Specific sections

What's needed to build this synchrotron?

There are three main things that need to be acquired before anything can be built:

  • Money

    The most perplexing. I'm not good at fundraising.

  • Space

    My garage is only a last ditch option for building this synchrotron. If it were built in my garage, it wouldn't have enough power, would drive away the insurance people, would get exceedingly dirty, and would possibly stir up anti-amateur-scientist resentment in the neighborhood. One person tried to bring a cyclotron to his house and faced a flurry of outrage, even though his installation was harmless.

    Ideally a space for the synchrotron would be tall enough to house some 42U racks and provide 10kW of power in 120V and 240V (or three-phase). It probably would take up a space about 5x5 meters, so that it could be comfortably operated and properly shielded.

    Unfortunately, lab space in colleges and universities come at a premium: the potential for lots of red tape and competition. For a lab space to work best, I need a professor who can back me through all of those.

    It's not out of the question, though.

  • A Design

    Working on it, see below.

Design Overview

Design Methodology

Since I'm working with limited resources, I figure the best way to engineer a synchrotron that'll get built with my timeframe and budget is to design around them.

Following are design specifications based on back-of-the-envelope calculations or restrictions.

1 to 5 MeV electrons

I'll probably be accelerating electrons because they're much easier to deflect, which means smaller magnets for a given energy and ring radius. Moreover, electron sources are simple: a filament with a couple of HV electrodes is enough to provide a (reasonably) focused beam. An example of a home-built electron gun showing a relatively collimated beam is available on Teralab.

Electrons suffer extremely high energy losses in circular accelerators, but only at high energies, due to synchrotron radiation. At 1 MeV synchrotron radiation losses should be minimal, and the radiation that is emitted is in the form of low energy photons like radio waves. There shouldn't be any X-ray hazards present within the accelerator itself, save for a beam dump or other occasion when the electrons strike a target. (Bremsstrahlung from electrons crashing into a target is pretty much proportional to energy, so any targets should have heavy shielding: 1MeV X-rays are pretty scary.)

Plus, since electrons are so light, they're relativistic even at low energies like 1MeV (they're at about .95c). This makes design of the RF cavity and circuits easier, since it limits the timing variance of the electrons.

In exchange for easier main-ring RF design, a RF linac will probably be needed for the injector. Maybe drift tubes with varying length/gaps? Another alternative may be a small betatron or a classical microtron.

1MeV is extremely low. Small betatrons can easily do 10MeV, so I might aim for a higher energy. The peak energy may end up depending on how much power is available for magnets and RF. Shielding becomes a major consideration when it comes to higher energies; the absolute higher bound on energy is the amount of shielding present---I cannot operate at higher energies than my shielding will allow!

2 cells of FODO lattice

Focusing optics are hard. The math involved in calculating exact particle orbits in a focusing lattice is way beyond me, so I'd like to keep the focusing as simple as possible while still remaining a full-fledged synchrotron. The Small Isochronous Ring is the closest match to what I have envisioned: a relatively simple 2-cell machine. In its inital configuration it had four magnetic quadrupoles and four electrostatic quadrupoles; I'll probably go for simpler design and have 4 magnetic quadrupoles total.

For clarification, the FODO lattice I'm thinking of using is the typical type: a focusing quadrupole followed by a drift space (dipole), followed by a defocusing quadrupole and another driftspace (dipole). A triple achromat cell would work too, but the FODO design is cheap, and I don't really need the astigmatism correction.

A 1-cell model is feasible but has a limited number of straight portions, limiting expansion.

A note: both SIR and UMER concentrated on studying space-charge effects; I'm just building it to work. Therefore, my job should be a lot easier: I can operate in any mode.

I have not run the numbers yet to find out about magnet design. At these energies it might be possible to get away with an air-core design at the expense of massive power supplies.

2.75" Conflat parts where possible!

I don't see the beam getting any bigger than an inch in diameter (I'd at least like to minimize emittance in case I want to run target experiments), so a 2.75" flange presents a big enough apeture for the beam.

Supply-wise, 2.75" CFs are ubiquitous. Plus, they're also cheap due to mass production, and are easy to find in surplus stores and on eBay.

It might be possible to build the entire synchrotron with minimal vacuum chamber machining; maybe I could make the main beamline out of Conflat right-angle bends and straights and Ys and clamp the magnets around the tubes. I haven't run the numbers yet, but I don't think that the radiation loss would be too bad if the dipoles were crammed onto a standard-sized bend (such as these parts from KJLC).

The injector and preaccelerator/linac might involve the most machining. The main ring and injector RF chambers will also be challenging.

2.75" CFs seem about right; any smaller and the gas conductance is low, making outgassing harder. Any bigger and the costs skyrocket.

A 7200 Watt limit

In the worst case scenario, I'll be building it in my garage, placing a 7200W limit on the electrical power. (I have one 120V 20A circuit and a 240V 20A circuit currently wired for experiments.) I'm not sure about the actual capacity of those sockets (while they're rated for 20A each, we have a single 100A service to our house), so the actual limit may be 6000W or less.

This may be the death knoll, since just the vacuum system itself could easily consume 1000W+. I'll have to decide whether to scale the rest of the parameters down to work in the garage, or to go on a prolonged search for lab space and more power from the mains.

A 1 meter ring diameter

I'd like to keep the ring small enough that it could fit in a small room. In a best-case scenario, the ring would be on a 1m square platform bolted down so that it doesn't move; the platform could be on wheels for transport/storage, while the electronics would be in standard 42U racks. It would be pretty neat to have a synchrotron that could be easily taken from university to university for demos and such, though it's not a requirement.

Safety, safety, safety.

Lead. Lots of lead.

Before anything else, though, the largest issue is safety. When a 1MeV electron beam crashes into anything, large amounts of highly energetic X-rays are released. If the beam is ever lost at energy, then X-rays will spew out of the ring in every which direction.

Therefore, adequate lead shielding is a top priority. As mentioned previously, my capability to shield from X-ray and beta radiation will be an absolute requirement. I don't plan on using any radioisotopes, so alpha and neutron radiation shouldn't show up.

Beta radiation is easily blocked by nearly anything, but I must take care: it's easy to stop the electrons themselves, but their energy continues on in the form of X-rays.

Large amounts of lead will be key.

Necessary Vacuum Equipment


As far as vacuum equipment goes, it should be pretty simple given proper equipment. Ideally there should be one or two turbomolecular pumps, hooked up to two scroll pumps. This would allow for an oil-free operating environment, reducing the need for baffles and making life much easier. The cleaner environment would ensure a long beam lifetime, allowing the bunch circulation to last for as long as the optics let it. Usually the residual gasses limit bunch lifetime.

A design with no consumables would be best. Having to fill the system with dry ice, acetone, alcohols, or LN2 would increase the operational cost very quickly and would be pretty inconvenient. For cooling systems, there might be a compressor-based or Peltier-based system that I could get away with instead.

In the worst case scenario, I can use a diffusion pump in series with a regular rotary vane pump. While it's old technology, I've dealt with both before, and maintenance is simple. The only issue would be oil contamination, which can be contained with proper use of baffles and clean oils (Santovac 5 if I can afford it). It'll probably lead to lower beam lifetime but that's a feature, not a requipment. For the initial testing purposes, a vacuum good enough for one or two turns is sufficient.


While Baratron gauges would be very nice for measurement, they're expensive. TC gauges will probably be used due to their simplicity, price, and availability. Standard-issue ion gauges will probably be used on the main beamline in one or two places to make sure that the vacuum is good enough.