glass jar (ie. baby-food jar).  Place that from your HV AC source to a scope probe, with a larger value (say 1nF) cap from probe to ground.  Calibrate the setup with some lower-voltage AC source that you can measure directly and through this capacitive divider.  (Either adjust capacitors to get a nice ratio, or just do the math on the resulting waveforms.)

For DC, my solution is similar.  Soldered a string of 100 1meg resistors together, then fed that into a polyethylene tube for insulation.  Later I bought some 10kV rated resistors, so my string isn't as long.  I use 30 of those in a string on the output of my multiplier so that the multiplier's stored energy isn't a hazard to the public when they are making their hair stand on end or whatever.

My finished multiplier has too high an impedance to measure accurately even with a 100-resistor string.  Instead I use small aluminum disks of varying thickness on top of the 10" (254mm) ball.  When the disk just lifts from the ball, its weight/area and the ball diameter can be used to calculate voltage.  I have ran it to 250kV, but typically run around 100kV for public static-electricity activities.

BTW, I initially used a 20kHz transformer to feed my multiplier, but the cheap ceramic capacitors were too lossy (just feeding stray capacitance of the multiplier string), so some overheated and failed.  Changed to a spark-coil running at 700Hz.

Here's my schematic. less the 30 x 10meg output isolation resistor string:


And a picture of the finished multiplier with ball at top:

[MRMILSTAR]

I have one of those commercial 40 KV DC probes with an analog voltmeter integral to it. I was thinking of using a full-wave bridge and a HV capacitor to convert the AC to DC so that I could measure it with this probe. Does this sound like a viable way to measure the flyback output voltage?

[davekni]

Yes, that will work fine too.  Of course, you'll get only the peak voltage, no waveform shape information.
For a flyback, you could half-wave rectify using a single diode.  That way you can separately measure the positive and negative peak voltage.  Flyback transformers don't mind that bit of DC load.

Make sure the diode is fast enough and low enough capacitance to not affect results significantly.  I've had good luck with the little 5mA 20kV 100ns 2CL77 diodes from China:
    https://www.ebay.com/itm/20pcs-5mA-20kV-High-Voltage-Diode-HV-Rectifier-2CL77/290637855659?hash=item43ab5dbbab:g:PgYAAOxyLiJR1nm7
For low currents, they work fine in series to get higher voltage.  They need to be taped or potted (generic epoxy or silicon rubber caulk works fine) to reach 20kV w/o arcs along the package.  Tape is fine for short-term uses.

[klugesmith]

Beautiful work there, Dave!

Rectifier doesn't need to be high voltage if it goes on the low side of voltage divider.
Before the divider's bottom R, so the rectifier's forward voltage drop subtracts from original high voltage instead of the divided voltage.  It's not so simple with a capacitive voltage divider.

Here's an AC/DC kilovoltmeter I made last year for characterizing NST's. 
* Started with a 10-resistor HV string acquired at the flea market. 
* Bridge rectifier part was first verified to have leakage current much smaller than its datasheet spec.
* Needs capacitor for AC operation, because panel meter is poor at rejecting 120-Hz ripple voltage.
* Both sides of panel meter circuit are not electrically common with either side of the HV input.


The floating AC/DC mA-meter in picture is another story.  There's a bridge rectifier inside box, between external terminals and current shunt resistor. 

[davekni]

Thank you for the compliment.

That resistive divider works well at line frequency (50/60Hz).  It also allows the option of scoping the divider output w/o diodes to see the AC waveform.

At higher frequencies as in typical flyback transformers, stray capacitance across the HV resistors can give falsely-high readings.  Accurately handling that capacitance is why HV scope probes cost more than HV meter probes.  HV diode(s) directly on the flyback output avoid that stray-capacitance issue.  The resistive divider sees only DC, so stray capacitance is not an issue.

Also, I expect the desired information for the flyback output is peak voltage, not average or RMS.  Direct HV rectification with HV capacitor measures peak voltage.

[MRMILSTAR]

Peak voltage is all that I am interested in so that I can determine the proper maximum voltage rating for the capacitors.

[klugesmith]

For the nominal Subject of this thread, qualified by "Peak", I stand with Dave.

With Device Under Test sharing a nice safe common ground with the instrument, you can use
* Half-wave rectifier with HV diode of sufficient speed and voltage. 
* HV capacitor to hold the peak voltage (positive or negative depending on diode orientation)
* HV meter, for example that analog TV service meter (need to reverse internal wires to read negative HV).

Diode needs to withstand the peak-to-peak voltage on transformer secondary.  For example, if capacitor holds a positive peak value of +12 kV at a moment when secondary is at -12 kV, there's 24 kV across the diode.

Positive and negative peak voltages have very different magnitudes in conventional flyback converters.
To feed a voltage multiplier, don't you want a 50% duty cycle, balanced voltage waveform?
We see lots of circuits with DIY center-tapped primary windings on a magnetic part that looks like a flyback "transformer".
In those circuits it's operated as a real transformer, and I think would often work better with no air gap in ferrite core.

[MRMILSTAR]

Indeed I am thinking of using an AC flyback transformer with a 10-turn center-tapped hand-wound primary coil on one leg of the core. The primary coil would then be powered by a ZVS driver. I also have a NST that I could use but I am more interested in the flyback arrangement with battery power for portability.

[davekni]

Sounds like fun! Since you will be driving the flyback transformer with a sine-wave, either full-wave bridge or the single diode into capacitor will work fine, giving the same result.  BTW: I'm presuming you're using an existing high-voltage winding and don't know its turn count, so need to measure?  Or, is it just to get more precision than a theoretical value?

Center-tapping the primary winding is great for Royer oscillators - single smaller inductor for power feed.

This is surprisingly related to my current project.  I've recently become intrigued with driving inverters with ZVS (Royer) oscillators.  A lot of simulation and some test results for small Jacob's ladder driving.  Is your flyback transformer already glued together, or do you have the ability to adjust the gap (inductance)?  If not adjustable, can you add a little extra primary series inductance to adjust net coupling factor?

The key issue is keeping the Royer oscillator running over the full range of load, including the initial short-circuit condition at startup when the load capacitors are discharged (or when the arc first strikes in my case), then remain running as the load voltage increases.  In simulation (and roughly verified experimentally), a transformer with <86% coupling factor works, and >86% doesn't.  I've adjusted my core gaps for 83% coupling - a bit of margin from 86%, but still good energy transfer.

With shorted load (ie. initial startup), the Royer oscillator frequency is set by the leakage inductance.  At light load, it is set by the normal transformer inductance. Between, where maximum power is transferred, it is between those two frequencies, and a bit distorted.  With coupling above 86%, it will function shorted (at high frequency), but drop out as the load voltage increases.  The Q becomes too low to sustain oscillation.

Have fun!  I'm anxious to hear how your project goes!

[MRMILSTAR]

I found several AC flyback transformers at a Teslafest recently for very little money so I bought 4 of them. One of them is quite large and is the focus of my attention right now. Unfortunately, I can find no manufacturer name on any of them, only some meaningless  long number. So I don't have any data sheets for any of them. I need to figure out a way to estimate the peak output voltage so that I can properly size the capacitors.

I plan using some 1.6 nF, 30 KV TDK doorknob capacitors that I have. If I can't figure out the flyback peak voltage I may have to put 2 of the caps in series for a 60 KV rating for each stage just to be safe.

[davekni]

Wonderful to have some AC flyback transformers!  I'd love to find a few.  Home-winding the HV side is difficult - to get low capacitance and high voltage w/o corona and arcing.  The HV potting info on this forum look interesting, but purchasing is easier.

A diode, capacitor, and your 40kV probe will work fine to measure peak voltage.  No concern about needing to guess.  (Of course, all the capacitors in the multiplier, except the first one, will see the peak-to-peak voltage, so twice the peak.)

If you have a scope available, measuring the secondary turn count would allow calculation of the primary turns needed to get 30kV peak-to-peak, so you can avoid needing to series-connect your caps.  Wrap 10 or 20 turns of scrap wire around the core.  If any source of kHz voltage is available (signal generator or other random project), connect that to the secondary.  If not, just charge a cap, then touch it to the secondary for a ring-down waveform.  Scope both the secondary (being used as a primary here) and your 10 or 20 turn coil to get the voltage ratio.  This is almost always the way I start with any unknown transformer.  (I prefer the ring-down, as it provides inductance also, from frequency and capacitance.  I use a TRIAC to connect the cap and coil for cases over 200V, as it makes a cleaner connection than just touching wires together.  Repeating with larger capacitance until the ring-down frequency isn't constant with voltage will provide saturation current.)

[profdc9]

I built a simple peak voltage meter this way.  Sorry, don't have schematic handy right now.

I used a 50 microamp analog DC current meter.  I made a full-wave-bridge of 4 1N4007 diodes into the DC current meter to rectify the AC to DC.   Then I made about a 600 megaohm resistor by putting 33 20 megaohm resistors in series and place this in series with the full wave bridge.  To prevent end-to-end arcing breakdown between the resistors, I laid them out in an undulating pattern and potted them all in epoxy.  This was able to read up to 30 kV peak, but you could add more resistance for a higher range.

[davekni]

I used a 50 microamp analog DC current meter.  I made a full-wave-bridge of 4 1N4007 diodes into the DC current meter to rectify the AC to DC.   Then I made about a 600 megaohm resistor by putting 33 20 megaohm resistors in series and place this in series with the full wave bridge.  To prevent end-to-end arcing breakdown between the resistors, I laid them out in an undulating pattern and potted them all in epoxy.  This was able to read up to 30 kV peak, but you could add more resistance for a higher range.

If I understand your circuit, it's measuring average rather than peak.  With the diode choice and stray capacitance of the resistance string, it would work well at line frequency, but not at typical flyback frequencies.