Yeah, I would advise you to not screw around with a magnetron, for fear of you frying your eyes or getting a high voltage shock.

Go play around with some nice 3V GaAs mmic chips if you want to have some fun.

Not really what I was looking for Biff.  I'm not (as you so eloquently put it) screwing around, but deep into the development of a new type of ignition system for a very lean-burn, heavily stratified-charge ic engine.  The concept works, now I need to down-cost and down-size from the lab-based kit; domestic oven magnetrons are a good place to start.  I was overjoyed when I found this forum as I considered I may be able to reduce costly mistakes by getting direct advice from industry professionals.  I just hope there's someone else out there can give me what I want.  

Be assured that although I understand that playing with microwave radiation has particularly unpleasant side effects if you get it wrong, I deal with very dangerous 'toys and technologies' every day, yet I'm still here to tell the tale.  I'm sure you'll be interested, so if I develop premature cataracts or suffer hideous burns I'll let you know.
 
Thanks for your input!  P

On some engineering questions that are asked, one has to be careful that the post is being placed by someone who is capable of knowing enough to know about the dangers involved. You are dealing with things that can cause injury or death.

In short, the output of an oven magentron extends into a rectangular waveguide which is then connected to the cooking chamber. If you instead connect the waveguide to another piece of waveguide which has a coaxial pickup 1/4 wavelength ahead of a short, you can create a transision.

A quick google of "magnetron" "transition" "coax" found the following example where the end wall is movable (to fine tune to the magetron frequency). http://www.iap.tuwien.ac.at/www/atomic/publications/reprints/WU_RSI94.pdf

However, you need a RF engineer or RF consultant involved in your project along with the proper equipment. Gaps can make nice radiators (cook your fingers or eyeballs). Incorrect dimensions or transition placement can create large reflections (arc, sparks, smoke). High voltages are used to power the magetron (zap!).

I worked in a company once where we had a big magnetron power supply.  We developed some high power switches, but despite having collectively over 50 years of microwave experience, no one was willing to turn the darned thing on.  I think it was the X-ray warning that caught all of our eyes!

Unless you are familiar with microwaves, you will certainly have high power leakage out of, what to yoy might look ok, the tiniest of cracks.  You can cause a fire, burn out your eyeballs, blow up your test equipment, and generally have a very bad day.  It would take about a year of college level courses to explain to you what the dangers are, where they can arise, and how to mitigate them.

Go hire an experienced high power microwave consulatant to at least get you started.

If you are determined to go ahead anyway, rent one of these

http://www.narda-sts.de/en/produkte/emr20.htm

treat the whole project like a mean and liver rattlesnake, and sniff for RF leakage EVERY time you turn the project on.

I would also buy yourself several 1m x 1m sections of radar absorber material (RAM). The pyramid surface type would be good. The point is that just shielding may not be enough. The high leakage signal will bounce around the lab and get you by a non-obvious non-line-of-sight path! Hide the experiment behind the RAM then you can safely use your safety EM field meter.

You want to put the magnetron output into coaxial cable. Forget that. The leakage from a standard coax cable at 2GHz will be too high, regardless of the integrity of the joints. You will need semi-rigid or rigid coax. The solid outer will be a much better shield than a braided coax screen.

Let's suppose you find some 50 ohm semi-rigid. 800W into 50ohm means an RMS voltage of 200V (say 300V peak). It also requires 4A which is not a small amount of current.

Do pay attention to the warnings.  But don't let them
scare you away from the project.  Just be careful, and
aware of the hazards.
Think waveguide-to-coax transitions, and you will get
closer faster to the goal.  There are other ways to
couple to an antenna that do not involve coaxial cable,
but do require precision machine work.

Look here for an example of what is involved:
http://www.comm.ttu.edu.tw/Chu/fly2ideas/fly2R&D/hypethermia/hyperthermia.htm

This vendor stocks 7/16 coax transitions, and shows
N-type and others as available as special order:
www.ibf-electronic.de/downloads/ibf_en.pdf

Keep us posted.
als

People... thank you all.  After a bit of a dodgy start (I'm sure you had my best interests at heart Biff44), you've collectively pointed me in the right direction.  I'll let you know how things work out.

Cheers and festive stuff... P

tinwelp,
I've taken power out of microwave ovens before using a waveguid to coaxial probe, common at Pasternack http://www.pasternack.com/detail.aspx?ID=72847. I didn't connect it directly to the oven, but spaced it apart since I only needed 100 watts.
Not sure how you plan to control the power level and duty cycle. The coax to waveguide transition is very simple for one frequency and easily screw tunable.

I assume you're creating an arc to explode some gasoline.

Kevin

Kevin,
I'm interested in your hinted solutions at power modulation.  My first plan was to reduce the collection antenna efficiency by tuning a little away from the optimum lambda/4, but it looks as if this will result in nothing but death and destruction (or at least something getting very hot).  Currently I'm looking at switching the HV supply to the magnetron or more likely the transformer input, but ideally I'd like to have a solution that would allow the magnetron to be modulated between 50W and 500W with continuous output.  Time restrictions and the holiday have prevented much in the way of progress, so I've yet to build anything and a directional change is not out of the question at this point.  If you have any suggestions, I'd be all ears.

The technique I'm using does not intentionally form a 'spark'.

Cheers... P

tinwelp,
I think you've got a challenge to modulate the RF in a simple commercial cost oriented manner. It'll take some effort.

Curious what timing is needed for on/off and duty cycle.

Most standard RF switching is high Power PIN diodes or ferrite switches. That's not a good commercial answer though due to cost.

I think that modulating the power source is the only economical way you can do it without blowing up the Magnetron. Since the present equipment already can turn things on and off, maybe you can optimize the speed or change the voltage to change the Magnetron output power.

I'd be willing to assist you. I'd proposed a similar technique for a new propulsion company in Goleta Ca. for their small 2 cylinder diesel engines, but they didn't go that route yet. You can get my email from the info below.

khiggins
www.toyon.com

tin, my experiences are somewhat parallel to yours working with oven magnetrons and directing their power in different ways.  One early lesson I learned was that leaking microwave energy can set off fire alarms. (Just another piece of the kind of advice we are looking for which could never come from a book.)  Another idea re power, the simplicity of the transformer supply has its merrits, but Panasonic uses an inverter solid state supply which I have standardized with because it offers power control (reduction) using a high frequency switching cycle.  Question is if your switching requirements excede those of the inverter. I wouldn't switch the high currents through the transformer style supply at the rate an engine would require.