Microwave oven: why use a voltage doubler?

Factual Questions
1

A friend’s microwave oven died last weekend. In his attempts to service it, he discovered that the power supply for the magnetron employs a transformer to step the voltage up to a couple thousand volts, and then a voltage doubler to further double that to 4000-5000 volts. Yikes. That’s the voltage that fires electrons into the magnetron, setting up the resonant EM fields that produce microwaves.

Question: why do they employ the voltage doubler device? Why not just put twice as many coils in the transformer’s secondary winding so that they get 4000-5000 volts right out of the transformer, and then run that through a rectifier?

2

WAG: Maybe adding twice the number of coils is more expensive than adding a capacitor and a diode?

3

That, and it increases the impedance of the secondary winding, limiting the amount of current the load can pull from it.

4

I’m guessing it is because it reduces stress on the transformer; it only has to withstand half the voltage while the voltage doubler gives you the peak-peak voltage. Although this means you need a large capacitor and diode (a magnetron will work on AC as it is basically a diode), which adds cost and another source of failure, plus the transformer has to output twice as much current, so you need thicker wire for the same amount of copper, so I don’t really see this being much more economical, if at all (perhaps because it takes longer to wind more turns of thinner wire; I have disassembled them for the wire and there is a LOT of wire, perhaps a mile or more).

Some newer microwaves don’t use a traditional transformer/doubler at all and use a switch-mode power supply instead, which can be cheaper despite being much more complex because you don’t have a huge transformer (a SMPS transformer may be 100 times smaller for the same power level due to the higher frequency); this is also why linear power supplies are rarely found in electronics nowadays, even many wall adapters now have SMPSs, unless 50/60 Hz AC is needed.

5

Would the output of a coil with more turns and a rectifier be equivalent to the output of a voltage doubler in every way?

I believe the output of the voltage doubler is pulsed DC, which means that the magnetron is being pulsed on and off at the frequency of the pulses in the AC input. Does this pulsing of the magnetron play a role in the way that the oven functions? Does it lower the amount of energy required to cook food? Does it prolong the life of the magnetron?

6

They add magnetic shunts to do just that, so a higher impedance would actually not be a problem, plus as I mentioned current is halved when you go from a doubler to straight HV (there is no free lunch, a doubler must draw twice as much current for the same amount of power).

Doubling the number of turns and keeping the size of the coil the same would increase resistance by a factor of four (twice as long and half as thick), but the power loss would stay the same since halving the current reduces power loss by a factor of four (P = R x I^2). Voltage drop would double however, but this may not be a problem given they already take measures to limit output current (see above). The output after rectification would be the same, and again, you might be able to omit the diode as well since a magnetron already is a diode. Using pulsed DC instead of straight DC may or may not affect life; pulsed DC has a high peak-to-average ratio thus for the same average current you have a much higher peak and RMS current, which increases power loss. On the other hand, a magnetron needs a high enough voltage to oscillate, so straight DC might cause too much power dissipation and average current draw (and you’d need a really big and expensive capacitor rated at the full voltage).

7

At those voltages you get problems with insulation failure of windings.

Look at the CRT tv ses, you have a flyback transformer that provides the line drive signal, and also feeds into the voltage tripler.

The tripler will bump the voltage up in excess of 24k to set the tue voltage.

One of the more common faults of tv sets was failure of the flyback transformer, and very rarely was it the voltage tripler.

You also get lower losses in a voltage doubler, there are no core losses, or winding losses and no induction limiting.

8

My WAG is so that one model can be made to run on 110V or 220V, the doubler is added for US and other voltage depressed markets of the globe while Europe and and the rest of the 220V world can also use that model without that addition.

9

Sorry, but this is totally wrong - the transformer itself must be rewound, at least the primary, for it to operate off of two different voltages (transformers that can run off of 120/240v use a split primary, connected in parallel for 120 and series for 240). Connecting 240v to a 120v transformer would immediately fry it or blow the breaker because the core saturates and you effectively have a resistor (very small in the case of a MOT) across the power line. Some equipment do use a voltage doubler, like computer power supplies, so you can run the same circuit off of 120/240v, but this is a different mechanism done to keep the input voltage constant (SMPSs can be made to operate over a wide range, but it is easier to design it for a fixed input (DC rectified) voltage). It is also possible to make a transformer for 240v and run it off of 120v with a doubler, but this would be uneconomic since it needs to be able to supply enough power at 120v but also current limit properly at 240 (a magnetron, like all diodes, will try to draw as much current as it can, limited by internal and external resistance).

The most likely reason is probably the cost of winding on an additional several thousand turns of wire, even if the overall size and copper costs can be kept the same (probably also more expensive to make twice as much wire even if it is thinner).

Also of note, most newer CRT TVs and monitors don’t have voltage triplers (a few have them built-in, but if you arc the HV, you can easily tell if it has one because it will snap and crackle due to the capacitors used; these flybacks are also much larger), so that isn’t really a reason to make a lower voltage transformer. Of course, the design of newer flybacks is different enough to be immediately recognizable from the older ones and in my experience they are still good with few exceptions when I pull them out of discarded TVs.

10

The circuit used allows it to work using only a single rectifier diode. The cap in series with the transformer insures that there is no DC flowing in the transformer winding, and that all of the transformer winding is utilized for both halves of the cycle.

To do this without the doubler, you’d need either a full wave bridge, or a center tapped transformer winding. The full wave bridge rectifier requires four diodes, and the center-tapped two. The center tapped version “wastes” half the copper, as only one half is used for each half cycle.(and you’d need four times as many turns)

You also get to ground one end of the transformer secondary, which is a good safety feature.

Also, the doubler configuration used is also known as a “charge pump” in other applications. The capacitor transfers a limited amount of charge each cycle, so acts as a non-dissipating current limiting device. This along with high leakage reactance in the transformer gives a fairly constant current drive to the magnetron, so you get pretty steady output (high utilization of the magnetron) without needing huge, heavy, and expensive filtering elements in the power supply.

But the main thing is making the transformer cheaper. A first order analysis indicates you need about the same amount of copper. (twice as many turns, but half the cross sectional area)

But practice is not so nice. Due to the higher voltage,the insulation has to be thicker, (by twice) and that takes up more room, and that means you need more copper, because the insulation takes up room, so the winding as a whole gets bigger, and the insulation blocks heat flow, So your wire actually has to be more than half the cross sectional area so that it has less resistance and generates less heat to have trouble getting out, and that fatter wire also makes the winding bigger, which means it has to be longer, which means it has to be even a little more fat.

The circuit has evolved over the years, and is actually fairly elegant: Cheap, simple, durable, and easy to trouble-shoot. (with the right training…poking around in a hot microwave oven is a good way for the untrained to become unliving!)

11

MOTs use ordinary magnet wire for their secondaries; the insulation is no different from that used on other transformers, which makes sense because you have many, many layers of wire, so each layer only has to withstand a fraction of the output voltage; the insulation itself is so thin (microns, as little as 0.0002 inches or 5 microns for 40 AWG, thicker for thicker wire) that it adds negligible thickness to the wire, hence why magnet wire is used. The only real concern would be the voltage on the outermost windings with respect to the core.

12

Lots of good responses. Would like to add (or reiterate) a few things.

Transformers are expensive. Especially high voltage transformers operating at 60 Hz. They tend to get very expensive - and heavy and bulky - when the secondary voltage exceeds around 2000 V.

If your load requires a high voltage and a substantial amount of current, then you’ll need to bite the bullet and use a transformer that supplies you with the voltage you need. (Or a fancy switching supply.) However, if your load requires a high voltage but not much current, it’s a lot cheaper to use a step-up transformer in conjunction with a voltage multiplier. (A multiplier circuit has a fairly high output impedance, and is thus only suitable for loads that don’t require much current. Such as a magnetron.) Diodes and caps are fairly cheap, and it’s pretty easy to float the whole multiplier circuit using insulated standoffs.

13

Right, but many actual high voltage transformers will add a layer of fish paper under the last half of each layer of wire (where the voltage difference between layers is the highest).

14

I have taken apart quite a few MOTs for the wire on the secondary but none of them used any additional insulation between layers, which makes sense because there are so many layers that the voltage between each layer isn’t that high, although if you doubled the number of turns with half the wire thickness (same overall dimensions), the number of layers would only increase by the square root (1.414 times more windings per layer and 1.414 times more layers).

Of note, the same is also true for the tiny transformers (similar voltage output) used to run CCFL lamps and newer flyback transformers, which split the winding into sections along the length of the core, as seen in this CCFL transformer (I have opened a newer flyback and found the same design, but IIRC with 20 or so sections, for about a kV per section), although older flybacks with the big disc-shaped secondaries do have insulation between each layer.

15

I found a MOT secondary that I haven’t started taking the wire off of yet and there are about 50 layers of wire on it, for about 100 volts between each layer (maximum between start and end of two layers wound left-right and right-left), which is a reasonable amount for magnet wire, which despite the thinness of the insulation can withstand hundreds of volts (for example, this wire is rated at 600 volts; two wires can withstand 1,200 volts between them, of course, you want to derate for reliability and peak AC voltage if that is a DC rating; the thickness of insulation on regular wire, especially power cords, is mainly for durability against physical damage).

16

This is a defect, too, to a manufacturer. Besides the additional cost in transporting heavier items, consumers don’t like heavy ones. Especially countertop ones like microwaves. So excessively heavy ones will hurt your sales.

17

Voltage doesn’t have much to do with the size of a transformer (within limits); a TV flyback transformer is a fraction of the size of a MOT yet produces 10 times the voltage (but at around 30 watts instead of 1 kW and higher frequency; that said, the transformer used in a microwave oven with a SMPS inverter looks to be similarly sized to a flyback), and CCFL lamps like those in LCD displays run at similar voltages to a MOT but are less than an inch across (that is a meter stick in this picture of a CCFL transformer; I have also found laptops that had even smaller transformers, less than half an inch wide and an inch long, these output 10 watts or so).

I’m still thinking it is because it would cost more to wind an additional 5,000+ turns of wire on the secondary than any insulation problems or the cost of a diode and capacitor (again, insulation thickness on magnet wire can be essentially ignored so you can make a winding the same size with half the thickness of wire and twice the number of turns, which will have the same losses at half the current despite having four times the resistance since power loss changes by the square of the current).

18

Did a little research. In a paper titled Parametric Design Guidelines for MW Oven Inverter the author (C. Bocchiola) says,

(Bolding added for emphasis.)

So it looks like a transformer that can produce 4 to 6 kV will be significantly more expensive than one that can produce 2 kV, primarily due to isolation requirements. For a microwave oven, it’s apparently much cheaper to use a 2 kV transformer and float a voltage doubling circuit at high voltage verses using a transformer that produces the required voltage from the get-go.

19

I don’t see how that would make much difference though, since like I said it is easy to make high voltage transformers that are very small; some black-and-white TV flyback transformers (in small portable TVs) can be less than 2 inches tall and produce around the same voltage, even the ones in large color TVs can fit in your hand; see this video for an example of a small one (being grossly overdriven, the uploader says it is rated for 5 kV in the comments). Here is a CCFL inverter circuit for a notebook computer; the transformer (on the right side) is about an inch long (the Wikipedia page on CCFL inverters also links to this picture of a “micro Tesla coil” which is even smaller).

Also, having a voltage doubler means you have to isolate the capacitor and diode, which really just complicates things; alternatively, you wouldn’t even need a diode, assuming that a magnetron doesn’t care about reverse polarity (but as I said, it is basically a diode itself), plus you have the same high voltages at the output of either.