I recently had the questionable pleasure of restoring a Eureka clock. These clocks are very desirable among a certain class of collector, I count myself among them, however their rarity and price makes them a little luxury.

If you don’t know Eureka clocks, the picture of the clock in question will show you the main feature of interest to the owner – the oversize balance wheel. In this case, the balance is behind the dial in a 4-glass timber framed case. Most Eureka clocks have a rate of 1.33333 seconds per beat, but this is one of the more unusual 1.5 second rate. They run on a 1.5volt battery – originally one of those big cylindrical batteries that readers of a certain age will remember from their high school science class days. 

The main problem with Eureka clocks is the sensitivity of the rate of the clock to battery voltage, and similarly to the condition of the electrical contacts for the impulse coil. For this reason, many people have tried over the years to make a ‘gizmo’ that will transform the clock into a good timekeeper. Brian Mumford (of Microset Clock Timer fame; see www.bmumford.com) has designed a device to make them keep good time, however for the last few years his website has said ‘not making them at the moment – please contact me and I’ll let you know when I’m making some more’. I’m still waiting.

Fig 1. The Eureka Clock

Fig. 2 The stuff removed from the clock

This clock had been tampered with by the late previous owner of the clock; the father of the current owner. He had removed the original two-ball balance bearings and replaced them with standard caged ball races, removed the original electrical contacts, first trying replacing them with a new set of contacts which apparently didn’t work as they weren’t connected, then following that failure, a Hall effect sensor (which works on a magnet attached to the balance). He had added a circuit board of about 10 IC’s which sensed when the clock was behind or ahead of the correct time, and changed the length of the impulse to the balance to compensate.

At first I thought that the clock probably had worked at some time since modification, and that I just needed to repair the existing system. After first working out how the electronic modification worked and that it was still working as designed, my next problem was lack of amplitude. I tried many things over a period of several months, including cleaning the ball races, replacing them, changing them to different sizes etc. etc. etc. After a long and unproductive series of attempts, I concluded that I had to take the bearings back to the original design and so I also decided to say goodbye to the electronics and take the clock back as close as possible to original. 

First step – the electrical contacts. The contact assembly required a jig made up to make the form of the springs, the springs were made and a flag of silver riveted to them. The contact on the balance wheel was then made up. Pretty good instructions on this matter can be found at davewestclocks.co.uk under the Antique Clock Repair link, so won’t be repeated here.

Next, the bearings. I had a copy of an article from “The Clockmaker” magazine which presented the plans of a Eureka clock so you could make your own.  The original bearings were a simple hardened steel outer race, two balls and the steel part of the balance arbor which rests on the ‘V’ formed by the two balls. The balls were the easy part – you just buy a packet from the bearing supplier (Miniature Bearings Australia in my case), then I made some temporary races in brass, assembled the clock and got good results.

Fig. 3 The bearings as they were

Fig. 3 The bearings as they were

Fig. 4 The new contacts and jig

The next step was to make the hardened steel outer races to replace the prototype brass ones. This, strangely enough was when it all started to go wrong again. The dimension given in the magazine article for the inner diameter of the outer race was 11/16” which translates to about 17.2mm. I had made the prototypes to this size. Somehow when making the steel versions, I just skimmed a little more off the diameter than intended, and the inner diameter ended up at 17.6mm. I thought to myself ‘that can’t make too much difference’.  I case hardened the races and assembled the clock and found of course that the balance sat a bit lower than it should have, but again it didn’t seem enough to make a difference. In operation however, I found that the balance would swing for a while then the balance pivots would force the balls apart and the balance would drop down and jam on the cross piece below the balance (which is part of the magnetic circuit). At this point I started to feel despondent and put the clock aside for about 12 months until fortuitously, another Eureka came in for repair.

I now took the opportunity to run a micrometer all over this unmolested Eureka and came back to that 0.4mm difference in diameter of the outer races. I couldn’t believe that this would be a problem, but I had run out of ideas and decided to make a new pair, paying attention to getting the size exactly right. With the new races turned up and case hardened, the clock was again assembled and at last ran reliably without the balance dropping down between the balls.

At this point, you’d think I’d be getting close, but there was still some time to go. I still had poor amplitude, and so turned my attention to the winding of the coil. The insulation on the outer layer was constantly flaking off, so I wondered whether there may be the odd shorted turn. Only one remedy – rewind the coil. The appropriate wire gauge was purchased and the coil rewound. Unfortunately this made no difference. Again, a little despondent I took the cowards way out and increased the voltage from 1.5v to 3.0 volts, and achieved reasonable amplitude, convincing myself that somehow this was OK.

I then let the clock run for a few weeks to ensure that the old “drop between the balls” problem didn’t recur, and the amplitude stayed reasonable. After that time, I was confident that the clock was now reliable, had reasonable amplitude, but was gaining about 15 minutes a week.  Now, attention was paid to regulating the clock. Eureka clocks have a conventional regulator with movable curb pins applied to the hairspring. Using this regulator seemed to make little difference, so I used the balance screws to regulate the clock. I could get the rate close to reasonable, but it was still a little erratic. So again, I ran it for a few weeks, still not quite satisfied, but convincing myself (yet again) that this was normal, because I have heard that Eureka clocks have a reputation for not being particularly good timekeepers.

And now, for the final piece of the puzzle.

I had the clock running in my TV room, for want of a better place to put it. It was late at night, the TV was off and I was reading a book. I slowly became aware of a regular clicking noise, clearly coming from the Eureka, and a click which shouldn’t occur at the time in the swing of the balance that I could hear it. I got out a torch and put my strong glasses on and carefully peered at the clock, and quickly found the source of the click. As mentioned before, the balance wheel bearing consists of two balls which run in an outer race, with the balance arbor sitting in the ‘V’ formed between the balls. To stop the balls rolling around to the top of the bearing (say when you’re moving the clock), there are two pins, one above each ball, to keep them in position. The click came from the balls hitting these pins. After hitting the pin, the residual amplitude of the balance was happening with the balance arbor sliding across the balls, rather than rolling the balls around. For some reason, the amplitude had increased since the clock was last assembled and was being limited by these bearing locating pins. I can’t explain why, but I was now able to reduce the voltage back down to 1.5v, yielding amplitude of close enough to 360 degrees; just small enough that the pins weren’t bumped by the balls at the peak of the swing. With the amplitude now limited by ‘natural causes’ rather than a frictional limit of the limit pins, the capacity of the regulator built into the clock to set the rate returned, and within a couple of days, I had the rate within 10-20 seconds a day, and reasonably stable.

I can’t explain why the amplitude increased over several weeks (or looking at it differently, the required running voltage reduced), but it did happen. I then made a voltage regulator to reduce the voltage from a surplus Nokia phone charger to 1.5v, with a switch and a battery holder to give the owner the option of running the clock from a ‘D’ cell.

So the moral of the story if you’re bringing a heavily modified back to original, the key things to remember are:

  • Persist
  • Make sure you keep to the exact dimensions of parts you have to make
  • Persist
  • Be patient
  • Make sure your bearings aren’t bumping up against the limit pins in the bearing assembly
  • If you’re doing it for someone else, make sure they’re patient as well.

Fig 6. The new contacts installed

Fig. 7 The new bearings installed


CALL NOW – 0433 830 058