The Rod-Lens Endoscope

Do the old design first, because the comparison is the point and you need the baseline honestly made. Set up a relay of small plano-convex lenses on a rail inside a tube. Each pair of lenses re-images the previous image plane onto the next, so the picture is carried down the tube in a series of hops. Place a resolution target at the object end and an eyepiece at the other. The two things to get right are spacing and stops. Space the lenses so each intermediate image falls between pairs, not on a glass surface, or every scratch and speck on that surface joins the picture permanently. And note carefully where the beam is widest — the aperture stop of the whole system is wherever the beam is most nearly filling the available diameter, and in a small-lens relay that is almost always at a lens rim rather than anywhere you chose. Measure two numbers before moving on. Photograph the target through the eyepiece with fixed camera settings and record the image brightness. Then find the finest target group still resolvable. Write both down. Everything in step 4 is a comparison against these.

Cut rods from borosilicate or, better, from a higher-index glass if you can get it. Length is not critical to a millimetre but the two end faces must be flat, polished, and square to the axis — this is the lens grinding skill again, applied to a short cylinder, and the flatness matters more than anything else you will do today. Grind a shallow convex curve onto the faces that need power, and leave flat those that do not. Work through the grits and finish on pitch with fine oxide. Check flatness by interference against a known flat if you have one, or by autocollimation off a mirror. Assemble the rods into the same tube, at the same overall length, imaging the same target. Everything except the relay itself must be identical to step 1 or the comparison is worthless — same illumination, same eyepiece, same camera settings, same target distance. Photograph and measure. Compare brightness against your step 1 baseline, and compare resolved target group. Expect a large brightness gain and also a resolution gain, the second because the wider aperture reduces diffraction as well as admitting more light. If your measured gain is well below the notebook's prediction, the usual cause is that your rod end faces are not flat and are scattering. Look at the bright field through the relay with no target: scattered light shows as a milky haze, and the haze is your polish quality made visible.

There is a second half to the instrument that is easy to forget. A relay this bright is useless without illumination at the far end, and a lamp small enough to fit down the tube gets hot enough to burn tissue. The answer is to leave the lamp outside and pipe the light in along a bundle of fibres running beside the relay. The heat stays at the lamp. This is why the modern instrument is called a cold light source, and it is the other half of the same insight — glass carries light better than air over distance. Add a fibre bundle alongside your relay, illuminate from a lamp at the bench end, and note that you can now hold the far end of the instrument against your palm indefinitely. That is the whole point. Where this stops. You have built two optical relays, measured the throughput difference, and separated the contributions to it. That is real optical engineering and it transfers to any long imaging path — borescopes for engines, inspection of castings, periscopes. It is not a medical endoscope. A clinical instrument must survive repeated sterilisation without the cement between elements clouding or the seals admitting fluid, must be leak-tested before every use because a breach admits contamination into the next patient, has a specified field of view and angle of view that the operator relies on for orientation, and is used by someone trained in where the instrument may and may not be advanced. Point yours down a length of pipe, into an engine, or at a casting. Not into a person or an animal.