|1830 Machine Learning Meetup #1 Jiri Materna|
|1900 BRMCW - vyuka telegrafie 2000 TalkNight|
This is an old revision of the document!
What began as the (now archived) project Tesla Coil has expanded into a whole high-voltage-oriented room, located downstairs of the main Brmlab premises. If you're interested in a tour, or would like to join in, come stop by either during regular tuesday evening meetups or contact chido or mrkva.
The small (~3x3m) room we occupy is situated below Brmlab, and was originally intended as a storage room until we came along. Now it serves as a place for all those experiments and projects that are too dangerous for people and/or equipment to be performed upstairs.
We try to keep it as safe as possible (it's in our own interest after all), but it's still a dangerous place, we therefore advise visitors to heed our instructions and don't touch (or turn on) anything without supervision.
The oldest (and longest functional) coil we have. Formerly the Tesla Coil project, it ended up being a tabletop coil with parts build from things scavenged around the lab.
(mrkva) type, specifications, components, schematics
TODO (chido) history, pictures
TODO (mrkva) type, specifications, components, schematics
Porek is the largest coil that resides at Teslab. It was donated to us by its author, Filip, and remains unfinished, mainly for our inability to construct suitable capacitors for it. For pictures, go here.
(mrkva) type, specifications, components, schematics
Petrzlen belongs to mrkva, who made the secondary coil years ago and now moved his project to Brmlab. chido built the wooden socket on wheels for it. It's still under heavy developement. Pictures here!
This is the second largest Tesla coil currently located in Brmlab. The topology is well proven DRSSTC (double-resonant solid state tesla coil) design.
Secondary is 92cm tall, wound with 0.355mm wire (~2400 turns). Toroidal topload made from flexible air duct tubing currently borrowed from the “Porek” coil. Resonant frequency of secondary is approx. 59kHz.
Primary side consists of 8×8 array of 1kV/0.5uF KPI-300-047 pulse capacitors made by ES Ostrava a.s. and 8-turn primary coil made of 10x1mm soft copper tube, with inner diameter 25cm, outer diameter 55cm. The coil is conical, with 45° slope. As switches on the primary side, we use a pair of 2MBI300-140 IGBT half bridge modules, rated 1.4kV, 800Apk. Primary voltage is regulated using old soviet 10A variac and connected on the H-bridge terminals through mains filter and voltage doubler with 400V/8.2mF electrolytic caps.
Driver electronics is custom-made using primary current feedback, fed into 4046 PLL IC, two stages of gatedrivers (first level is pair of IC gatedrivers boosted by the two discrete pairs of N/P MOSFETs. Output drives directly gates of IGBTs using GDT. The board is capable of both interrupted and CW mode of operation, with overcurrent limiting and audio modulation (untested yet).
Regulated output from variac is fed onto L,N,PE input connector, filtered using pi-filter, and rectified in voltage doubler. Ouput of the doubler is connected to two half-bridge modules using aluminium L-profiles, to minimize any resistance between large filtering capacitors and bridge. On terminals of each module are two 1.5kW 400V unidirectional TVS in series. Gate of the transistors are connected directly to the GDT output and are protected using 600W 18V bidirectional TVS. Between primary tank capacitors (C_PRI) and primary coil (L_PRI) are two ferrite toroids, one working as primary current feedbeck, another one as sensor for primary current limiter.
Caution: there is a serius flaw in the schematics regarding the feedback Try to find it :)
The driver circuit might look too complicated, or at least more complicated than other DRSSTC designs. The feedback from primary current sensing coil is clamped to 0/+5V voltage rails and fed directly to input of 4046 PLL IC. We've chosen 4046 C,R1,R2 (C7,R11,R12) values so the resonant frequency of primary circuit will fit within operating range of the 4046.
Other signal from current limiting coil is rectified using Graetz bridge from Schottky diodes, filtered and compared to reference voltage, if it is greater LM311 comparator (IC4) sets its output high, forcing interrupter output low, effectively stopping operation.
D-type flip-flop (IC3A) takes as input value from interrupter and as clock output from 4046 and is effectively synchronizing interrupter singal with feedback from 4046, preventing any hard switching.
Signal is then fed to pair of UCC27322(1) gatedrivers that were supposed to drive the GDT. This was however proven to be almost impossible - even at low interrupted duty cycle UCCs were getting really hot and worse - at maximum recommended input for UCCs (15V) the voltage across gates of IGBTs was only ~10V. Given the fact that those big IGBT bricks have gate saturation voltage approx. 8V, it was really dangerous thing to do - one small voltage drop below saturation level and total power dissipation on that IGBT bricks would destroy them. The final solution was to add a small H-bridge made of complementary MOSFETs that are driving the GDT. It worked really nice, however when we've decided to raise the GGDT driving voltage a little bit, GDT became saturated and had to be rewound. Now it works really nice - see the “Measurements” section!
Yep, hard switching is a bitch, but after tuning phase-shift a little bit, output looks much nicer:
TODO (mrkva): more pri/sec measurements
(the Array, Nikolka, experiments with glass and plastic wrap)
(what do we have?)
Goal of this project is to build a small Marx generator. Marx generators are pulsed voltage multipliers, that can be used to experiment and play with voltage levels that are difficult to reach by other means. High voltage pulsed discharge also usually generates smaller EM pulse - another thing to experiment with. Marx generators can be also used as a power supply for plasma (and fusion) experiments!
We do all sorts of high-voltage related experiments, here's some examples:
Prompted by the most remarkable (at least for a biologist) metabolic pathway of fructose, chido bought some and a series of experiments on cooking using high voltage ensued. These included (among a failed attempt at making popcorn and a more successful one at roasting marshmallows) making caramellized fructose using an electric arc. For a detailed recipe, see this handy how-to.
We've disemboweled a couple of microwaves, not letting the magnetrons go to waste:
And tried to make art:
Parsley viewed from inside a faraday cage, by bluebear