Anatomy of a design failure

I serviced this Fortis limited edition watch last year, and it had a failure that I thought you might like to see, as it shows how a small modification can go wrong:





It was sent in with the specific complaint that the date change no longer worked. I had expected this to be an ETA 2824-2, but it ended up being an ETA 2836-2 instead:



It's interesting to see how companies change a basic off the shelf movement to suit their design requirements. Some can involve rearranging wheels inside the movement (Oris for example), or completely changing the way the watch operates (Hublot), and then there are "easy" modification like this one. The watch is fitted with an oversized dial, which required a spacer to be added to the outside diameter of the movement. The red arrow points to the dial screws that hold the large dial in place:



With the dial removed, you can see that the date indicator is also very large:



I give Fortis credit here, because we have all seen that when companies do this larger dial thing, having the date window crowding the middle of the dial isn't a good look, so the design intent here was spot on in my view.

Just moving the movement around to take photos, a portion of the date indicator driving wheel fell off:



So fixing the faulty date change was simply a matter of getting a new date indicator driving wheel:



But I wondered why this happened, and also why they chose this movement. I called this a date indicator driving wheel, but it actually drives both the day and date, because the 2836-2 is a day date movement. So why would they use this movement for a simple date watch instead of a 2824-2?

Well, the answer is the difference between the 2824-2 and the 2836-2 date mechanism. On the 2824-2 it is a semi-instantaneous date change. This means that the date will start to creep forwards in the date window a bit, before completing the flip somewhere around midnight.

The 2836-2 is a truly instantaneous date change, so it flips from one date to the next in the blink of an eye. Unlike the date indicator driving wheel in the 2824-s that just rotates once per day in a constant slow movement, the 2836's wheel does turn for part of the day, but then stops as tension builds up in a spring. There is a cam that eventually allows the spring to release, and it flips the date all in one shot.

In both of these movements, the date indicator is advanced by a small finger that is on the driving wheel. On the 2824-2 the finger is spring loaded, but on the 2836-2 it is not. This is the key to the decisions Fortis made to use this movement.

Here’s a look at the dial side of an ETA 2824-2, also from a Fortis:



The red arrow points to the finger on the date indicator driving wheel. This finger is spring loaded, and the reason this is done is to prevent damage to the date indicator if you accidentally quick set the date when the date is already in the process of changing,

So here is an SW200-1 and I am demonstrating that the finger in the date indicator driving wheel (same design as the 2924-2) can collapse if the date is quick set when it’s in the progress of changing.


Note that I don’t recommend doing this a lot, but if it happens, you won’t get the same kind of damage you might in other watches, where the finger isn’t spring loaded, like on this 7750:



The finger had sheared right off on that one.

So back to our limited edition Fortis. Here is the Fortis date indicator, next to a date indicator for a 2824-2:



Not only is this date indicator larger, but it is substantially thicker (large one on top):







Some quick calculations show that the larger date indicator is approx. 5 times heavier than the type found in a 2824-2. So if you tried to use a date indicator driving wheel with a spring loaded finger for this very large date indicator, it most likely wouldn’t even drive the date change as the finger would just collapse. The spring for the finger wouldn't be strong enough.

So they used the 2836-2 with the design that can’t collapse, but in the end the weight of the date indicator was just too much even for the 2836-2's design. I've never had to replace the date indicator driving wheel on the 2826-2 before, so this is in my experience a unique failure, but I suspect that this won't be a unique experience for those who may own one of these LE's.

If the larger date indicator had been made the same thickness as the original 2824-2 indicator was, we likely wouldn't see this happen.

Cheers, Al

Thanks Al for a good read. I have a couple of Fortis LEs….will have to look at them more closely. At this point no issues requiring revision😗

Very interesting!

Damned Physics and Material Science, always getting in the way!

Great investigation and documentation!

I wonder if the larger date wheel also leads to more friction?

Yes. This can be mitigated somewhat by not having the large date indicator ride on a wide, flat surface, to reduce surface area contact.

Ah, right. I did not think of that. Thank you!

Reminds me of the Bettlach movement I was practicing on in december. Date wheels often seem like they are an afterthought added by the marketing department. Sometime I need to re do that watch as the date wheel friction is too much for the roskopf offset friction drive, so the motion works tend to slip when they get near the date shift.

I have found on a lot of the date watches I have that friction really is the bane of the system. Sometime I need to pull apart the constellation I gave my dad decades ago and see why the date system causes the watch to stop. My dad however would rather have me work on the rolux he got on the Yangsee river or what ever it was before called they flooded it with that big dam. I cleaned it shortly after he got it, was full of cat or some other sort of similar lint. The chairman Mao watch on the other hand worked when we wound it up.

Another great piece, thanks Al.

Interesting, movement info was cool but then I got stuck on that dial. I like weird unique things but I’m having a hard time appreciating that dial.

Another couple of the Sachs pieces, have not needed revisions….

The foggy one appears to have a normal date indicator, and the other has no date, so unlikely either will have this problem. 😉

Does the plate have a ring/ridge for the date wheel to rise on or is it a flat surface to flat surface interface?

I also suspect that the date wheel detent spring may be stronger than normal to overcome the significant momentum of the date wheel flicking over.

It's just flat.


No, the date jumper was just the normal one. Note that it's already fairly sturdy, because it has to stop a rapidly changing date in the normal version of the 2836. I think the weight and friction of this date indicator offsets any inertia issues that might come from the larger mass.

Here's a video of the quick set, and then normal date change...