The following pic shows the board with the heatsink removed. The components that are heatsinked are a GBJ3510 35A/1kV bridge rectifier (wider rectangle, all black) and two parallel FGA25N120 25A/1200V IGBTs. The circuit analysis is as follows:
Starting from the upper right, you can see the AC input screw lugs, one right above the vertically-mounted xfrmr and one near the hole in the PCB a little ways in. That inboard lug passes through the black shrink-wrapped horizontal fuse and then both AC lines pass into the toroidal chokes at the upper right. One choke per line, of course – 300uH each.
After the chokes (which prevent switching noise from passing back into your home wiring), the AC lines are filtered by the big yellow X-cap at the top of the PCB. The Neutral line passes next into the 3000:1 current sense xfrmr (horizontal xfrmr just below the Xcap) and then both lines go to the heatsinked bridge rectifier to be converted from 60Hz AC to full-wave rectified 120Hz “lumps”.
The full-wave rectified current is filtered by the 8uF cap and the negative side connects to the IGBT emitters as well as acting as the GND rail for the entire power stage. The positive side of the rectified current passes through the horizontally mounted toroidal choke (400uH) and then to the top lug of the work coil connection and one side of the two 0.33uF resonant caps. The other work coil screw lug is connected to the low side of the two resonant caps as well as the two IGBT collectors. The work coil, of course, attaches to the two screw lugs and is connected in parallel to the 2x 0.33uF caps. So one side of the LC tank is held high, and the other side is pulled down by the two parallel IGBT’s.
Check it out folks – GROUND referenced resonant switching!! We were really, 100%, expecting a half bridge when we saw those two IGBT’s and it was very exciting to see a nice, 1.8kW commercial IH using such a simple and hobby-friendly power stage. Next, we used a 3-turn external coil on a 50mOhm resistor to pick up the current waveforms and read them on a scope – let’s take a look.
Fig 12 – Clear View of the Power Stage PCB
Here’s the 50,000 foot view. The current rises and falls in sync with the rectified AC line, meaning that when the AC input voltage is high the current is also high. This is really great because the AC power is used for 100% of it’s cycle and the input load of the device just looks like a resistor – it’s called good power factor and it gives two advantages.
- You can get more power from a single plug. Like we said, you’re actually using all the power delivered by the AC line. This is different than when a bridge rectifier is filtered by a huge cap to try to make DC. That circuit only takes power from the AC line at it’s peak, and it gets no useful power for most of the AC wave so the total power available is much less.
- The power company likes it better. Only taking current at the peak draws a big surge and ugly’s-up the power company’s nice waveform. It also means that they need to use really big wires to provide the peak surge current but the rest of the cycle is basically dead, making those wires a big waste. For a big power user like a factory, you’ll pay huge penalties if you do something dumb like this so good power factor is a must.
For little household users like us, it’s mostly good manners to strive for good power factor. Our surge is a drop in the bucket, but when added up to all the homes in the neighborhood – it’s gotta count! The more important reason is that for a fixed 20A circuit, a good power factor allows us to get more power out.
The interesting thing is that this topology seems to (if implemented correctly) give PFC for free!
Here’s a scope shot showing the envelope of the current drive – as you can see, the amplitude is directly proportional to the rectified AC line. These scope shots were obtained by winding up a cliplead into a 4-turn coil and clipping across a 50mOhm resistor. So the cliplead acts as a tiny transformer winding to pick up the electromagnetic field from the work coil. The cliplead coil measured 850nH, for those that are interested.
Fig 13 – Long Timescale View of the Work Coil’s Field
In this next shot, we can see a closer view of the field signal from the work coil. Now since the sense resistor was 0.050 Ohm (1%), this measurement of 500mV of signal means that our little 850nH sense coil is circulating 10A! Imagine how much current is circulating in the pan, which should only measure a few nH at best – massive!
Fig 14 – Closer view of the Work Coil’s Field
And the final scope shot shows a close-up view of the work coil’s field. It’s not exactly sinusoidal, although it’s pretty close. What we are picking up here is a replica of the current through the work coil. The little glitch at the bottom right of each cycle is probably due to the diode turnoff/switch turnon, which makes sense when you consider that the current begins to rise right after the glitch – the switch must be on!
Fig 15 – Zoomed in View of the Work Coil’s Field.
Stay tuned for more analysis of the circuit, and some theories on how to build a controller for this type of device. This article is already WAY too long to even begin to delve into the circuit diagram. But trust that we will be back very soon with our ratty hand-drawn schematics converted into pretty PDF’s for your viewing. Until then…