Dangers/risks of Mazilli ZVS driver?

From your post I'm not certain as to what you're planning on powering... is it a flyback transformer or an induction heating coil? I have built essentially the same circuit and used it for an induction heater, with moderate success. I was able to heat nails to redness in around 10-15 seconds, and slightly larger metal pieces to orange heat with longer heating.



The actual ZVS driver is mounted to the right of the capacitor array. I soldered a copper tab onto each pipe, drilled a hole and then mounted each MOSFET on the tabs. This way the water coolant served to keep both the MOSFET and workcoil cool without large heatsinks.

I used a slightly different circuit configuration (see schematic), with 2 inductors instead of 1 to avoid center-tapping the work coil. I would recommend this configuration.

I can post more info on the capacitors and # of turns, etc, if you are interested. Do you have a high-current capacity power supply? I used a rewound microwave oven transformer. To take advantage of the full potential of this design you will need around 20A current.

I did not have problems with things exploding like you mentioned, However, with larger workpieces the setup sometimes began to drift off resonance, causing a huge surge in current and loud buzzing noise. In these instances I could usually save the circuit by unplugging it immediately. For this reason I would not recommend the circuit for higher power applications... I could never draw more than ~10A without running off resonance, but that's probably due to the way my LC tank turned out.

This is a fun project, but if you want more useful output I would go with a half bridge/full bridge configuration (http://uzzors2k.4hv.org/index.php?page=pllinductionheater1). Also, I will say the ZVS driver works for flybacks, but again I had the same problem with drifting off resonance...

**Also per your question about current without load, it drew only about 0.5A without a workpiece. As progressively larger workpieces were introduced into the coil, the current draw increased to a~10A. Parallel LC tanks will always work this way (my configuration), but if you use a series configuration there will be a large current draw without load.

[Edited on 30-6-2014 by bob800]

I don't plan on posing this way normally, but I'm playing with my new surface pro.









If I have some time I'll throw this together and see if it works... or someone else "knowledgeable in the art" can try it.

Quote: Originally posted by woelen

I also can obtain IRFP260 devices for less money. What is your opinion on those?

Quote: Originally posted by woelen

I also saw IRFP264 transistors from Chinese sellers, for less than 1 euro per piece and free shipping. It sounds too good to be true. Are these good quality components?

Another thought after studying the circuit is if one does not properly consider the switching speed of the two diodes there is a danger of the Mosfets conducting destructively. I suggest you consider the MUR 460, a 600V, 4A ultra-fast diode. Just a thought anyway.

Found a couple pics of the old chopper transistors I mentioned in another post. One is PNP Ge and the other is NPN Si, included it because case style is correct and it's hard to find images of these old devices. AFAIK only the PNP Ge types were used in the HV supplies for tube mobile gear. Also used in the later 50's Delco-Remy car radios to get away from the old vibrator supplies. So if your junking an old Cadillac save the AM radio those chopper transistors are becoming extinct.

Going for voltage, this is one option:



The switching transistor can be salvaged from an old PC supply, usually a C2625 or something like that. +V can be the same +12V, or anything else. (Actual CRT TVs and monitors use 100-160V, or probably 320V in 240VAC countries, and an internal winding; just put on your own winding, make sure of course you find a flyback transformer that you can still get wire around the core.)

For dumb oscillator purposes, i prefer a variation something like this (you can ignore the top half of the circuit):



The 0.1 capacitor between transistor drains should be selected to resonate with the primary at the same frequency as the secondary.

Tim

Woelen your question was about "Dangers/risks"

Being too impatient to wait for the parts on order, I decided to try the circuit using some 500 volt Mosfets which are rated at 4 amperes I had on hand. I built the circuit to operate at lower current than the specs posted in this thread. Not only does the circuit work well, it imparts greater energy density into the windings than other circuits I have tried over the years, when considering the power input. Meaning efficiency is far higher than your 'typical' designs. In fact the design kicks ass fairly well, I like it. Waiting to get my orders and build a much higher current version. Upon power up a virtual short circuit exists for both fets. As the current inrush builds magnetic field in the coil the back EMF shuts off the opposite fet from each end of the coil. If it did not destruction is impossible to avoid. When the 'ringing' voltage is large the .68 uF capacitor will destruct if it's dissipation is high, or voltage rating is low. If a short circuit occurs from insufficient back EMF to shut it off every cycle (or if the fet shorts from over current or over voltage), the fet explodes. The greater the energy input the 'bigger the bang' obviously. So I see two dangers. One is simple shrapnel from either a fet or the capacitor. Avoided very simply by a shield to keep a high velocity chunk from putting out an eye.

The second danger is electrocution as the power level at high voltage (back EMF and/or transformer secondary) is quite high. That's it, nothing to worry about if safe procedures are followed combined with common sense. A chunk of Mosfet at firearm velocities can be very dangerous yet minor shielding is adequate since the mass of the plastic pieces is low.

If you look at inductive heating they typically use 8 turns center tapped of Cu tubing around 6" in diameter. Going by pictures in some of the ebay auctions for factory built driver boards. For a circuit using a transformer primary they use 10 turns CT on a ferrite core. Being proven to work I would call this the 'starting value' of the end design. So calculate the 'safe' inductance, using the values shown to work. If L is too low not enough back EMF will develop to shut off the Mosfets guaranteeing destruction of the devices. If L is too large the excessive back EMF is going to explode the capacitor or cause a destructive short in one or more Mosfets. It is as simple as this. Which also includes consideration of the voltage/current input from your supply obviously. Start with proven values for the turns, and for current draw since this will cause either over current destruction of the Mosfets or again excessive back EMF to develop. I imagine most having problems with parts exploding are either running too much power into the circuit or not considering the load and the needed back EMF to turn off their Mosfets, i.e., using too few turns. Or as stated too high of a back EMF.

An oscilloscope would be useful to monitor the back EMF so one can calculate proper component ratings for safe operation. Another thing is looking at some of the factory made units I see on ebay, they are using 1 uF not the 0.68 uF indicated in this thread. I am sure a range of values will work here, worth considering the frequency of operation since for example many TV flyback transformers operate with greatest efficiency in the 15 KHZ to 17 KHZ range. Playing around with the frequency of operation may be useful when considering what it is you plan on driving with this circuit. I am interested in the use for inductive heating. Since I imagine Tim (12AX7) has more knowledge than other members about inductive heating circuits I wonder what he might have to say about what the best frequency of operation is. Also what is the optimum inductance for the Cu tubing coil. I am sure there is an optimum range for L/C ratio and wonder what his thoughts are about this.

"- How does the driver behave when there is no load? Does it just draw a small current from the power supply in that case?- If I connect a transformer to it (e.g. 5 + 5 windings primary, 2500 windings secondary) and have no load attached to that secondary, how does the driver behave in that case? Actually, I really would like to limit the current capabilities of the driver in my first experiments. If I limit the current with a big power resistor of e.g. 3.3 Ohm in series with the 12 V power supply, does it still work, albeit with lower energy?"

If you study the circuit the Mosfets are biased fully on by design at power up. If no windings are connected to the Mosfets there is no drain current. But if there is, the circuit is going to run at fairly high power depending upon power input, DC resistance and developed back EMF. It may merely be generating a large magnetic field and high voltages from back EMF and loading will increase power draw, but I do not see how this circuit is going to 'loaf' at little power just because you have not loaded it. This design is not going to behave as a simple transformer which is not loaded on it's secondary. You will no doubt see a range of power draw depending upon how much resistance you put in series with the input and how heavily you load it. However I think you must consider the fact that if your developed back EMF is too low to fully shut off the Mosfets, basically you have two turned on devices presenting a dead short to the power input over too much of the cycle or just plain 'on' meaning a dead short all the time. From this one can conclude even at light loading it must still be running at a high enough power level to create the needed back EMF to turn off the devices and maintain nonstop oscillation. I imagine you need to consider this in your design. Operating at low voltages is possible so long as you keep the circuit in oscillation at all times when powered.

For clarity my last comments pertain to looking at both the circuit in the starting post, and the schematic I borrowed from the auction listed below.

http://www.ebay.com/itm/12V-24V-36V-Flyback-Driver-Board-Zer...

zvs driver

Quote: Originally posted by froot

Another consideration for overheating mosfets is the rise and fall slew rates of the pulses at the gates of the mosfets. The faster the slew rate the less heat the mosfet will produce. Ideally the mosfet must be fully on or fully off in this application so anything other than vertical lines and horizontal lines above Vgs(on) in the Vgs waveform translates to hot transistors. A perfect square wave is your goal which can be suprisingly difficult to achieve.

Here I have chronicled some of my meager forays into the domain of inductive heating. Hopefully this will leave everyone amused, and maybe it will even be a little informative.
Buried in a dusty corner of my garage, I too have a coil of copper refrigerator tubing, bought back in the day when silver was $4 an ounce. I could use it, but that would be boring. Everyone else is using it, after all. So I decided to use copper wire. Not only copper wire, but braided enameled copper wire. Yes, I made my own watered-down version of Litzendraht.

I used 32 AWG magnet wire, with heat-strippable insulation. It’s bright red in color. If you use the other stuff, you will regret the day you thought to try this, as you will be stripping each wire individually. No fun.

A couple of empty spools sacrificed their lives for the cause of science, as seen below. Six holes were drilled around the periphery of the spools to enable the wire and springs to pass through.




A couple of springs were set up for tension on the center conductor:




And a copper “washer” was soldered to a length of copper pipe on a hot plate.



Here the pipe is inserted through on one of the spools, on which it rotates freely. The pipe is clamped into the vise behind the spool. The washer on front of the pipe keeps the spool captive, and the springs maintain tension on the wires. Here we have a center conductor, with six wires spaced around it.



At the other end the wires are threaded through the spool, and twisted together at the ends. The spacing between the twist and the spool should be maintained at about 7”. This is easier if you have a friend to help you with this, as the wire that I used was 9’ long.




If you do it by yourself, you’ll have to walk back and forth every five turns, to keep the spool spaced properly (ask me how I know).
Using a clever trick of photography, I made it look like I actually did this!




Actually, I did do it; it only took about 20 minutes to twist all nine feet of it. It didn’t take as long as I thought it would. In fact, it was so much fun that I made two of them!




To be continued...

Now here is the next step. I took each one of these wires, and folded them over twice, cutting the loops at each end. This left a total of eight strands, four from each wire. Then, four strands each where taken and braided into approximately two foot lengths, leaving two braided cables of four strands each. If you have a younger sister, get her to do this for you. Pay her if you must. She will be way better at this braiding thing than you. As for me, I have a trained feline. She does a good job, and I can bribe her with tuna.




Maybe you can also see why it’s important to use the heat-strippable version of wire. There are 28 strands in each cable.
After all that work we can have a little bit of fun now. Both cables were soldered together in parallel, and wound around a dowel, both cables side-by-side. This left a nice (but floppy) inductor, with some rather nice specs:

Ls=0.83uH , Rs=6.9mohms @ 1kHz
Ls=0.83uH, Rs=9.6mohms@ 100kHz

At 100 kHz, we see that the AC resistance is only 0.0096 ohms, which is only marginally higher than the 0.0069 ohms at 1kHz. Normally because of skin effect the AC resistance increases dramatically, but here it does not, because the individual strands are thin enough for the frequency (32 AWG at 100kHz), and each strand spends an equal amount of time in the cable as it does on the outside edge, cancelling the circulating currents in the wire.





Because this coil structure is too soft, a ribbon was cut out of copper sheet, and the braided wire was tied on with pieces of enameled wire. Each end of the coil was soldered to the copper ribbon, and the finished coil was tested.

Ls=0.60uH, Rs=2.9mohms @ 10kHz
Ls=0.55uH, Rs=8.9mohms @ 100kHz

Here I goofed and measured at 10kHz instead of 1kHz, but still the dramatic affect of having the copper ribbon in parallel with the coil is illustrated. Low frequency AC resistance improves because of the bulk copper in the ribbon, but at 100kHz the resistance is unchanged from the earlier coil configuration. This is because at 100kHz, the ribbon’s AC resistance is so much higher than the braided cable, it is electrically not part of the circuit.

The coil was installed in a basic type of circuit as seen below. It is driven only on one half-cycle, and uses a single FS70UM mosfet. The FET is driven directly from a signal generator, and is driven rather poorly, I should add. The signal looks more like a triangular wave on the gate, instead of the desired square wave.





At 10V on the supply, 0.3A is drawn at resonance (110kHz), unloaded. In a full-wave configuration, this would be about 0.6A, similar to what Cheddite Cheese is seeing. There is a key difference in the way our respective tank coils respond to a load, and when I have time, I throw some math out there to show why. It’s getting rather late right now, though.

Anyway, here I have the circuit operating in a beaker of DI water. The water has to be very clean, as you don’t want it to conduct electricity, only heat. With this type of a setup everything in the circuit was kept cool, even while dumping 30-40 watts into a piece of steel. The piece of steel that I was heating was sitting on a piece of firebrick contained within a test tube, which you can see in the picture. I’ll have to run some numbers, I may be able to bump this up to a 100 watts. Right now I’m limited by the power supply I’m using.

Quote: Originally posted by jock88

Hope they are not on the proverbial slow boat from....

Quote: Originally posted by papaya

Don't be excited too much - works only on ferromagnetics. copper will not get too hot for example - actually the effect is not from inductive currents heating..

Before I use my high power devices, I did some playing around with a miniature version of the ZVS driver, just to get some experience with this stuff before I start with the real thing, and I must say that this works quite well. If something goes wrong with this lighter/smaller stuff, then the damage is less severe.

I have some IRF540's lying around. These are not nearly as powerful as the IRFP250 and IRFP264, but they are decent devices in a TO-220 package. I used two BA159 diodes and 12V zeners (normal small zeners) and two 330 Ohm power resistors and a coil from an old power supply. The primary windings are 2x5 turns, around the core of a small flyback transformer from Nokia with unknown number of turns in the secondary winding and unknown primary windings (it has 10 or so connections, I do not use any of them, except one which is the ground end of the secondary winding).

There is no hissing and no corona discharge, but I can light a glass ampoule with neon gas in it and I also can light a similar glass ampoule with helium gas in it. The ampoules have a length of 10 cm and no internal electrodes, they are just sealed glass ampoules with some gas in them. I simply turn some metal wire around one end of the glass ampoule and around the other end and connect this with the two poles of the flyback transformer. This gives a bright nice orange glow for the neon ampoule and a fainter pinkish/orange glow with the helium ampoule. The ampoule is this one (I already have this some time, ordered from this chinese eBay seller as a nice and robust sample of neon):

http://www.ebay.nl/itm/281267028284?ssPageName=STRK:MEWAX:IT...

I never had it glowing before, but with this little experiment I could witness the nice orange glow.

This shows that I have at least a few kV, probably 5 kV or so, given the length (and the thickness of glass) of the ampoules.

I use small heat sinks with heat transfer paste on the transistors (20x30 mm) and when I have the glass tube lit for a few minutes, then these heatsinks are warm, but not hot.

These are my first little steps in home made high voltage drivers, more will follow with more powerful stuff

This makes 2 of us coming to this conclusion thus far. I wondered about it when I first looked at the design using IIRC 0.68 uF, before I tried the circuit myself. I am guessing your even higher C means yours is working better than mine. Guess I'll risk some Mosfets and see how high I can go while staying within reason. Also forgot to mention my factory board with my way too large tubing coil is running around 114 KHZ. Damn Cu tubing is getting expensive but I must get a 50' roll of some skinnier stuff. My 3/8' tubing sucks trying to coil a small diameter with 9 or thereabouts turns. Not being able to quickly obtain a roll of say 1/4 " and 1/8" cu tubing right now (car needs rear differential and it's too hot to put the used one I found in right now) I gave up on heating ideas and went back to flybacks for awhile. If you study the schematics Tim posted on page 1, these are what I call 'typical circuits'. The kind or similar I have toyed with for decades.

I will never go back to one like it again whether 555 driven bipolars, fets, u name it; or oscillator types using all of the previous or even choppers. Or whatever comes to mind. Even tube circuits. Done. I am so amazed with the efficiency of this ZVS driver. At 600 ma 14 V input I can draw an arc from 2 inches away which immediately goes to a metal melting thick arc with power input rising to an ampere at most. I have tried many variations with this same flyback at input currents of several amperes and overheating devices that produced neither the potential nor the power in the arc I am seeing right now. The efficiency and ability to transfer raw energy into a flyback at very low power input is beyond anything I for one have seen in simple low parts count circuitry. Using this driver circuit design for the heart of a stun gun would produce a very dangerous no longer non lethal self defense device of this I have no doubt.

Having taken many high end commercial models apart completely reverse engineering them in the past I am amazed not one has so far used the principle of the ZVS driver. They all use the oscillator type of what I now term forever the 'typical circuits' driving a spark gap and multiplier circuit. I am going to dig out my old 300 KV stun gun and rip the board out to replace it with a ZVS design for the base high voltage source with that driving the spark gap and multiplier. I get the feeling the output factory components will not be able to handle this new input even when run from a couple Alkaline 9 volt batteries. May well end up having to redesign and rebuild the gap and multiplier output sections as well.



Tossed this together on my bench quick and dirty just to take a couple pics. The flashlight served the useful purpose of stopping the big fat flexible HT wire from moving around. Circuit does a fair job setting my bench carpet on fire and the arc will go further still melting copper wire. At this time voltage input is 13.8 volts and I ~= 900 ma. Don't remember where I got these great flybacks, some surplus outlet where I bought a dozen at IIRC $8 each. Best ones I have found, easy to take apart core and create various circuit designs with. Great output in terms of high voltage and current.




[Edited on 7-24-2014 by IrC]

I reduced the images to 640x480 to avoid going past the post limit.

Seems interest in this thread has vanished lately, I wonder if woelen has built his ZVS circuit yet?

Anyway, the pics show a supply I built back in 2004. I would also post a schematic but after a decade I don't know where it is buried in my junk. Not really needed, anyone even slightly skilled in the art can see how straightforward this circuit is. To be honest a diagram is not really needed. Similar to what Tim posted a couple pages back and the supply is just a 2 watt 10K potentiometer (top 25 volts, center base, bottom ground) driving the base of a Tip3055 which drives a 2n3055, with a few capacitors here and there. Powered by a 19 Vac 12 ampere transformer into a 50 ampere 100 volt bridge (giving full wave output) which loaded at a few amperes gives a steady 25 volts DC output filtered by a few thousand uF. One 2N3055 is the low voltage DC power supply pass transistor, the other two each drive a flyback of the type in the top post on this page. Been 10 years since I looked inside this supply, at least I got my answer as to who made them it's on the top of each flyback (pic 7a). The circuit board: one Tip3055 drives the power supply 2N3055, allows me an adjustable zero to 25 volts which can do 8 amperes all day and probably 2 or more times that for shorter durations, until I lose the 2N3055 that is. The 555 circuit is on a separate Zener regulated 12 volt circuit, it drives another Tip3055 which drives the other two 2N3055's. You can see them on the circuit board each with a clip on heat sink. Each flyback primary is opposed in phase. The last pic is the type of He-Ne connectors I used, made by Alden.

Circuit ground and chassis ground are one, connected to the banana jack in front (and AC plug ground). From this I get half the maximum potential to each connector. One port on each plug (bottom) is the AC high tension from a flyback (~17 KHZ), the other (top) is from the multiplier. So from connector (bottom) to connector (bottom) I get double the AC high voltage on bottom, and double the multipliers on top. Each multiplier is enclosed in a cylinder of fairly thick plastic (rolled up 54 oz plastic soft drink cup). These worked well, never once in a decade have I ever had any arcing inside. I opened one up so one of two identical plexiglass mounted multiplier circuits is visible. All in all this has been for years a very useful 0-100 KV adjustable supply for various mad-sci experiments.

Showing the before as it is now. It works as I designed, was going for a zero to 50 KV to ground from each connector and double that between connectors. Just wanted to show the before as next I am going to take out the electronics and rebuilt this as a ZVS driven supply. I will parallel all three 2N3055's for the zero to 25 volt supply at greater current rating. Needed for a ZVS driver circuit. May take a few weeks in my spare time but I am curious to see the final output comparison between this 'typical' circuit and a ZVS version, all other things being relatively equal. Still giving thought about how to drive the flybacks. I want them running 180 out from each other so I can still utilize the double the AC high voltage output. While not important for the multiplier outputs, I want double the AC high tension from the flyback windings from plug to plug. If you look at the plug in pic 5a, the center is the multiplier and the shield is the direct flyback secondary out. The reason for this was so that I can use the supply to drive other multiplier (or whatever) circuits as opposed to using the internal multipliers.

Seems difficult to have two independent ZVS circuits synchronized. This leaves me thinking I should rewire the primaries, make them identical, drive them opposing in parallel from a single higher rated version of a single ZVS circuit. Any thoughts from people as to how I can best approach this concept?

Forgot to mention but the only arcing problem I had was the Alden plugs are not rated for 50 KV. I overcame this by cutting out a big rectangle in front, filling the space with a plexiglass window drilled for the connectors and epoxied in (makes it little ugly but no problem for me I just wanted a trouble free HV supply). The meter indicates zero to 25 volts output from the pass transistor (now all three in parallel as I have rebuilt it tonight since I no longer need the other two 2N3055's for running flybacks). The Alden plug on the left top comes from a negative multiplier stage and top right plug comes from a positive multiplier stage. In this way I could double the high voltage output. Basically a 100 KV supply grounded in the middle for a true dual polarity setup. I have on occasion found this arrangement to be useful in some experiments.

I write this while taking a break from it, I just finished removing all the 555 circuit components to give room to build the ZVS circuit on the same board. Saves time as I already have it setup with bottom bracket to mount to the floor of the chassis. For now I will try driving two 10 turn CT primaries (opposite phase on each flyback as I still want the AC outputs to double plug to plug) in parallel from a single ZVS circuit. Not really sure how this will work out but if it does it will save time and parts, space as well. Thought about using the factory made ZVS board but I decided to leave it for my metal melting coil. Dug out my 50 volt 50 ampere supply and even with my larger coil it works well enough. Have not gone above 25 volts, no point in blowing up such a nicely made board and it states 36 volts maximum.


[Edited on 8-3-2014 by IrC]

I added another 1 uF 630 volt cap in series to give 0.5 uF 1260 volt rating. Current flow dropped significantly. Oscillation starts at a lower voltage and output voltage increased quite a bit. No doubt the flyback cores were saturating badly. This is with both flyback primaries (each 16 turn CT) in parallel but 180 phasing. No way to measure the AC voltage between flyback secondaries but it is amazingly high. Much higher than before at 1 uF. The circuit runs cooler and more stable. Looking at the pics, the 260 fet on the right runs cool the left hand one slightly warm. Everything else runs cool except the three 2N3055's but they are heatsinked well and within temperature tolerances. The trimmer bottom right adjusts the voltmeter (was a watt meter in another life of the chassis). The transistor at bottom in clip on sink runs the three parallel 2N3055's.

At this point I must stop, I cannot find what I need to build two multipliers and encase them in oil while still all fitting in the cabinet allowing me to put the top cover back on. I need just the right tubes (to enclose multipliers yet hold oil), and capacitors, diodes with higher than 6 KV ratings.



Well my wandering the hardware and grocery stores did no good. Found some items with plastic tubes yet metal lids. Must be all plastic with screw on top so I can drain, rebuild multipliers if need be. Without gluing the end caps on. Must be 6 or 7 inches long and around 2 inches diameter, but no more so they will fit in cabinet. I guess I could encase them in RTV but self healing is important for long term reliability meaning oil is better.


[Edited on 8-4-2014 by IrC]

Being stuck on the dual supply until I can rebuild the multipliers I decided to alter another supply I built long ago to a ZVS driver design. This one outputs only the AC high voltage at around 17 KHZ. The first 5 pics are what it was, originally starting as a 21 amp power supply. I modified the LM723 circuit so the pot would adjust the output from 7 to 21 volts and used the old 555 circuit plus transistors to drive the flyback (which when together mounts upside down to the top cover). From the labels and holes you can tell it has been quite a few different projects over the years. I kept one RCA jack for frequency and waveform measurements.



Pic 6 shows the flyback primary rewired from 28 turns to a CT 14 turn coil. Pic 7 shows the ZVS board built in place of the old 555 circuit. Pic 8 shows it all put together and operating at 21 volts while it sets the alligator clip insulation on fire. I ran it for a long time wide open at 21 volts melting and burning away with no trace of overheating or problems. Outstanding performance. This supply was for running external multipliers or Marx banks. Or whatever comes to mind. I used a pair of the 264 fets in this one. My 470 ohm 3 watt resistors were in so I used 2 of them. Two 1 uF 630 volt capacitors in series, at the junction of the two I run through a 47 pF 1 KV SM to the RCA jack. Hot lead grounded though 68K for a slight load so measurements of the frequency and wave-shape will be more stable. Worked out very well and now I have a much more powerful supply.

IrC, you are much further in this area than I am, but I now can show my first results with the true ZVS driver.

The circuit is shown in the picture below. I used the PCB on which I posted a link before, and a flyback transformer on which I made 2x5 turns, just as in the standard documents about the ZVS driver. I used simple loudspeaker cable for the primary, with one side marked with a red stripe.

I do not have a really good power supply for this kind of experiments, I only have 12 V and 13.8 V fixed, max. 10 A. I took the 13.8 V power supply and connected it to the driver.

This is a picture of the driver with the transformer and one of my element samples (neon gas in a thick walled glass ampoule).






When the circuit is switched on, then the glass ampoule starts glowing orange near the electrodes and blue in the bulk of the tube. The ampoule is quite large (total length is well over 10 cm) and even under these very unfavorable conditions I get a decent glow in the tube. This means that the output of the driver must be at least several kV. Unfortunately I cannot measure the voltage. There is, however, a strong hissing noise and a smell of ozone, so I think there is quite some corona discharge around the electrodes.

The transisitors in the driver do not get hot at all, the glass tube becomes quite hot and I dare not let it run for extended periods of time. I do not want to destroy my nice sample of neon gas (I also have samples of the other noble gases in such tubes and they also glow when they are treated in the same way). The picture below shows the circuit in action.



Next step is to draw some arcs with the driver with my 13.8 V power supply.

I put a fuse of 5 A in series with my driver, applied 13.8 V and then used the circuit for making an arc. When I do this, then I get a fat and noisy arc for a fraction of a second and then the fuse blows out (it gives a brief flash of light and then it is gone). I have the impression that making arcs with the driver is equivalent to almost short circuiting the power supply. This is not good at all. I am out of fuses now. I hope to receive my step up converters with current limiting soon. I have the impression that something is wrong, drawing such huge currents at only 13.8 V with a fuse in series (which also takes a certain voltage drop) sounds weird to me. What is your opinion on that?

I also ordered another two PCBs for making ZVS drivers. For the other ZVS driver I will use IRFP264's. The driver I have now has IRFP250's (I do not have any left of these transistors, I only had two of them).

As you probably know, the voltage in an arc is very low. Once the air in the gap fully breaks down, it appears as a very low resistance to whatever is driving it. The winding resistance in the secondary is the most significant part of the load getting reflected back to the primary (the driver circuit) in this case. Since the turns ratio is very large in the flyback transformer, the impedance transformation is equal to the turns ratio squared, and the coupling in the flyback transformer is so good; the load looks like a short circuit to the driver.

Do you know how this is handled in commercial transformers? Neon sign transformers limit output current by decreasing the coupling between the primary and secondary of the transformer. By diverting part of the flux away from the secondary winding, the coupling is decreased, and this electrically adds a separate, uncoupled inductance in series with the windings. This uncoupled inductance provides 'ballast', and limits the short-circuit current.

You can do the same thing with your flyback transformer by adding an external inductance to either the primary, secondary, or both. For a Mazilli-type driver, two extra inductors of equal value on either end of the primary will work. This does, however, mean that you have to adjust two inductors every time you want to change the ballast in the circuit. Only one inductor is needed on the secondary side, but if it is in series with the secondary winding, it will need to have a very high inductance, and high voltage isolation, because of the large step-up ratio in the flyback.

A better compromise is to add a second transformer with a 1:1 turns ratio, with ballast inductance between both the driver and flyback transformers. This allows simple inductor design, maybe even using an air-core, if your operating frequency is high enough.



I think it is best to ensure that the reactance of the ballast that is reflected back to the primary is at least 10 times that of the primary reactance. This should minimize the pulling effect that varying loads would have on operating frequency.

[Edited on 8-11-2014 by WGTR]

In principle, yes.

When designing a transformer, usually you want the permeability of the core to be as high as practical. This minimizes losses associated with magnetizing current and hysteresis. In this case, however, the primary of the transformer is part of a resonant circuit. The inductance and capacitance needed in the tank circuit will be determined by the desired operating frequency and the load resistance.

There is more than one way to do this, but I'll attempt to explain this in a way that uses your iron powder torroidal core.

As a rough approximation, this is the design of the passive elements in the circuit:


If L₂>>L₁, then L₁ and C determine the parallel resonant frequency of the circuit. This is because the larger L₂ is compared to L₁, the less effect L₂ has on the circuit. I could show this mathematically if needed. In effect, if the load is 0 ohms, then L₁ and L₂ are in parallel with each other.

Since the input is driven in push-pull, we have a center-tapped L₁. L₂ can be denoted as a balanced configuration of two equal inductors.


Expanding this out to an isolated configuration, L₁ becomes the primary of a transformer, and L₂ appears on the secondary side of the circuit.


This is where we left off before. So the questions are if a iron powder torroidal core can be used as T₁, and how many turns are needed.

First of all, it is necessary to ensure that the core does not saturate at the frequency and voltage that it is being driven at. Assuming that we are driving the transformer with a square wave:

t = 1/(2f),
where t = time in seconds,
f = frequency in hertz

For a 100kHz square wave, this gives us t = 0.000,005 seconds.

Going further:

Wb = (V * t)/n,
where Wb = Webers,
V = average square wave driving voltage across the winding during interval 't'
t = time in seconds of applied voltage before flux reversal,
and n = the number of wire turns through the core.

At 13.8V, at 100kHz with 10 turns, this gives us Wb = 0.000,0069.

To calculate the flux density from this:

T = Wb/A_min,
where T = peak-to-peak flux density in Teslas
Wb = Webers (flux),
and A_min is the cross-sectional area of the core in square meters.

You can determine A_min by measuring the core with a pair of calipers, but oftentimes it is available directly on the manufacturer's data sheet for the core.

Assuming A_min is 100mm^2:
A_min = 0.000,1m^2.

Therefore:
T = 0.000,0069/0.000,1 = 0.069 Teslas

This is the peak-to-peak flux density. Since the core is being driven alternately in both directions, the flux swing is centered about 0. Therefore the peak flux density is 1/2 of the peak-to-peak number, i.e., 0.0345T in this case.

The peak flux density for a core that is being driven with a square wave AC signal would then be:

T_PEAK =(V * t) / (n * A_min * 2)

If one wishes to input the frequency directly into the equation, you would get this:

T_PEAK =V / (n * A_min * 4 * f)

It is this peak rating that is considered in manufacturer's datasheets.

Most iron powder cores will handle 0.5-1T before saturating, so we are well within the operating limits under these conditions. It's generally better to keep the flux density low at high frequencies, due to hysteresis losses. These losses are especially high in iron powder cores.

The above example is for a primary of 20 turns, center-tapped, with 13.8V being applied to one-half the winding at a time.

The next issue is the inductance of this winding. It depends on the A_L rating of the core, and this varies from one model to the next. It is almost always specified from the manufacturer, if you know the origins of the core. If you don't, then you can put ten turns of wire through the core and measure the inductance with a meter.

L = A_L * n^2,
where L = inductance in Henries,
n = number of wire turns
and A_L = inductance of one turn on a given core.

Note that doubling the number of turns causes the inductance to quadruple. Determining A_L allows you to solve for the number of turns for a desired inductance.

In the real world, you might end up with a core that requires 0.25 turns to give you the inductance that you want. This would be because the permeability is too high. If it were even possible to make 0.25 turns, this would be undesirable, as it would cause the core to saturate easily. You can buy cores according to their permeability, but if you pulled one out of an old power supply you might not be able to be so choosy.

In this case there is yet another option. L₁ and T1 can be physically separate parts. As long as the inductance of T1 is high relative to both L₁ and L₂, then it transfers power without appreciably affecting the tuning of C and L₁. In other words, L₁ << L₂ << T1; L₁ has the lowest inductance, and T1 has the highest.


So in this case your torroidal transformer would have a 1:1 turns ratio, and have as many turns as you can fit onto it efficiently. As the number of turns goes up, wire diameter decreases and length increases, increasing DC resistance in the winding. More turns decrease losses from hysteresis and prevents core saturation. An iron powder torroidal core will run quite hot at high frequencies, with high flux density. Increasing the number of turns decreases the losses associated with this.

There's some more that I'd like to add, but I'm running out of steam for the day. Iron powder should work, if the core permeability is high. It would be better, however, to use ferrite instead of iron powder for T1. The material has much higher permeability, with lower losses.

What types of cores/materials do you have laying around?

[Edited on 8-14-2014 by WGTR]

Yes, this circuit is a constant voltage output. So, unloaded, the supply current draw is generally low, and as you bring loads in (rusty nails, hissing corona, etc...), current draw increases.

If the load increases much more than your circuit is comfortable with, excess supply current will be drawn. If your power source is not current controlled, you'll have a problem, either from the power supply browning out (e.g., most flyback type power adapters, when heavily loaded, go into a 'pulse skip' mode which delivers very little average power into a very low to shorted load resistance), or your circuit burning up (I wouldn't recommend running one from battery power, not without a current limiting resistor anyway).

An example of controlling one of these circuits might be,



This is a 10W, 300-2000V DC power supply I made for testing purposes. (I haven't finished the tester it's intended to power, so it's actually useless right now...) The oscillator is recognizable at the bottom, and instead of a direct AC output, or an induction load, a high voltage secondary feeds a full wave voltage doubler.

It's actually more like quasi-resonant, because I intended the transformer to have more leakage inductance (to allow the primary and secondary to resonate with some freedom), but instead, it sort of goes into a 'high pitched whine' mode when shorted (namely, it runs at ~200kHz instead of the normal ~60kHz). Whatever.

In any case, the important aspect is controllability. The output voltage is sensed with a high value resistor divider, buffered with an op-amp, then compared to an adjustable reference voltage (0.19 to 2.08V) with a compensated error amplifier. The amplifier, in turn, drives a power amplifier stage, which has a foldback current limit, 3A maximum, down to about 1A in a short circuit. Finally, a series inductor supplies the oscillator, to provide "squishiness" necessary for commutation* and resonance.

*Commutation, as in, the instant in time when the transistors swap roles. In this case, the moment when one transistor turns on, followed by the other turning off. This "push pull" circuit requires "shoot through" (both on for a short period), whereas most "bridge" circuits require "dead time" (both off for a short period).

Pic:

http://seventransistorlabs.com/Images/HVPower2.jpg

Deadbug construction of course! The oscillator FETs aren't heatsinked, because they don't usually dissipate much power (the worst occurs under short circuit conditions, when load current is high, supply voltage is low, and therefore, the transistors do not saturate fully but remain in the linear range). The series pass transistor gets a nice big heatsink, because it can dissipate 10W easily (the overall efficiency of this circuit is fairly poor, often under 50% IIRC).

An astute observer might note that a switching power supply would be more efficient than the linear pass transistor: that is true. I didn't do this, to keep the circuit simple, and because efficiency doesn't matter for this purpose. I don't need to debug a buck converter as well as an oscillator, plus their possible interactions (the oscillator expects clean DC).



(Funny, last time I posted in this thread, I gave general advice; a few months later and I've now built an example of my own, independently and for unrelated purposes.)

Tim

Ed: eh, that pic's kind of wide

[Edited on 8-13-2014 by 12AX7]

You're very welcome woelen. I'm trying not be overly confusing about all this, but also trying not to be overly simple either. Such a simple thing as an inductor is actually a complicated beast, and has driven many a university student to frustration. I have an entire (well-worn) binder full of notes just on soft magnetics.

If you have trouble getting the driver going, let me know. It may be easier starting out just to add an equal (small) amount of inductance in series with both halves of your primary coil (the one that is currently wrapped around the flyback). That would at least show you whether the circuit will still oscillate and limit output loading at the same time. You want to keep the inductances low relative to the flyback primary inductance to start out with, and then increase it for more load limiting. The circuit will work if the driver is just a simple push-pull driver, driven by an external oscillator, but I don't know how a Mazilli driver will behave with the extra inductances in there.

Here's a useful nugget on wire design. Properly spec'd litz wire has an AC resistance that is the same as its DC resistance at a given set of frequencies:
http://www.newenglandwire.com/~/media/Files/Litz_Theory.pdf



I'm going back to school this Fall, so in a couple of weeks I may be a little scarce around here. One can get old, or one can get old and educated. But getting old is pretty much a given either way.


[Edited on 8-15-2014 by WGTR]

The circuit I made is destroyed I replaced the fuse and it immediately is blown out again, as soon as I switch on the power supply. Just doing the arcing apparently blew out one of the FETs. I do not understand why, I see videos on Youtube with impressive arcs from 24 V or even 36 V power supplies. I just have 13.8 V. So, the experiment with the inductors in series with the transformer is postponed, I first need to replace the transistor(s). I also need to understand why my circuit is destroyed, while others have impressive results with their circuit.

[Edited on 21-8-14 by woelen]

Quote: Originally posted by woelen

The circuit I made is destroyed I replaced the fuse and it immediately is blown out again, as soon as I switch on the power supply. Just doing the arcing apparently blew out one of the FETs. I do not understand why, I see videos on Youtube with impressive arcs from 24 V or even 36 V power supplies. I just have 13.8 V. So, the experiment with the inductors in series with the transformer is postponed, I first need to replace the transistor(s). I also need to understand why my circuit is destroyed, while others have impressive results with their circuit. [Edited on 21-8-14 by woelen]

latchup

Has anybody found a way to improve the circuit so it doesn't latchup and blow mosfets?

I've built this circuit years ago and it worked fine, recently I brought it out and found that if the power supply doesn't rise fast enough, the circuit latches up and one mosfet conducts heavily, while the other one remains shut off. I'm using a linear benchtop supply so current it limited to 3A and no mosfet were blown.

I'm not sure I understand this circuit correctly. When powered up, the voltage on both mosfets will rise to about 3V, and one of them (A) will start to conduct more first. At this point the current from the power supply will flow both sides, because the capacitor is being charged on the B side. The positive feedback will turn mosfet A fully on, thus shutting mosfet B fully off.

Suppose the B side now charges fully, and current stops flowing into the B side. If the center tap of the inductor is at voltage V, then since mosfet A is at 0V, mosfet B would be at 2V. The current through A side will continue to increase following V = L*dI/dt, until either the supply limits or the inductor saturates.

If the supply is limited, the inductor current will have to come from the B side, thus starting the cycle. However if it is due to saturated inductor, it just means the L value becomes small, thus the supply will have to provide larger dI/dt. Therefore the need for a RFC choke on the supply.

Still can't see why latchup occurs....

latchup

Quote: Originally posted by seilgu

Has anybody found a way to improve the circuit so it doesn't latchup and blow mosfets? I've built this circuit years ago and it worked fine, recently I brought it out and found that if the power supply doesn't rise fast enough, the circuit latches up and one mosfet conducts heavily, while the other one remains shut off. I'm using a linear benchtop supply so current it limited to 3A and no mosfet were blown. I'm not sure I understand this circuit correctly. When powered up, the voltage on both mosfets will rise to about 3V, and one of them (A) will start to conduct more first. At this point the current from the power supply will flow both sides, because the capacitor is being charged on the B side. The positive feedback will turn mosfet A fully on, thus shutting mosfet B fully off. Suppose the B side now charges fully, and current stops flowing into the B side. If the center tap of the inductor is at voltage V, then since mosfet A is at 0V, mosfet B would be at 2V. The current through A side will continue to increase following V = L*dI/dt, until either the supply limits or the inductor saturates. If the supply is limited, the inductor current will have to come from the B side, thus starting the cycle. However if it is due to saturated inductor, it just means the L value becomes small, thus the supply will have to provide larger dI/dt. Therefore the need for a RFC choke on the supply. Still can't see why latchup occurs....

latchup

latchup

Quote: Originally posted by WGTR

On slow power supply ramp up, if both MOSFETs conduct at the same time, the tank circuit is effectively shorted out. This means it has practically no inductance, if the windings are well-coupled. (Temporary) current limiting now depends on the series choke, which will probably saturate more quickly than you can ramp the supply voltage. At that point you have a big, saturated, steaming pile of silicon and ferrite (unless you have a current-limited power supply, under which conditions it merely simmers with mild annoyance). To make this work, the series choke needs to be able to handle without saturating whatever current the power supply can deliver. Also, the gate drive to both MOSFETs needs to be kept at zero when the device is not oscillating. This latter point is the opposite of how a typical Mazilli driver works, as it supplies gate drive to both devices on power up, and relies on one device out-competing the other to get oscillation going. If oscillation doesn't start, well, then both devices smoke. Think about this: given that both MOSFETs should be "OFF" when not oscillating, but with the functioning circuit remaining within the design constraints of a Mazilli-type design, could you modify the design to work with, say, a temporary pushbutton that could get the oscillator started? And a circuit that, if it stopped oscillating under load, would allow the gate drive to drift to zero?

Quote: Originally posted by seilgu

I don't think there is the remote possibility of both MOSFETs smoking. When one turns on it immediately turns the other off, they can't be on at the same time. They can't be both off, either. When not oscillating, one is on and the other is off, and it doesn't switch.

Quote: Originally posted by seilgu

The way to get is restarted is for the feedbackloop to have a gain larger than 1 at the desired frequency so that any noise of the frequency will get amplified. The other way to prevent latchup is to detect latchup either by checking how long the MOSFETs turn on or how hot the thing gets. Although if used with large power supplies, you'd be already too late when it feels hot.

So I've been working on a zvs driver myself. Finally decided to put together parts that I bought a few years back not sure if they are all in the same category of power consumption for the proper circuit. Parts are as follows: mosfet- IRFP250N, fast diode- UF4007, zen diode- 1N-5349B, 1uF MKP10 cap 1000-600~, 5w 470 ohm, 1/4w 10k ohm,

Ended up copying an online schematic and making my own PCB board. Got everything etched and soldered in place except for the inductor coil. For a non electronics guy some of this stuff is kind of hard to pick up, like figuring out on the parameters to wind your own. I can usually gather enough info just reading online, but lately life gives me my spare time in numerous shorter lengths. While working away I only get to use my cell phone for internet Leads to learning simple things many times over, and little time for theory.

Any suggestions for whether or not the yellow/white T106-26 core would be good? Pretty sure it was from a ATX PSU.
I do have T 106 26 white yellow iron powder core, a few ferrite cores unmarked and bare. I also have three spools of magnet wire from RadioShack the thickest is 22 gauge but I do also have 26 and 30 gauge

Also, etching.g was done pretty easily. One SOS pad, muriatic acid, little bit of rust and some hydrogen peroxide 3%. The solution was ready in about half an hour, and etched rather rapidly. 10 minutes was almost too long as the traces we're getting undercut.

Any input on the inductor core would be great, so far the online calculators I am seeing say between 38 and 42 raps on the core with the 22 gauge wire that I have( for about 150 uH). But the schematic that I had only states between 47 and 200 Micro Henrys of inductance, is a pretty wide open guess for someone who doesn't have the know-how or eyes for mismatch problem.

I have a fly back to try it on, but will eventually be building an induction heater. I don't reall need big sparks for fun, But melting metal sounds productive

Thanks for the help there IrC, much appreciated. I cut wire to the specs of the coil calculator and came up a couple turns shy of the 150uH rating. ~135 as it was two turns under 150 and two turns over 120. By online calculator, my multimeter isn't that multiple in function soon enough.

It's all wound, soldered, the 5+5 on fly back done... Be home tomorrow to throw it on a power supply

Then on to the next planning stage, the induction heater. Side note, solder or flux seems to be eating my irons rather rapidly. Butane one I've had for years and a new harbor freight one see damage immediately. Not a fan of that.

I have 10 of the 470KΩ resistors, 20x 4007 fast diodes, 10 IRFP205N, 5x IN5349b zener-d, 2 of those caps ordered almost 2 years ago, and a pile of various bits ripped off boards that weren't new. have a pile of the big microwave caps, at least 6. a number of other much smaller MKP caps/a fist full of several KV range ceramic caps.

then on the way are some
MOSFET IRFP460N x10, IN4742 12V 1W zener x10, IN4767 18V 1W zener x20, IN5822 3A 40V schotkeys x20, for the heck of it considered getting 2x tiny 12v fans, like the size of a US penny,... but 16$ free shipping will have to wait would look cool on the green heatsinks
http://www.ebay.com/itm/131500008965?_trksid=p2060353.m1438....

I'm hoping to get more time to read on something bigger than my phone... sux only having one page open, reloading any tab every time you swap to another one open, on such a small screen. it makes learning stuff like this FRUSTRATING!!! also my father is a ham operator, and can give some advice, if the concept is similar to his past projects ill get answers. though most my projects bear little resemblance to his.

I've come to the conclusion I just need to start an electronics notebook to go with my chem one. much nicer to find last times notes in one spot not peppered with random junk unrelated. I have a number of decent books at my disposal, have to dig them out. many are either WAY to simplified, or over my head. high school electronics was long ago, time to get back to it eh.
I really appreciate the help here guys, been trying to do my homework and testing before posting more questions. it's the practical knowledge I want. guess it's time to read up and wait for the mail. thanks

Edited in some pics


1)Some of the parts recently ripped off boards.
2)weird diode
3) 3 transformers from the same board, had a pile of 2.2kv ceramic caps too
4) sintered glass diodes, byv76, byw 34-tfk, some with 228-ph end. Kinda cool, haven't seen them often.

[Edited on 24-2-2016 by violet sin]

Quote: Originally posted by woelen

This thing worked very well for electrolysis experiments and was very robust in terms of overloading it with low resistance or even shorting it, but when I did my first high voltage experiments with a small 15 kV DC power supply, salvaged from an old monitor, it quickly died. I did only draw a few hundreds of milliamps from the 12 V terminal. I am afraid that some low power, but high voltage statics destroyed some sensitive CMOS circuit in the power supply. Could this also happen when I want to use the 24 volt unit with switched regulator for my ZVS driver?

not much love from my project just got in some lower watt zeners 12/18v, the right 2W resistors, IFRP460's and some other non related bits.

I tried with the correct value resistors, but perhaps some of the components were more damaged than my multimeter led me to believe? no arc's issued forth, no hum of flowing electricity, but no smoke/heat either. all ambient temp. just does NOTHING. ATX PSU isnt dropping dead when connected, measured correct voltage on rail so no Vdroop, quick amperage test showed pleny( and quickly warmed test leads). next up is winding a better inductor with thick wire, looking for better caps, and redesigning my PCB ( the main reason it's unattended ATM ). not that this one is bad, just want a batter parts packing geometry for an easily packaged unit. a guy doesn't need too many 'random bits on board' type projects dumped in a box for use *sometime* in the future.

there is a LOT to learn about the various semiconductor parts available... soo much reading. projects on the table now are LED desk lamp(95% done), some sort of power supply to take abuse ( instead of the 70$ ebay unit I love), understanding and attempting to build my own custom LED driver for nightlights.

basically my time now is spent on electronics projects and research, instead of chem. got a note book for documenting the process, starting from small simple units around the house and from ebay. 200W scr dimmer, CFL bulbs, 3w led driver, motion detector light sensor, kids toys etc. sketch the circuit, read, understand and fix if possible. if not you learn stuff. then I will get back to the harder stuff like the stepper driver I failed miserably on.

just easier for me to mess with electronics right now then chem it up at home, and I always went back and forth any way every couple months or so.

funny little side note, I tried to get a spec sheet for an 8-pin BP9011 chip(on the 3x 1W CC LED driver) from an alexpres seller. the part is not listed in any datasheet search engine, assuming it is a generic overseas chip name instead of the proper manufacturer designation. anyway, he quoted me 10$ ea for the chips, couldn't provide a spec sheet and said only "if the part number is the same it will work".. meanwhile the whole led driver costs 1-2$ US on ebay with free shipping, LOL . they do have great prices on triac's of considerable power though