Search the Community

Salutations everybody, My present update still refers to previous posts based on VaporGenie's unique core material selection, .e.g. space-age Silicon Carbide Foam -- Unsurpassed since Y2K IMO! VaporGenie pipe(s) as Prototyping Platform for Semi-DiY Pulse-Heating challenge (2020-May-18) Egzoset's LAVA IH-driven concept adapted to fix failed products (as Loto Labs, Dr. Dabber, etc.) {2019-Mar-30} There was an opportunity to refresh my "intro" recently which now ties it all together into 1 single (but very long) publication which i made available here via my dedicated album (it can be read off-line with a tablet after taking advantage of this board's Image Tools / Download functions), or the reader can also find a copy in my Google Photos and Flickr collections anyway: [ https:// photos.app.goo.gl/bLEHV2asFupTvoxJA ] Egzoset's intro on HV Forum (2020-jul-2) N. B.: Please don't bother registering out there just to exchange public DiY correspondance with me as i happened to cancel the account kind of early. Nonetheless this was a fine occasion to browse through some intense ZVS reading, motivated by catastrophic hardware failure reports from DV VC IH-driven enthousiasts still busy evaluating their popular modules inspired by the work of Vladimiro Mazzilli and George Royer before him, eventually confusing over-simplification vs KISS designer rationale... A 1st hint occured after i noticed the contribution of Bob Nash about his addition of "Gate Drivers" in order to fix that approach, mainly by preventing "lockups" via mutual-exclusion logic circuitry; a creative guy named Dany K. (cz) got intuitive enough to protect his project with a series-connected power lamp, who then applied the virtues of more series-connected ("Snubber RC") components, to improve stability in the power stage... In terms of miniature integration it seems to me the "Fast ZVS Mazzilli Driver" of Francisco Piernas Diaz failed to attract due attention, already more than a couple years ago; somewhat closer to us the work of Oleg Kutkov added subtle refinements, mainly with his mutually-exclusive failsafe implementation where failure modes get strictly restricted by redistributing the circuit as multiple independant elements - the contrary of integration - but doing so i figured damage to 1 component is much less likely to escalate into a "lockup". Personally i prefer to use the "crowbar" expression as we're talking about power devices incidentally putting a heavy Short-Ciruit across High-Voltage power tracks, which should be a forbiden state and hence i was forced to conclude the Mazzilli/Royer modules are actually flawed-by-design! The worst part of it being how the "community" chose to address such odd situation: by circumventing it instead of making durable corrections, which may seem to be a cure but only until that also fails to handle the cause they left intact! As a matter of fact i'm now convinced in this particular case the "Stupid" part in "KISS" can prove most deceiving as that ain't necessarily synonymous of over-simplification; the basic observation that sometimes "ZVS" isn't really ZVS anymore should have rang a few bells, but nobody paid due attention nor even listened i guess... DiY experimentation is great, yet i keep thinking just stop repeating the same errors relentlessly!! After all, those self-regenerating resonator loops were never supposed to end up exceeding the operational parameters of expensive semiconductors, starting with "lock-ups" although the intrinsic sense of "Zero Voltage Switching" should imply a physical impossibility to simultaneously drive 2 power devices of a very same arm, NEVER! Too bad but it's quite about time to step back and acquire additional perspective once it's notorious and repeatedly verified that those popular over-simplified Mazilli/Royer modules allow such catastrophic failure by design. IMHO it must NOT be an option to sacrifice reliability in exchange for trivial trade-offs - but that's me. So, looking for alternatives i eventually realized people got blind to beneficial features of the whole forest just for having their nose stuck to the nearest tree: law of least effort prevails universally i'm sorry to admit! The thing is IH-drivers can hardly gain popularity proportional to merit in a context where both the delivery method and power source result in smoking hardware - a serious deterrent for any vaporist!... So, my suggestion of remedy has been to insist over proper gate driving as a foundation, then comes the problem of component selection since the choice is vast, with many variations on a theme. For example Infineon's iR2153 has on -chip low-side/high-side gate drivers mutually-excluding each other via digital logic interfacing with some generic "555" multivibrator connected to suitable level-shifters, etc. A newer iRS2153 version even offers the convenience of a self-contained "bootstrap" diode! Etc, etc. Strangely enough none of this shows up on modules with an unusually-durable carreer, perhaps not for the best reasons. What's clear to me is that gate-driving chips such as the iRS2153 could as well support a local VCO just through the adjunction of a simple "varactor" diode, hence opening the door to completing this self-resonant loop through a venerable "4046" PLL chip, for example. My proposed scenario implies auto-detection of a delivery tool based on the workcoil + capacitor (LC) natural peak frequency response being shifted: #1) at insertion/removal time; #2) by virtue of the Curie effect if a corresponding alloy is present. Think of a 2-levels Power Stage controlled by exploiting a carrier-lock condition... Initially the "boostrap" phase inhibits all driving, then brief bursting salves of low intensity are generated to scan the workload cavity at relatively long intervals, which should always reflect a non-locked state while empty or after it's ready. Once feedback EMF is finally picked up the VCO will eventually "capture" the loop's natural resonance frequency, aided by those short/weak bursts of the preliminary "stand-by" phase, until it's time to switch the high Voltage source into its final IH injection mode. A timer or temperature function must end that Pre-Heat cycle, necessarily, so i personally like to think of synchronuous quadrature demodulation since there's a local clock reference readily available at lock time anyway. If Curie alloys are included then electromagnetic vs susceptor interactions will be reflected as "pick-up" signals, because these give access to a temperature-dependant Curie effect only possible with IH-drivers, IMO similar to this: [ http://www.elegron.pl/files/AFCinductionfurnace.pdf ] (p64) In conclusion it's all clearly suggesting a button-less/display-less user interface and there's more... Add a permanent continuous magnetic "bias" field (as from a magnet) and the delivery tool shall most probably behave the same as a bi-directional transducer (...), capable of implementing haptic/audible feedback on top of tap-touch commands! Briefly put actuate the power switch to enjoy ZERO lock-ups ZVS with performant energy conversion, insert wand to quit stand-by/scan mode so to begin power IH-driving until the Curie effect kicks in, at which point the unit may want to repeat its "power muting" risk-elimination procedure (if allowed by design). Using my LAVACapsule that's going to be safe for finger skin and even accessible to single-handed/visually-impared persons i believe. All this being a bonus resulting from a necessity to correct some fundamental/obsolete design flaw actually calling for constructive scrutiny + criticism anyway! ... Good day, have fun!!