Tests with Silicon Hairsprings

Report on trials with silicon hairsprings
at the Musée International d'Horologie (2):

© text: MIH, translated by M. Hanke, pic: MIH, 2002. 2006



Movement used: Unitas 6445, 18 1/2, 18.000 A/h, lift angle 49°

We have received 25 hairsprings made from silicon and measured their dimensions with a Hauser pointing machine. The thickness of the silicon plates varied between 0.025 and 0.12 mm.

After having produced a movable stud, and with a modification of the balance staff, a first test worked. The spring was glued onto the balance staff, and a brass stud was also glued on the other end of the spring plate. However, the balance wheel’s frequency was far too low. This spring broke, while we tried to dissolve the glue (# 19, 0.050mm thickness).

The second test, conducted with hairspring # 20 (0.080mm), was successful. The performance was checked with a Witschi Wicomètre on which the frequency can be adjusted. With a defined frequency of 18.375 A/h, the results were the following (compared with a movement with conventional hairspring).

The observational position is horizontal (face down), the movement with silicon spring is wound with only three turns (of the mainspring) since it rebanks at maximum winding.

spring temperature amplitude deviation
silicon 22° C 282° -6s
37° C 283° -59s
5° C 271° +47s
conventional 22° C 294° +35s
37° C 295° +33s
5° C 279s +47s

The results in different positions at a temperature 22 degrees Celsius are the following, taking into account that there are no regulating pins to modify the positional variations.

Dial up -5 s 282° ampl.
Crown up -15 s 244° ampl.
Crown left -19 s 250° ampl.
Crown right -12 s 247° ampl.


The silicon springs are very fragile. They break easily, and have to be handled with great care. The glueing can be achieved troublefree with a superglue (cyanolit), but it is very difficult to dissolve the glue again (dimethylformamid), and to clean the hairsprings for re-use.

The performance at a fixed temperature of 22 degrees Celsius is good. Without further regulating device, with the hairspring simply glued to the balance staff, the results at different positions show a variation span of 14 seconds at maximum.

However, the variations at different temperatures are very large - 106 seconds between 5 and 37 degrees Celsius - whereas a movement with conventional Nivarox hairspring only varies for 14 seconds.

These different tests were executed with only one silicon and one conventional hairspring. However, the results are conclusive enough in order not to insist with a larger sample.

Ph. Pellaton, February/March 2002


Copyright January 2006 - ThePuristS.com - all rights reserved