Duty cycle when driving a CRT TV flyback transformer

[John123]

Hi, new here but I've dabbled with this sort of thing in the past but I'm a little rusty.

When driving a CRT TV flyback transformer externally what would be the optimum duty cycle for arc drawing? I've read all sorts on 4hv.org (RIP) from they work best at 20%-30% all the way to 90%! (https://4hv.org/e107_plugins/forum/forum_viewtopic.php?180096),

Your flyback will work better in the 10-20% duty cycle range. You're assuming flybacks work well with 50% duty and that just causes a lot of heat buildup in the switches (Mosfets).

Frequency is 15.7khz and I'm using that robust current mode control driver based on the UC3844 by Jan Martis. Obviously the duty cycle will change with the current feedback but I'm still curious.

What kind of duty cycle would they run at in the TV? The answer to this question doesn't seem to show up on google.

Thanks.

[davekni]

If it's a TV flyback (not from a computer monitor), the two common standards have very similar horizontal timing (NTSC in the US and PAL in Europe).  I recall PAL at the moment: 15625Hz horizontal, 64us total, of which 52us is active picture and 12us is blanking (retrace) time.  A high-quality TV showing the entire picture would have slightly less than 12us retrace time, say 10us out of 64 total.  Cheaper TVs often used the entire time and cropped the image edges a bit in the process.

Are you defining duty cycle as the drive transistor on-time?  Presuming so, then normal TV use would have it on for 52-54 out of 64us, or 81-84% duty cycle.  For non-TV use, the optimum depends on many factors, so there's no one correct answer.  It depends on frequency, DC input voltage, transistor voltage capability, and the nature of the desired arc.  Low voltage at high current needs wires closer to start an arc, but can sometimes stretch to longer after started and be thicker.

Sine waves from a ZVS circuit can work well if you are after high current at a lower voltage (ie. 50-60% of rated flyback output voltage). ZVS may be better driving a a user-added primary winding on the exposed ferrite core leg (lower transformer coupling factor), with a center-tap on the primary so the ZVS has only a single inductor for power feed.

For driving in flyback mode, my preferred topology is one with a current-sense resistor in the source/emitter of the drive transistor, so each drive period lasts until the current reaches the flyback's rated input current.  If spec's aren't available, a cap discharge ring-down test can determine the core saturation current.  The spec current is less than that, perhaps 60-80% of saturation.  If the core gets over 100C, then it's being driven too hard.  60-80C would be safer to account for hotter inside the windings.  This topology auto-corrects duty cycle for maximum current.

[John123]

Thanks for the help Dave, I've got a few further questions which I've highlighted in bold to make them easier to spot.

If it's a TV flyback (not from a computer monitor), the two common standards have very similar horizontal timing (NTSC in the US and PAL in Europe).  I recall PAL at the moment: 15625Hz horizontal, 64us total, of which 52us is active picture and 12us is blanking (retrace) time.  A high-quality TV showing the entire picture would have slightly less than 12us retrace time, say 10us out of 64 total.  Cheaper TVs often used the entire time and cropped the image edges a bit in the process.

Are you defining duty cycle as the drive transistor on-time?  Presuming so, then normal TV use would have it on for 52-54 out of 64us, or 81-84% duty cycle.  For non-TV use, the optimum depends on many factors, so there's no one correct answer.  It depends on frequency, DC input voltage, transistor voltage capability, and the nature of the desired arc.  Low voltage at high current needs wires closer to start an arc, but can sometimes stretch to longer after started and be thicker.

Thanks these were the exact numbers I was looking for, it's one of those black line output transformers from a CRT TV. I was just going to try and keep the frequency the same as what it was in the TV so 15khz is right. Does core saturation current stay fixed no matter what DC input voltage, frequency and number of turns I use? 36v and external windings on the core would be easiest for me to supply at any decent power level.

Just out of interest, what do the CRT PC monitor line output transformers usually run at? They often didn't put out as much IIRC.

Sine waves from a ZVS circuit can work well if you are after high current at a lower voltage (ie. 50-60% of rated flyback output voltage). ZVS may be better driving a a user-added primary winding on the exposed ferrite core leg (lower transformer coupling factor), with a center-tap on the primary so the ZVS has only a single inductor for power feed.

For driving in flyback mode, my preferred topology is one with a current-sense resistor in the source/emitter of the drive transistor, so each drive period lasts until the current reaches the flyback's rated input current.  If spec's aren't available, a cap discharge ring-down test can determine the core saturation current.  The spec current is less than that, perhaps 60-80% of saturation.  If the core gets over 100C, then it's being driven too hard.  60-80C would be safer to account for hotter inside the windings.  This topology auto-corrects duty cycle for maximum current.

That cap discharge ring-down test looks interesting, didn't know they were a thing. What would be a typical ballpark saturation current for 10 windings externally wound on a gapped flyback transformer core?



This is what I've used in the past for a reliable true flyback mode driver but with a 600v MOSFET and smaller primary capacitor to allow it to ring up a bit more and give more output voltage. This thing was a beast and pretty reliable once a good layout and miller clamp was added to the MOSFET gate, I could use just 12v DC and tune the primary coil turns with a low switching frequency to get arcs to ignite over a 7cm gap! It wasn't even specific to one flyback transformer, all of them were capable of this but sadly only a few could survive such treatment. One transformer from a CRT monitor even shorted its internal capacitor with this punishment. All this whilst the peak drain voltage was kept within MOSFET limits with plenty of headroom.

Over the years I've also played with most of the popular topologies, ZVS gave me over 1ft arcs with 48v and bridge lets you hit the magic frequency where corona covers the whole thing and shorts the internal diode and audio modulation was fun too.

Anyway enough of the good old days as all I've got to show for it is a box of destroyed flybacks, these days I'm more into designing things within their design limits. Do you have any reliable flyback mode driver schematics you can share? I'm guessing there are probably much better IC's out there these days, Are there any newer go-to MOSFETs you could recommend for this too? Back in my day IRFP260 and 460 were all the rage!

[Mads Barnkob]

Thanks these were the exact numbers I was looking for, it's one of those black line output transformers from a CRT TV. I was just going to try and keep the frequency the same as what it was in the TV so 15khz is right. Does DC input voltage affect how low a frequency I can go and does core saturation current stay fixed no matter what DC input voltage I use? 36v and external windings on the core would be easiest for me to supply at any decent power level.

Just out of interest, what do the CRT PC monitor line output transformers usually run at? They often didn't put out as much IIRC.

This dates back from 2013, which is close to the last time I played around with TV/PC monitor flybacks.

http://kaizerpowerelectronics.dk/high-voltage/tl494-flyback-driver/

And to quote myself from that article: "Flyback transformers from a CRT TV are typically driven at 15 kHz and flyback transformers from computer monitors are typically driven between 30 to 150 kHz."

[John123]

Thanks these were the exact numbers I was looking for, it's one of those black line output transformers from a CRT TV. I was just going to try and keep the frequency the same as what it was in the TV so 15khz is right. Does DC input voltage affect how low a frequency I can go and does core saturation current stay fixed no matter what DC input voltage I use? 36v and external windings on the core would be easiest for me to supply at any decent power level.

Just out of interest, what do the CRT PC monitor line output transformers usually run at? They often didn't put out as much IIRC.

This dates back from 2013, which is close to the last time I played around with TV/PC monitor flybacks.

http://kaizerpowerelectronics.dk/high-voltage/tl494-flyback-driver/

And to quote myself from that article: "Flyback transformers from a CRT TV are typically driven at 15 kHz and flyback transformers from computer monitors are typically driven between 30 to 150 kHz."

I remember your website now, just came flooding back to me. All this high voltage stuff is like a blast from the past, the single ended design with a 555 timer was the bane of my existence in 2011 lol.

[davekni]

"Does core saturation current stay fixed no matter what DC input voltage, frequency and number of turns I use?"

The input voltage and frequency affect the actual current, but not the saturation limit.  Turns does affect saturation current, inversely proportional.  In other words, the saturation level is ampere-turns.

"What would be a typical ballpark saturation current for 10 windings externally wound on a gapped flyback transformer core?"

One flyback transformer I have is ~38T on the primary, with saturation around 1A.  So, a 10 turn coil would have ~3.8A saturation current.

The circuit you posted looks fine - controlled by current limit.  My circuits are mostly made of discrete transistors, so not something anyone would want to copy.

[John123]

One flyback transformer I have is ~38T on the primary, with saturation around 1A.  So, a 10 turn coil would have ~3.8A saturation current.

That's weird, I would of assumed saturation current to be lower with less turns. How the come saturation current goes down with more turns?

[davekni]

If you wind one turn around a core, and send 1 amp through it, it creates one ampere-turn, which induces a specific magnetic flux, depending on the core geometry/gap and material properties.  If you wind a second turn, with one ampere, then there is a total of two ampere-turns, which creates twice the magnetic flux.  That's two turns with one ampere.  If you go back to a single turn and send two amperes through it, that's another way to get the same two ampere-turns, so the same flux.

Perhaps what you are thinking of is the volt-second rating of a winding.  That does go up linearly with turn count.  A typical flyback core may have a volt-second rating of 50uVs/turn (50 microvolt-seconds per turn).   Say it also has a rating of 1uH/turn^2 (one microhenry per turn squared).  A one-turn winding will have 1uH.  Applying 50V to it for 1us will ramp current to 50A, which would be saturation (50uVs).  A two-turn winding will have 4uH.  Applying 100V to it for 1us will ramp current to 25A, again saturation (100uVs / 2 turns = 50uVs/turn).

Hope this helps.  If not, please feel free to ask again.

[John123]

If you wind one turn around a core, and send 1 amp through it, it creates one ampere-turn, which induces a specific magnetic flux, depending on the core geometry/gap and material properties.  If you wind a second turn, with one ampere, then there is a total of two ampere-turns, which creates twice the magnetic flux.  That's two turns with one ampere.  If you go back to a single turn and send two amperes through it, that's another way to get the same two ampere-turns, so the same flux.

I think I'm starting to get it now, so when people use those rectified mains drivers with lots of primary turns they're really just spreading less primary current over more windings. Say 300mA over 70 turns vs a couple of amps over less turns, this gives the roughly same amount of flux?

[klugesmith]

Yes. Dave explained it very well in previous post.

To carry the message one step farther,
consider the 3-D space occupied by some coil of wire in a transformer, electric motor, electromagnet, etc.  The coil designer chose a wire gauge and turns count, to fit some design point in voltage, current, and magnetic flux.  There's an associated winding resistance, inductance, and power loss.

As a rule,
the coil can be rewound with different gauge wire to change the voltage-current design point. Designing for higher or lower voltage does not change the magnetic flux, coil volume and mass, power dissipation, or electrical efficiency.  (Magnet wire is available in odd AWG sizes and even 1/2-AWG sizes, allowing designers to adjust the voltage-current tradeoff very finely.)

Here's an example in the direction you talked about, to increase working voltage and decrease working current.   Like Dave's numerical examples with real ampere-turn and microvolt-second values.

Reference winding has a voltage, current, resistance and I^2*R power loss, magnetic flux F, turns count N, winding cross section A, etc.  Its interaction with the core, armature, yoke, etc. can be represented in terms of ampere-turns and volts per turn.

Now rewind the coil using wire with half the cross-sectional area. 
It will take twice as many turns, and twice as much wire length, to fill up the winding volume. 
Coil resistance and inductance will be 4 times larger than the original coil.
For magnetic flux to be same as original, ampere-turns and volts per turn are same as original.
So the new coil wants twice the voltage and half the current as original.
You can see that the power losses, inductance energy at saturation, maximum volt-seconds per turn for driving circuit, etc. are all the same as original.

[edit] Here is another angle from which to look at it.
A designer could start by considering each winding as a single turn with cross-sectional area A.  The shape might be rectangular or irregular to fit the allocated part of window through core.  That coil's terminal voltage and current will be identical to the volts-per-turn and ampere-turn values from magnetic circuit design.
Step 2: Divide the winding area A into N equal sections, and connect all sections in series, as in a practical  coil.  That increases the terminal voltage, and reduces the terminal current, by factor of N.
Wire mass, current density, power loss, etc are not changed by the N-way partitioning. Except for fill-factor effects when using round wire, insulation between turns, etc.
 

[John123]

Wow, lots of good information there for my brainlet brain to take in. Might take a while but its slowly sinking in I think.

[jturnerkc]

Wow, lots of good information there for my brainlet brain to take in. Might take a while but its slowly sinking in I think.

Ya, I've got to let this marinate as well.

Thanks for all the insight, all, and good question John!

[John123]

Ya, I've got to let this marinate as well.

Thanks for all the insight, all, and good question John!

I think a practical exercise is in order, sometimes they help spark a lightbulb moment where the theory just clicks into place.

I'm not sure about you but in the past winding coils were just a case of winding a few turns and adding and removing some to tune it, never really gave much thought into the deeper theory behind it. Like I'd heard of saturation but just assumed something would go bang if that was the case.