Microwave oven PWM frequency

[Zero999]

By coincidence, we replaced our Panasonic Inverter Microwave today. From the IC datecodes the old one appears to have been manufactured in 2005. I remember it was still the only brand offering inverter models. The only hard failure over its life was a failed turntable motor (a cheap and easy replacement due to the snip-out panel in the bottom).

Oddly enough a friend of my mum's had a similar problem with her microwave, so I ordered a second hand motor off ebay or Amazon (I can't remember which) and it's still working six years later.

Anyway, it had recently begun to suffer dropping power as cooking time increased, with the odd cracking sound. I had it apart a couple of times but couldn't see any problem - other than the filament glow of the magnetron appearing to dim slowly over several minutes. I wasn't sure if this was due to the inverter or the magnetron (inverter seemed more likely) so effectively BER.

With the nice new one in place, it was time to 'scavenge' the old one. This time I removed the magnetron and found cracked magnets and loose cooling fins that I hadn't noticed previously - a sure sign of overheating. I'm kicking myself now, because I could have got a brand new Panasonic magnetron for £44 on ebay and probably got another 10 years of life out of it! Still, the new one is very shiny, no dents and we don't need the same volume these days (it's still 950W though).

I found the attached Panasonic training guide on the web today. Apparently the inverter controller senses the Anode current using a current transformer to control the power output. With the damaged magnetron, it was presumably dialling back the power as the (non-microwave-generating) anode dissipation increased, hence the slowly dimming filament. Ah, hindsight!

From the training guide, it appears that the PWM frequency varies between approx 20kHz and 45kHz depending on output power, the whole thing running on 100Hz un-smoothed bridge rectified mains. Surprisingly there was only a single Electrolytic in the whole thing, a small Panasonic one on the controller logic supply, rated at 105'C despite having good ambient airflow from the cooling fan (take note treez).   I was expecting it to be chock full of the things!

Now, what to do with a 1kW rated IGBT driven ferrite cored transformer (once the nasty 2kV secondary is unwound)? 

Do you know how much you could get for it second hand, if you repaired it? I wonder if it's worth it, for the satisfaction, even if you break even.

[coppice]

Real men just wind their own magnetrons.

Actually, what part normally fails in them?  Is it usually the windings, or more involved?  I know they use some rather special (and dangerous) materials so probably don't want to handle them more than you need, but I suppose in theory one could rewind one.  There's a pretty serious science to it though, to get the right resonance etc. So probably would not recommend. 

Winding? I think you've got the wrong device. Its vacuum equipment you need to make your own magnetron.

[Gyro]

Do you know how much you could get for it second hand, if you repaired it? I wonder if it's worth it, for the satisfaction, even if you break even.

It is tempting, the going rate on ebay looks like about £45 for a pre-owned one. Balancing that is the worry about a potential buyer claiming that it's faulty when he puts it in his mis-diagnosed microwave and demanding a refund.

On the other hand, it is an interesting looking board with a simple interface, Mains in and a 3 wire opto isolated interface - Gnd, input control, and output power sense signals. The primary side is obviously not isolated, but the secondary side is grounded. I have no appetite for 4kV at 300mA, so that would have to go, but it has a nice high current 3V filament winding. It might possibly be a candidate for an induction heater.



Maybe time for a little research before I decide. 


EDIT: Ah, sorry, you meant the Microwave. I'm not sure if I'm a bit too twitchy, but I'd be worried about selling on a microwave appliance that I'd had apart and done magnetron related work on. Likewise a charity shop. The new Panasonic one only cost £144, and you can get cheap ones for less than £70, so by the time you factor in cost of a new magnetron and shipping a bulky 30kg lump...

[Gyro]

Just as a follow-up. I found the attached PDF which relates to this specific inverter board - and somebody evaluating it for a valve linear amplifier PSU. It gives a good level of detail on operational characteristics and drive requirements (a 555 timer will do).

I'm not clear on the topology - I'm not sure that it counts as a forward converter, despite having two IGBTs (of different sizes but both very high rated) as the emitter of the top (smaller) one is capacitively coupled to the primary and the core is gapped. On later inverters, Panasonic went to a single IGBT flyback converter so this is a bit of an early iteration (circa 2000). It doesn't appear to mind its output rectifier polarity being switched either.

From the figures in the PDF, it is a constant power inverter rather than constant voltage or constant current, and seems to be able to maintain a constant power output to remarkable accuracy (1W mentioned) over a considerable load variation.

The usual safety warnings should apply to anyone reading the PDF, the output is truly lethal and I'm certainly not planning on any lethal HT PSUs. The transformer does appear to be a split (concentric) bobbin though, so I am tempted to try winding a replacement higher current, low voltage secondary.

EDIT: https://streampowers.blogspot.com/2012/09/panasonic-microwave-oven-inverter-hv.html    (same author, less detail)

[Gyro]

I seem to be doing a lot of following up here...

Anyway, definitive proof of the failing Magnetron, artistically arranged with evidence of the true extent of the magnet cracking:



If you look carefully you can see that alternate anode segments, the ones on the antenna driving ring, have been reduced to little shiny molten blobs (melting copper in a vacuum leaves a really nice finish!). Magnetron internals are certainly things of beauty, the closer you look, the more internal structure you see.

It's surprising that the magnetron was still producing a decent amount of power from cold. Presumably the blurring of anode segment edge definition resulted in excessive anode dissipation over cooking time - it's hard to know whether the segments were re-melting each time, there had been no specific abuse, other than the passage of time and the cracking of the magnets.

On the subject of inverter Induction heater construction, I found a 'proof of concept' video on youtube. The inverter uses a newer transformer format, but it is still the older dual IGBT version of the circuit. I can't say that I approve of the isolation between the primary and improvised secondary, but it seems to work rather well - especially later in the video where he adds an additional impedance matching transformer. It seems that the normal 'self tuning' royer converter isn't that important and that fixed frequency operation is sufficient, certainly saving some on some very highly stressed capacitors.



P.S. In the spirit of staying vaguely on topic, the PWM frequency is 30kHz. 

[Gyro]

Out of interest, I remembered that I still had the core of the magnetron from our previous microwave, which died of control panel failure around 15 years ago. I had salvaged the (intact) magnets and thrown it in a drawer! Irrc, this one was rated at 800W (despite being physically the same size as the 950W Panasonic one), and driven by a standard MOT.

Here's a photo of the Anode - showing absolutely no signs of wear:



Together, they cover over 30 years of microwave ownership. Given the lack of abuse,  I have the uncomfortably feeling that, if it hadn't failed for other reasons, the original monster would still be lurking in the kitchen!

I'm bound to conclude that modern inverter units do put significantly more strain on their magnetrons than the older generation. My instinct would have been exactly the opposite, with slamming the magnetron on and off every few seconds with the filament having to warm up each time being much harder on them than a slowly varying PWM supply.

[SeanB]

I find that the magnetron typically fails by either going open circuit on the filament, or by overheating from a clogged cooling fan. Typically the common faults are the door switches, the latches for them, and the diode failing, along with fuses failing from thermal cycling.  I have a largish collection of slightly used magnetrons, from ovens that failed mostly from switches and cavities rusting out, and in general with the common microwave irrespective of the make you typically find 3 different magnetrons in use, only varying in the orientation of the input power connector and the mounting holes for the magnetron, and then you find the shrouds used vary if there is a thermal switch on the shroud or on the magnetron body, though they all have the holes for it already there. Other than that no difference in them, they all use a similar size capacitor, diode and perhaps a high voltage fuse.

Power wise you cannot operate a magnetron at low power easily, you need a certain anode current for it to oscillate, below that it simply is a power diode, and the frequency stability varies with voltage, temperature and anode current, which is why the ISM band at 2.4GHz is so wide, so that the manufacturers can keep the tolerances of mass produced punched plates inside down. 2.4GHz chosen as you could get it back then with cheap components and magnetics, as the first magnetrons used Permalloy magnets, so had to be kept large enough to keep the magnets cool. Ferrites made the magnetron assembly smaller, but the actual body size didn't change, just the magnetic circuit. Probably about the only thing you find that has not been made cheaper by using copper coated steel, or CCA, though the transformers certainly have been optimised to the bare minimum volume and windings, relying on the forced air cooling to survive, along with class H insulation.

[Zero999]

Do you know how much you could get for it second hand, if you repaired it? I wonder if it's worth it, for the satisfaction, even if you break even.

It is tempting, the going rate on ebay looks like about £45 for a pre-owned one. Balancing that is the worry about a potential buyer claiming that it's faulty when he puts it in his mis-diagnosed microwave and demanding a refund.

On the other hand, it is an interesting looking board with a simple interface, Mains in and a 3 wire opto isolated interface - Gnd, input control, and output power sense signals. The primary side is obviously not isolated, but the secondary side is grounded. I have no appetite for 4kV at 300mA, so that would have to go, but it has a nice high current 3V filament winding. It might possibly be a candidate for an induction heater.



Maybe time for a little research before I decide. 


EDIT: Ah, sorry, you meant the Microwave. I'm not sure if I'm a bit too twitchy, but I'd be worried about selling on a microwave appliance that I'd had apart and done magnetron related work on. Likewise a charity shop. The new Panasonic one only cost £144, and you can get cheap ones for less than £70, so by the time you factor in cost of a new magnetron and shipping a bulky 30kg lump...

C704 & C705 appear to be the wrong values. If they were really 8200µF, at that voltage, they'd each be the size of the entire microwave.

[Gyro]

Haha, you've got sharp eyes! I managed to fight my way close enough to the PCB to confirm that they are indeed pF. 

[Gyro]

I find that the magnetron typically fails by either going open circuit on the filament, or by overheating from a clogged cooling fan. Typically the common faults are the door switches, the latches for them, and the diode failing, along with fuses failing from thermal cycling.  I have a largish collection of slightly used magnetrons, from ovens that failed mostly from switches and cavities rusting out, and in general with the common microwave irrespective of the make you typically find 3 different magnetrons in use, only varying in the orientation of the input power connector and the mounting holes for the magnetron, and then you find the shrouds used vary if there is a thermal switch on the shroud or on the magnetron body, though they all have the holes for it already there. Other than that no difference in them, they all use a similar size capacitor, diode and perhaps a high voltage fuse.

Power wise you cannot operate a magnetron at low power easily, you need a certain anode current for it to oscillate, below that it simply is a power diode, and the frequency stability varies with voltage, temperature and anode current, which is why the ISM band at 2.4GHz is so wide, so that the manufacturers can keep the tolerances of mass produced punched plates inside down. 2.4GHz chosen as you could get it back then with cheap components and magnetics, as the first magnetrons used Permalloy magnets, so had to be kept large enough to keep the magnets cool. Ferrites made the magnetron assembly smaller, but the actual body size didn't change, just the magnetic circuit. Probably about the only thing you find that has not been made cheaper by using copper coated steel, or CCA, though the transformers certainly have been optimised to the bare minimum volume and windings, relying on the forced air cooling to survive, along with class H insulation.

Agreed, most failure modes seem to be surprisingly trivial considering the power conversion that is going on. As I mentioned above the only failure previous to the magnetron melt-down had been the turntable motor... and the previous one, a fuse due to an accidental door latch shorting switch bounce. Both Magnetrons were ferrite magnet, from the feel of the fragmented ones, I suspect that they may have been stronger, hard to tell now. I suspect that ratings have climbed a bit on the basis of experience and expected appliance life - 15 years a time isn't that bad. On the melted one there was minimal dust contamination, I suspect that a factor may be differential expansion of the copper anode and aluminium fins - there were at least a couple that were a loose fit.

From what I've read the Panasonic inverters run variable power down to about 50% and then then cycle on and off to get the lower outputs (down to 100W). One thing have noticed is that the current models have a thermistor screwed to the magnetron body, together with a thermal switch - the failed one had neither, so had no feedback of magnetron temperature or airflow.

One failure that does surprise me is the pictures I've seen of magnetrons is the meltdown of the metal antenna cap. The one on my melted anode magnetron was still in perfect shape but there are many images of burnt antennas, eg:



This is presumably due to arcing within the waveguide, maybe from energy being reflected by running empty or contamination? It's hard to think that anode damage wouldn't also result. It seems to be frequent enough failure for replacement caps to be available in large quantities on ebay and the usual Chinese sites!