Separator thoughts -- R. Baartman

Separator mass resolution is an engineering issue, not a beam dynamics issue. Given correct mechanical, electric and magnetic field stability, can always get the resolution if not concerned about transmission. Transmission is a beam physics issue. Optically, it's meaningless to talk about resolution apart from phase space acceptance.

Experience with our 135 degree separator

After 9 years of using the Chalk River separator magnet, we've never gotten close to the calculated resolution of 10,000. This is not because of anything intrinsically wrong with the separator magnet.

We tune to get something like this: (0.5mm by 12mrad) Other examples: taller, latest-best,, envelopes S-shaped Octupole movie
But the high resolution design was 0.13mm by 60mrad!:
Or, comparing on the same scale: (N.B.: 120 mrad full width is 24 cm wide beam in dipole centre.)

We've never even gotten close. Why not?

We did not finish the engineering:

• slits not controllable to fractions of mm,
• hall probe only good to .3 gauss (that's 1:5000 for A=24),
• aberrations are bad: optics incorrect,
• insufficient intensity i.e. lack of an offline source
• ...

Why did we never carry out the good intentions?

Get the science out. Given a choice:

1. Beam now, with some contamination, or
2. allow beam physicists a couple weeks to commission high resolution,

Can we set it up once and forget it? No. A high resolution tune is hard to maintain because it is very fussy/sensitive to everything.

So where's the rationale for a resolution 10,000 separator?

What did we do right?

• Allison Emittance scanner at mass slit. It allows operators to set up first and second (sextupole) order correction without help from physicists.

What have we learned (that we would do differently)?

• Provide an offline source upstream of separator so we can commission it independently.
• Use vertical constraints (not just horizontal slits) to aid reproducibility.
• More diagnostics to help match out of separator. This is an often-overlooked issue. Magnification from mass slit to transport system is >> 1.
• Should have given more thought to low resolution setup. As it is, it takes too much time to set up a simple low res tune.
• Should arrange for ample commissioning time.

What did we know beforehand (but still have not verified)?

• Source energy: ΔV < 1 volt.
• Slit position, width, etc.: Δx < 50 μm
• Magnetic field (including AC noise): ΔB < 0.1 Gauss

Cost scaling

So clearly it's not economically sensible to go to extremely high resolution.

Possible Design