Getting the two diamond culets flat against each other and centred before you load the cell is the single step that most often decides whether your anvils survive. It costs a few minutes and saves expensive diamonds.
Two diamond anvils only work as a pair. When their flat faces aren't parallel, the load stops spreading evenly across the culet and piles up along one edge — and that edge is exactly where a diamond chips or shatters. Take the time here and you get a cleaner pressure distribution, a more even sample chamber, and a real chance of reaching your target pressure. Rush it, and you can lose the anvils on the very first turn of the screw. Whatever your cell looks like, this step is the same.
It helps to think of alignment as two separate jobs. The first is tilt — making the two culet faces truly parallel. The second is centring — getting one culet to sit directly over the other, sharing the same axis. Almost every diamond cell, whatever its design, gives you a way to do both: one adjustment swings the angle of an anvil (often a rocking seat), another slides it sideways (a translating seat or a set of screws). The order matters — fix the tilt first, because that's what the fringes will show you, then come back and check the centring.
Here's the trick that turns an invisible angle into something you can read at a glance. Bring the two culets together with no gasket between them, almost touching, and shine light through. The thin sliver of air trapped between the faces behaves like an optical wedge, and the light reflecting off the two surfaces interferes — painting a pattern of light and dark bands called fringes. Each band marks a step in the gap of half a wavelength, so the bands are really a contour map of how the gap opens up. The closer you are to parallel, the further apart they spread.
The spacing of the fringes, β, depends directly on the leftover tilt angle θ between the faces:
and the number of bands across a culet of diameter D is simply N = 2·D·θ / λ. That leaves you three pictures to recognise:
A note on the light source. White light gives coloured tints — easy to judge as "one even colour, good" but awkward to count. A single-colour lamp (a sodium or a green source works well) gives crisp, countable bands instead. As a rule of thumb, around green wavelengths each band stands for roughly a quarter of a micron of gap across the culet.
The working target is easy to remember: zero to one band across the flat. If the fringe spacing is wider than the culet itself, no full band fits and you're parallel to within a fraction of a wavelength. For a typical 500 µm culet that means a tilt under about half a milliradian — well under a twentieth of a degree. One catch is worth knowing: the test gets less fussy as the culet gets smaller. On a tiny high-pressure culet a surprisingly large tilt can still hide under a single band, so don't trust a bare "no fringes" there — aim for a perfectly even tint and look hard. The two tools below turn a band count into an angle, and give you the tilt budget for your own culet size.
A routine that works on essentially any cell:
Don't be surprised if it takes a few tries to feel which way to nudge — that's completely normal. With a little patience you'll land on a flat, even field every time.
Once the faces are parallel, the last job is to line up their axes with the sideways adjustment. Most round, many-faceted or cone-backed anvils are symmetric, so you only need to slide them, not rotate. If your anvils aren't round, you may also have to turn one to match — mark the parts so you always rebuild the cell the same way. With bevelled culets, line up the central flats and keep the bevel rings concentric. An off-centre culet, or a sample hole that isn't centred, pushes the pressure lopsided across the chamber.
Once you're aligned, you're set for the step that follows — a few ruby spheres in the chamber will read your pressure from here on.