I am working on a microwave oven that starts to emit an odor after 2 minutes of full heating despite otherwise working correctly. The smell is the same as that "sweet", overheated coil varnish smell that anyone with much electronics experience knows well. It seems to be coming from the magnetron (inside the area where all the electronics are mounted, not the cooking cavity).


When I measured what seemed to be low voltage going to the magnetron, it dawned on me: Since magnetrons are somewhat mystical in their operation to me, is it possible that a lower voltage could cause them to overheat? Would it matter if the electrons took a different path inside the tube?


To assuage responders' questions about my methods, the following details are presented.

The magnetron's wave-guide looks clean via a bore-scope and the cooling fan spins.

With a multi-meter, I tested the magnetron, high voltage (HV) transformer, HV diode and HV capacitor while each was disconnected; all test fine that way.

I swapped, one at a time, the HV capacitor and the wave-guide cover, but the smell recurs after the 2 minutes. I replaced the magentron, too, and the smell changed slightly but was still present.

I can't see the HV diode causing such an issue, but I may try swapping it next.

I have an old, 30 kV, analog, DC voltmeter that I used to test the HV circuitry (I don't know of its accuracy). According to the oven's service manual for the oven, approximately 2400 volts should come from the HV transformer which is then "doubled" to about 4000 V and then fed to the magnetron. When disconnected, I measured 3000 V (maybe less) on the lead that is normally connected to the magnetron.

A new, replacement HV transformer is not available and a good used one may be hard to come by.

(If relevant, I remember replacing the HV capacitor years ago and noticed that that capacitor is rated for only 2100 V [what the appliance parts shop ordered for me back then]--I do not recall the original working voltage of that capacitor and it is not in the manual.)

Edit: Repair

For those who are curious about the end result of the repair, I fixed the magnetron overheating smell with an exact replacement, HV diode change. I didn't expect this. Again, the old diode behaves normally on a meter, forward-biased vs. reversed (though it measured 1/3 lower resistance in the conducting direction than did the new diode--I assume it has gone out of spec.). But I now measure 1 kV DC with the magnetron connected (versus 1.5 kV DC with the old diode), 2 kV DC when it's disconnected (versus 3 kV DC with the old diode) and there is no smell even after several minutes of full-power use. Thank you all for your helpful comments.

  • 2
    \$\begingroup\$ Do realize that magnetrons deteriorate when used, their output power decreases over lifetime, see: hunker.com/12592740/do-microwave-ovens-lose-power-with-time and: electronics.stackexchange.com/questions/380899/… so if this magnetron already has many hours of operation it might be a good time to just replace it. \$\endgroup\$ – Bimpelrekkie Sep 26 '20 at 20:54
  • \$\begingroup\$ @Bimpelrekkie Thank you, but that doesn't address "smell". \$\endgroup\$ – kackle123 dances with Monica Sep 26 '20 at 23:22
  • \$\begingroup\$ And that's why it is a comment and not an answer. Without a thorough investigation it is impossible to answer without making assumptions and/or guessing. \$\endgroup\$ – Bimpelrekkie Sep 27 '20 at 12:40
  • \$\begingroup\$ @Bimpelrekkie I understand. I also originally mentioned that I replaced the magnetron and still had the smell. \$\endgroup\$ – kackle123 dances with Monica Sep 27 '20 at 15:09
  • \$\begingroup\$ Are you sure an insect hasn't crawled into the transformer (or somewhere else that gets hot) and died? That can have a similar smell. \$\endgroup\$ – pericynthion Sep 27 '20 at 18:13

A cavity magnetron will be overheated by it's cathode heater if there is not sufficient bias voltage to operate correctly. Correct operation depends on thermal electrons being emitted in sufficient quantity and swept away to keep the cathode cool.

For this reason a 'conventional' microwave oven operates in On/Off mode to get reduced power (for defrosting).

An 'inverter' microwave oven has independent control of the cathode heater, and turns down the heater current to operate the magnetron at reduced voltage for reduced power.


Magnetrons don't have varnish in them, it is likely to be the transformer. Try measuring it's Temperature, it shouldn't be more than 70 centigrade.

  • 1
    \$\begingroup\$ Too short of an answer (actually, it's not really an answer so much as it is a suggestion), but you hit the nail on the head with respect to the right next question for the OP. It's temperature! I almost could not believe all of the details provided by the OP and yet to have the temperature unmeasured. \$\endgroup\$ – jonk Sep 26 '20 at 21:04
  • \$\begingroup\$ jonk, I plea guilty for both crimes, I was tempted to make contribution based on experience. \$\endgroup\$ – Moty Sep 26 '20 at 21:22
  • \$\begingroup\$ @Moty When you look at a magnetron dissection on-line, you can see very coarse/short filter coils right at the inputs. They look to be copper and are, therefore, likely varnished. I'm am 99% sure the smell is at the magnetron. That said, you're right; I should verify that with a thermometer! (Although, I don't have access to an IR thermometer until after the pandemic, unless I buy one.) And it still doesn't answer my question about lower voltage causing flying electrons to miss their target. But thanks for your reply! \$\endgroup\$ – kackle123 dances with Monica Sep 26 '20 at 23:42
  • \$\begingroup\$ I wasn't clear with my advice. For security run the oven full power for 3 minutes. Disconnect it from mains and measure temperatures with any thermometer. The temperature vary slowly and you don't need accurate readings. \$\endgroup\$ – Moty Sep 27 '20 at 12:50
  • \$\begingroup\$ The magnetron is always powered by a voltage doubler that gives pulses of HV , the voltage always vary between HV to 0V 50 times a sec, so it's not likely to take more current at low voltage. \$\endgroup\$ – Moty Sep 27 '20 at 13:07

Seems unlikely to me. The value of voltage you measure at the magnetron with an ordinary DMM will be a bit suspect because it's half-wave pulsating DC.

Check that the fan that cools the transformer and magnetron is working properly and air flow is not obstructed. They run the transformers in old-fashioned (non-switching power supply) microwave ovens very hard to save on cost and weight. If you compare the size and weight of a similar VA transformer that is rated conservatively for continuous operation without forced convection you'll see a large discrepancy. The smell is more likely to be coming from the fan motor or the transformer than the magnetron I would guess, because both of those have windings that are exposed to the air.

  • \$\begingroup\$ Good point about the pulsed DC on my analog meter. I'm 99% sure the smell is from the magnetron. The fan does not cool the large transformer in this oven, it is pointed only at the magnetron. Its air flow is not obstructed. The transformer doesn't get as hot as the magnetron (by feeling the air around it). The fan is spinning, but I haven't measured it with an RPM meter before so I can't tell if it's slightly slow--it looks proper from my memory. Regardless, I just ran the oven with another fan added, and the smell still returned. \$\endgroup\$ – kackle123 dances with Monica Sep 26 '20 at 23:29

It is normal for a magnetron to heat up way more than the transformer.

Assuming here the usual domestic microwave oven with 1000..1200W input power and 700..800W microwave power.

~1 kW transformer is 95..98% efficient and has 20..50W of heat to dissipate, the magnetron is like 70-75% and has 250..350W heat. That's why it has fins and fan for cooling and the transformer doesn't.

Whatever coils there are in the magnetron, they are not insulated by varnish. Neither the working temperature nor the voltages present inside suggest the use of the same varnish that is used in transformers. They are also sealed. This means either no smell or different smell when they do overheat.

Measuring a distorted half-wave-voltage-multiplied-non-linearily-loaded voltage over the magnetron is hardly going to show anything. It is not the usual thermionic valve power supply that is ironed out by capacitors and inductors.

Measuring the AC output of the unloaded transformer may tell more.

The temperature of the transformer (or even the magnetron) can be measured by fingers (with some care taken) or by contact thermometer. Both of them are grounded outside. Then again the usual failure mode for a transformer is a short between windings that is not going to significantly increase the overall transformer temperature.

These people likely know what they are doing and they are not affraid neither of lowering the anode supply voltage nor of heating the magnetron with an independent power supply while doing something to the anode power. In fact, I am yet to see some tube schematics that modulates the heating according to the load. Some tubes even DO overheat when subjected to lower than the required heating.

Lowered anode power supply is also how microwave ovens are modified to be used in RVs/trucks with limited available peak power.

In short, if it smells like a burnt transformer, ...

p.s. I wish I was that brave (or qualified) to measure the anode voltage of a magnetron for a stake that low (a new microwave oven is like $50 and can be scavenged for all parts if the old one carries some sentiment). The last time I did something to a microwave oven, I connected mains AC to the output of the transformer and measured the lower voltage at the input and the heating voltage.

  • \$\begingroup\$ Re: varnish, I understand what you mean, but it's hard to say for certain either way; copper must be oxidation-protected somehow. The sealed-tube section of the magnetron is in the airflow, the bottom with the air-holes is not (you can see the input filter coils through the holes); the bottom seems to be where the smell is. There is no smell near the transformer, but I may externalize/separate the magnetron to be sure. I wish I had an HV AC meter. I am a coward when it comes to HV, so I play it safely (no hands). The cost is irrelevant; I must understand. I will see your URL; thank you! \$\endgroup\$ – kackle123 dances with Monica Sep 27 '20 at 15:49
  • \$\begingroup\$ I looked at the paper pointed to by your URL. The schematic seems to show that they reduce the filament's power (see reference "K") as well. \$\endgroup\$ – kackle123 dances with Monica Sep 27 '20 at 19:06

Magnetrons don't conduct until the voltage is within about 100V of their rated voltage, because the magnetic field that spirals the electrons around inside steers them back toward the cathode. The voltage across the magnetron is an almost perfect square wave going from about +10V (the diode has about 10 junctions in series) down to -4kV, so, with a meter, you will read about 2kV. If you have a CRO with x10 probes and 18 10MΩ resistors you can look at the waveform across the magnetron by connecting 9 of the resistors in series with the CRO x10 probe. The capacitor has a 2.1kV sine wave across it. that's why it is rated at 2.1kV. You can look at its voltage by using 2 strings of 10 meg resistors and both channels set in A+B mode with Ch B inverted. The reading will be 100x the difference between Ch A & Ch B. Click here for more details. What are the bare minimum parts needed to operate a magnetron? enter image description here

Source of diagram: my own experiments.

  • \$\begingroup\$ Very interesting. I originally measured ~ 3 kV unloaded with the DC voltmeter. Do you expect 2 kV DC with magnetron to be connected or disconnected? And are you saying the 9 resistors make it safe for the oscilloscope? \$\endgroup\$ – kackle123 dances with Monica Oct 13 '20 at 2:27
  • \$\begingroup\$ Connected. Otherwise you won't get a square wave. You will get a level shifted sine wave. I'm not sure exactly what voltage to expect, but 3kV looks about right. The 9 resistors should be safe. Just keep them insulated from everything else and don't use ones that are too physically small. The voltage across each one & on the oscilloscope x10 probe will be 0 to -400V square wave. Check the voltage rating of the probe. It should be good for about 700V. \$\endgroup\$ – Peter R. McMahon Oct 16 '20 at 7:55
  • \$\begingroup\$ @kackle123 dances with Monica My CRO probes are rated at 600Vdc or AC peak. If you live in Australia, you can get packs of 8 10M 1% 0.5W resistors for 85c a packet from Jaycar Electronics. They will stand 400V easily. Cheers. \$\endgroup\$ – Peter R. McMahon Oct 17 '20 at 1:03
  • \$\begingroup\$ @"Peter R. McMahon" Thanks for the replies. I was using this math to guesstimate, but it probably doesn't apply to a square wave: electronics-tutorials.ws/diode/diode_5.html Can the oven and the CRO use the same mains/ground? Are 1/4-watt resistors okay? I will try something, but it'll take a bit due to an injury. \$\endgroup\$ – kackle123 dances with Monica Oct 17 '20 at 19:09
  • \$\begingroup\$ @kackle123 dances with Monica. Sine wave only. If you are measuring the average DC of the magnetron voltage the meter, on DC, will give you the average DC voltage (-2kV for a 0 to -4kV square wave). Use the same ground for the CRO. If you are looking at the waveform on the transformer, or capacitor, the divided down peak voltage(s) above ground might still be a bit high for your x10 probes. Instead, make up 9M from 2 10M in parallel + 2.2M + 1.8M. The 1/4W should be ok, but, just in case, try them to ground first & run the oven for a minute. If they don't go poof!, they are ok. \$\endgroup\$ – Peter R. McMahon Oct 19 '20 at 0:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.