The short version. An energy window narrower than the factory default doesn’t buy you precision — it buys you false alarms. A photopeak is wide, and it moves a little every day as a healthy detector settles. A properly set window rides over that movement; a tight window lets every small shift spill counts across its edge, which your efficiency check then reads as drift. Unless you have two isotopes whose peaks sit close together, the full factory window is the safer setting — which is exactly why we calibrate the defaults the way we do.

What an energy window is

When your counter measures an isotope, it doesn’t count every pulse the detector produces. It counts only the pulses that fall between a lower and an upper energy bound — the energy window — set around that isotope’s photopeak. For Cs-137 the photopeak sits at 662 keV; for Co-57 it sits at 122 keV. The window is the pair of fences you put on either side of that peak, and the counts that land inside the fences are your measurement.

Set the window well and it captures the whole photopeak. Set it badly — in particular, too narrow — and it captures only part of the peak, and becomes sensitive to exactly where the peak is sitting that day.

The ±10% rule of thumb — and its limit

A common, perfectly reasonable starting point is to set a window to roughly ±10% of the photopeak energy. For a 662 keV Cs-137 peak that is about 596–728 keV. As a general guide for a standard detector, that is fine.

But it is a rule of thumb, not a target to tighten toward — and it is a floor, not a ceiling. There is almost never a penalty for going wider than ±10%, and there is a real penalty for going narrower. The factory windows we ship are deliberately wider than the ±10% rule for that reason.

Why a tight window backfires

Two facts about a real photopeak explain it.

A photopeak has width. On a sodium-iodide detector, a Cs-137 photopeak is roughly 50 keV wide at half its height, with tails that run out well beyond that. A meaningful share of your counts live in those tails. A window that hugs the peak too closely simply leaves some of them out.

A photopeak moves. A healthy photomultiplier does not hold its gain to the last decimal place day after day. The peak drifts a few keV one way or the other as the tube settles — especially in a detector’s first year — and that small movement is entirely normal and within specification.

Now put those together. When the peak shifts a few keV inside a properly set window, almost nothing changes: the peak and its tails are still comfortably inside the fences, so the count stays steady. When the same peak shifts the same few keV inside a window that has been narrowed, a much larger fraction of the peak crosses the window edge — so the count swings. Your efficiency check sees that swing and reports it as drift, even though the detector is doing nothing wrong.

The tell. If one well looks rock-steady while one or two others appear to wander, and the “wandering” wells are the ones whose photopeak position moves a little between checks, suspect the window before you suspect the tube. A stable peak rides out a tight window; a normally-moving peak does not. Widen the windows to the factory default and the apparent drift usually disappears, because you have stopped measuring the peak’s position and gone back to measuring its area.

Why the proper window wins

A properly set window — one calibrated to the full factory width — captures the entire photopeak and its tails, and rides over the normal day-to-day movement of the peak. You give up nothing in accuracy — you are still counting the same isotope’s photopeak — and you gain a great deal in stability. Your well-to-well spread tightens, your efficiency check stops chasing normal motion, and a genuinely drifting detector still stands out clearly because real drift is far larger than the few-keV jitter the wide window was absorbing.

The one time you keep a window tight

There is exactly one reason to hold a window narrow: when a second isotope’s peak sits close enough to spill into it. If you are counting two isotopes whose photopeaks are near each other in energy, or a high-energy isotope whose scatter (its Compton continuum) reaches down into a low-energy isotope’s window, a wide window would let counts from the wrong isotope leak in — spill-up or spill-down. In that situation you trade some stability for separation, and a tighter window is the right call.

But that is a narrow exception. When your isotopes are far apart in energy — Cs-137 at 662 keV and Co-57 at 122 keV have no realistic chance of overlapping — there is nothing to spill, and nothing to gain from a tight window. In that case, wider is purely an advantage.

Larger and high-energy detectors want wider still

Bigger detectors — for example our high-energy build with the 2-inch photomultipliers — produce a slightly broader photopeak and run at a higher operating voltage, so they tend to show a little more normal peak movement, particularly in their first year of service. That is expected for the configuration, and it is one more reason to leave the windows at full factory width on those units rather than tightening them.

What to do

  • Use the factory default windows. They are set wide on purpose. On the LTI isotope library the Cs-137 default is 500–800 keV and the Co-57 default is 90–170 keV.
  • If someone has narrowed them, restore them. Open the isotope library, edit the isotope, and set the lower and upper window back to the factory values. Re-run your efficiency check afterward.
  • Only tighten for overlap. Keep a window narrow solely when a neighbouring isotope’s peak or scatter would spill into it — not as a way to make a measurement “cleaner.”
  • Don’t chase normal motion. A few keV of photopeak movement between checks is healthy. Your window should be wide enough to make it invisible.

Further reading