Induction hob: What is the effect of suddenly removing the pan?

[nali]

Electrically speaking of course. Not sure whether this belongs in Repair or here, but I'm more interested in the analysis than the actual repair.

I've attached the reverse-engineered driver schematic of a hob I am looking at for a friend - the issue is that it will happily sit there boliing a pan of water, but if I suddenly lift the pan off the ring while it's going full power the IGBT violently lets out the magic smoke and all the lights go out.

As far as I can see the hob coil and 0.33uF act as a parallel resonant circuit when no load is applied, so high-Z. When a pan is on the ring it now becomes a low-z transformer. But what happens during the condition that triggers the smoke?

[Note there is no flyback diode in the schematic, the IGBT incorporates a power diode)

This appliance has two identical PCBs each driving two rings, the issue has happened on each. Nothing in common apart from a fuse and power relay switching the line voltage. The only slight anomaly is the line voltage is a bit high here at 248~250V though still within UK spec.

[peter-h]

What is the model of the hob?

There seems to be no detection of the no-load condition, which is bizzare.

[T3sl4co1l]

I know those things get pretty goddamn cheap, but I had no idea there were actual open-loop unregulated and unmonitored versions out there.

Well, I guess there are, they just don't sell well enough to show up that often.

I wonder if there's a reason for that.

Failure mode most likely overvoltage triggering avalanche destruction, and subsequent short circuit.  There's no fuse either, it takes out the breaker.

It's a danger to itself and it's a danger to others. Throw it on the heap ASAP.

Tim

[tooki]

I know those things get pretty goddamn cheap, but I had no idea there were actual open-loop unregulated and unmonitored versions out there.

Given that $40 induction hot plates (which actually manage 1800-2000W, exceeding many “real” stove burners) from IKEA or Aldi manage to have perfectly functional pan detection, I’d be surprised if it was designed to be open-loop, but rather that the thing’s circuit design sucks in some unexpected way.

[nali]

I know those things get pretty goddamn cheap, but I had no idea there were actual open-loop unregulated and unmonitored versions out there.

Given that $40 induction hot plates (which actually manage 1800-2000W, exceeding many “real” stove burners) from IKEA or Aldi manage to have perfectly functional pan detection, I’d be surprised if it was designed to be open-loop, but rather that the thing’s circuit design sucks in some unexpected way.

You're right guys - it isn't open-loop! I did think it was a bit crude (double check toroid to see if it was in fact a current txf - nope) but never looked inside one of these before. The PCB is single-sided and the IGBT emitter is connected to power return through what looked like a wire link, but is obviously a shunt (see pic, next to IGBT highlighted in red) as there is a connection to it leading to an analogue input on the MCU via some RC filtering.. updated schematic below.

Nevertheless both identical PCBs have failed so there's something more fundamental than a simple fault, so I think the design is maybe borderline, exacerbated by running at 250V, 100% power and removing the load (pan). More than one corner case here.

Peter-h the make/model is Gionien GIT470SP, it's a 4-ring 2.8kW unit.

The owner has already decided to scrap the unit, I just don't like leaving things unexplained, they bug me 

[mag_therm]

If you want to spend the time on it and have test gear **, , replace the IHW20N135 (1350V 20 A) (and probably at the 18V zener).
Check that the normal IGBT gate ON voltage is at least 10V
Run at reduced voltage ( 50%)  using a variac. DSO Scope probes on V_drain ** and the current sensor, trigger on either rising high.
Do the pan trick and try to distinguish which of  voltage or current is closer to the datasheet max ratings

It could be that the current feedback on both hobs is either failed, or never worked properly.

**100:1 probe rated 1500V needed

[nali]

Well there's 3 other identical channels to play with...

This is the 3rd failure now; the original, my friend already replaced the IGBT before he gave it to me, then I re-replaced the IGBT plus the 18V zener as both had fused, and finally I replaced the Chinese bridge rectifier with a Vishay part as a precaution. The board I repaired is working, this is now on the other identical board.

Sadly variac & HV probe are two items I don't have at home. I've been promising myself a variac for a little while now, I'll have to have a word with myself about the probe

[peter-h]

What frequency does this thing run at?

It can't be in the audio range because you would hear it, due to magnetostriction. It could be 50Hz or 100Hz.

[T3sl4co1l]

I see 820k resistors near the output terminals as well. Are you sure there are not any other connections laying around?

Tim

[T3sl4co1l]

What frequency does this thing run at?

It can't be in the audio range because you would hear it, due to magnetostriction. It could be 50Hz or 100Hz.

It's class E, with 5-20uH coil inductance and 0.3uF capacitor.  That gives say 90kHz, but the quasi-resonant class-E operation drops that (longer on-time in return for higher peak voltage).

Tim

[peter-h]

How do they get it through EMC approvals? It will be radiating 90kHz 180kHz 270kHz and all the way up

[mag_therm]

A typical "proper" induction heating inverter has a safe operating area surrounded by about 6 limit lines.
( Capacitor Volt, Semiconductor volt Semiconductor current, Semiconductor temperature, rectifier power, turn off time)
During startup and load changes and ramp down,, each of these limits might be approached.
So there are 6 independent fast acting sensors/regulators. Any one can pull the firing frequency back into safe zone

With your cooking units they seem to be using just Semiconductor current, Semiconductor temperature.
When the pot is removed, for example, don't know, the drain voltage might fly up into danger zone while the Semiconductor current stays fairly constant.

[Someone]

What does the instruction manual say?

"cookware must not be removed while power is on" ?

[PCB.Wiz]

I've attached the reverse-engineered driver schematic of a hob I am looking at for a friend - the issue is that it will happily sit there boliing a pan of water, but if I suddenly lift the pan off the ring while it's going full power the IGBT violently lets out the magic smoke and all the lights go out.
As far as I can see the hob coil and 0.33uF act as a parallel resonant circuit when no load is applied, so high-Z. When a pan is on the ring it now becomes a low-z transformer. But what happens during the condition that triggers the smoke?

The clamping scheme here is the resonant circuit and the reverse conducting diode, which is somewhat Hail-Mary.
At low drive duty cycles, that limits to sine-ish ~ 2x Vcc, but as the ON time increases it becomes more like a TV flyback 

A half bridge driver always knows what the device swing will be, but a single device flyback part depends on the loading and release current.
The worst case will be high drive current, with sudden load+damping removal, and you hit the jackpot here 

 
I guess the manual says to turn off before removing the pot ! 

[PCB.Wiz]

It's class E, with 5-20uH coil inductance and 0.3uF capacitor.  That gives say 90kHz, but the quasi-resonant class-E operation drops that (longer on-time in return for higher peak voltage).

Put those numbers into spice and this results, as we vary the damping/equivalent loading

ESR  Vpeak 10uS on time
4R = 700V
3R = 836V
2R = 1.03kV
1.5R = 1.15kV
1R = 1.288kV

If we then vary the ON time, you can see a good response time matters, especially at high drive levels.

1R 5us = 961V
1R 15uS = 1.605kV

[nali]

I see 820k resistors near the output terminals as well. Are you sure there are not any other connections laying around?

Tim

Those are directly across the coil terminal blocks, presumably to ensure the 0.33uF capacitor is discharged if a tech disconnects the coil.

I measured the coil BTW, it's 90uH, and the working frequency is 30kHz which pans out [pun intended].

The stuff mid-bottom is a small auxilary +5V SMPS for the logic.

I'll probably shelve it for now until I get more suitable TE (I have isolation txf & dim bulb tester but that's about it for mains work) Annoying, because at my last job I had access to an abundance of HV diff probes, current probes, variacs etc etc..

Even so it looks like trying to improve it is going to be a bit like putting lipstick on a pig as they say

[peter-h]

30kHz... I bet the cats and dogs love you

[David Hess]

I measured my Nuwave at about 30 kHz with a big cast iron pan and the power is modulated to follow the line voltage so it presents a power factor corrected load to the AC line.  It is suppose to be 1500 watts but only draws about 900 watts, making it underpowered for much cooking.

[IanB]

I measured my Nuwave at about 30 kHz with a big cast iron pan and the power is modulated to follow the line voltage so it presents a power factor corrected load to the AC line.  It is supposed to be 1500 watts but only draws about 900 watts, making it underpowered for much cooking.

That's interesting. I tried to measure the power delivery of my gas stove once, and found it to be remarkably low. I put a measured amount of cold water in my 3 kW electric kettle and measured the time to boil (2-3 minutes). I did the same with a stove-top kettle and with the gas burner on full power. It took much longer, I think 10-12 minutes to boil. I concluded that the actual power delivered to the kettle/pot/saucepan by the gas burner was far less than I would have guessed, much less than 1000 watts.

Given that an induction hob directly heats the pan with no heat wasted, delivering up to 900 watts into the pan would seem reasonable.

[soldar]

That's interesting. I tried to measure the power delivery of my gas stove once, and found it to be remarkably low. I put a measured amount of cold water in my 3 kW electric kettle and measured the time to boil (2-3 minutes). I did the same with a stove-top kettle and with the gas burner on full power. It took much longer, I think 10-12 minutes to boil. I concluded that the actual power delivered to the kettle/pot/saucepan by the gas burner was far less than I would have guessed, much less than 1000 watts.

Given that an induction hob directly heats the pan with no heat wasted, delivering up to 900 watts into the pan would seem reasonable.

Technology Connections

[coppice]

I measured my Nuwave at about 30 kHz with a big cast iron pan and the power is modulated to follow the line voltage so it presents a power factor corrected load to the AC line.  It is suppose to be 1500 watts but only draws about 900 watts, making it underpowered for much cooking.

Have you tried with a stainless steel bottomed pan? They are usually what the makers optimise for, although different alloys can change performance there, too.

[IanB]

Technology Connections

Yes, actually, I have watched that.

I just repeated the experiment.

1 kg water (1 L), from room temperature (25°C, measured) to rolling boil (100°C).

Electric kettle: 2m0s = 120 s
Gas ring, stove-top kettle: 6m15s = 375 s

Power estimate for electric kettle: Power = 1 kg x 4.2 kJ/kg/degC x (100 - 25) degC / 120 s = 2.6 kW

Pro-rata power for gas ring: Power = 2.6 kW x 120 s / 375 s = 0.83 kW

(My 3 kW kettle seems not to meet its specification.)

[amyk]

More relevant videos on this topic:

[David Hess]

Given that an induction hob directly heats the pan with no heat wasted, delivering up to 900 watts into the pan would seem reasonable.

My power measurement was made on the 120 volt AC side, which led me to investigate the output waveform with an oscilloscope.  So power applied to the pan is 900 watts minus efficiency losses.

[David Hess]

Have you tried with a stainless steel bottomed pan? They are usually what the makers optimise for, although different alloys can change performance there, too.

I get identical results with aluminum and stainless pans which are designed for induction stovetops.

I did just notice however that the Nuwave is indexing the power to the temperature control, even though it has closed loop temperature feedback.  So if I set the temperature to the highest possible setting, the input power does go up to 1500 watts.  (1) But this means that the recovery time at lower temperature is extended.  The temperature control is very poor, for instance it overshoots the temperature when starting so it is best to wait almost half an hour before cooking, and then temperature recovery is very bad making cooking inconsistent.

The design problem here is that the temperature sensor is measuring the temperature of the insulated cooktop, which is considerably delayed from the temperature of the bottom of the pan.  It would actually be better without the closed loop temperature feedback.

As an aside, I originally bought it for cooking candies and found it to be completely unsuitable.  Because of the way it operates, the minimum change in temperature (10F) is the difference between too little heat and way too much.  And an external temperature probe cannot be used accurately because the induction heats the probe directly.

(1) Anything above 370F tends to burn stuff, limiting power to 900 watts at the plug.