Help with PLL induction heater driver

[dbach]

Hello all,

I have been working on my induction heater and have been recently working on the driver. I have modified The Backyard Scientists design (with permission) and have added a full bridge and ucc type driver for ease of use (nearly a copy of the UD type tesla coil driver power sections as it is a tried and true circuit) If anyone could look over the (awful) schematic and let me know if I made and stupid mistakes before I have them fabricated that would be much appreciated! 

thanks so much, Davis.

[petespaco]

What are you going to use the induction heater for?
What frequency range and  power output will it have?

Pete Stanaitis
---------------

[dbach]

Induction heater is for general hobby use, melting an forging and whatnot.

Frequency range will be around 50-100khz, output power goal is around 15-20kw (will send photos of the inverter and other things shortly).

re did the design with SMD components and better component choice in general, attached are photos of the circuit and (poorly) routed PCB. This was my first time routing traces so I hope I didn't botch it too badly.

Thanks so much, Davis.

[Anders Mikkelsen]

Your modifications should be fine, but the original design has some issues that can cause erratic operation. Kevin's video shows that it works well in his case, but with 300 A IGBTs and a low operating frequency, you can get away with a lot of sins.

Having the ICs drawn as the physical chip rather than schematic symbol makes the schematic very hard to read. Especially the circuitry around the double op-amps are hard to read without redrawing the circuit.

The 4046 PCII has trouble controlling resonant tank circuits like tesla coils and induction heaters, due to the fact that it counts edges. Any extra edge on either input can push it into a state where the frequency is pushed to the end of the range and it doesn't recover. There's some more detail here https://4hv.org/e107_plugins/forum/forum_viewtopic.php?156966 and here https://4hv.org/e107_plugins/forum/forum_viewtopic.php?p=1&id=173532

180k + 1N4007 diodes is not ideal for the capacitor voltage phase sensing. The capacitance of the diodes along with the high source impedance adds around 60 degrees of phase shift at 50 kHz, and 80 degrees or more at 100 kHz. This is not accounting for the slow recovery time of the diodes which might make it even worse. The phase shift can always be calibrated out, but it eats into the range of the phase detector and adds frequency-dependent phase shift, requiring retuning whenever the work coil is changed.

On the PCB, the ground pin of the 4046 should be connected to the ground plane as close as possible, right now the path is pretty long and thin. The same goes for the ground connection of the decoupling capacitor next to the chip. Some ceramic decoupling here would be nice, but 40xx series CMOS is pretty forgiving given the high operating voltage and low switching speed. Some decoupling near each chip is generally recommended, with as short of a path to the power pins as practical.

I don't see any means of regulating the coil current or voltage in this circuit, it appears to apply constant voltage to the series resonant tank, at the resonant frequency. This means the heater will draw maximum power (and put maximum stress on the components) when unloaded, and it will draw progressively less power as you load it more. I would recommend having a method of regulating the tank current or voltage, to make it draw more power as it is loaded. This would involve sensing the voltage and current, and using this signal to regulate the effective drive voltage, for example by detuning the operating frequency.

[dbach]

Hello,

Thank you very much for the advice. I will add means of tank power regulation in the Rev 2 board, luckily I am using some even more powerful bricks (400a 1200v) with lots of protection  .

I have had the boards manufactured by JLC PCB (somehow cheaper then buying components separately!) after lots of layout changes and fixes, although I will definitely work on the capacitor sense system for next time.

For the next revision I might be replacing the 4046 chip with either a microprocessor (makes regulation easy with pulse skipping) or any other PLL chip. I was using the 4046 chip simply since there is a good deal of information out there on it.

As for layout issues, hopefully I didn't make them worse (images below), and again for next time I will add some decoupling chips and fix some routing errors.

Thank you so much for your help, will be posting pictures of the finished tank circuit soon!