Seiko 6105 restauration

This is an article I wrote last year about a restauration of my Seiko 6105 diver watch. Unfortunately my account in the forum I wrote it for got locked over a trifling issue (although the forum in question prides itself on non-oppressive moderation), so I’m having my article again. It is a bit rough round the edges, being re-posted here verbatim, at least until I find time to give it a once-over.

I have reported a couple of weeks ago that I finally got hold of a Seiko 6105. Just what I wanted, all original parts, but with problems that I had to fix. I have fixed the most severe now and she’s in working order. So here’s a bit of a report and some pictures to look at.

I won’t assume you all saw my initial post, so a reminder, this is how I got it:

There is the black slime on the lume particular to Seikos from this period. Weird stuff, seems to grow out and spread across the neighbouring regions. It is corrosive as well, there are marks on the stainless steel hands beneath it. It seems to do this to the chromed indices as well, but to a lesser extent. The bezel insert is a bit scratched, so is the case, but they are in surprisingly good nick when you compare them to the crystal, which is terribly scratched. You can also see that the crystal looks a bit milky – this I found out to be dirt on the inside of the crystal, probably caused by some compound inside decaying and gassing out the stuff that then settled on the crystal. I suspect the lume. But the biggest problem was that it was a non-runner. I was not deceived, I bought it on the SC, the seller let me know exactly what I was buying, and that was exactly and precisely what I got (thanks again David). It was working, but the timing was way off and it kept stopping. So:

Just to see how bad it is:

Well, ra-ther. Both the timing and the beat were horribly out of whack. Well, that was the least of its problems for now, as it was about to be serviced anyway. By me. So on to the disassembly, here’s the date wheel off:

What strikes me with Seikos of this vintage is how well made they look. I don’t have any experience to speak of, but the hour wheel is very nicely machined, there is no friction washer on it – seems like they were so confident in their machining that they thought it unnecessary. Indeed, there is hardly any play in it. Towards seven-ish from the centre you can see the shock protection (diashock), and next to it is a big gaping hole in the base plate. I only figured out what that is for while looking through pictures for this post. It’s to open the regulator pins so you can separate the balance from the balance cock. I did that the hard way. Between the hour wheel and the Diashock there is the reduction gearing and the wheel for the date change mechanism. That one struck me as a bit needlessly complicated. On the wheel there is a spring-loaded finger, the other end of which runs on the eccentrically mounted cam/washer/screw thing:

The purpose of that is to allow the finger to extend when it needs to mesh with the date wheel teeth, and to retract when pointing toward the centre. Which is nice – except that I can’t see what it would interfere with if it did not retract. They were probably just too smart to make it simple.

With that out of the way (I like to remove the vulnerable bits on the dial side first so I don’t have to worry about scratching something vulnerable and visible while putting it in the movement holder) I could start on the time bits.

The balance is out already. That is the next vulnerable bit that I absolutely want to have out of the way as soon as possible. Then I had a look around it. It was quite pleasing. No scratches or marks, bit of wear on the corner of the plates where the rotor would strike them when you bump it. Predictable dirt. But underneath all that a nice healthy looking movement. I could only hope it would look like that all the way through.

It did. I would not be the first to eulogise over Seiko’s magic finger auto winding system, but you can’t pass on it either. Just look at that auto bridge – that’s all there is to it. An eccentric pin moving the magic finger pawls around, the wheel they’re pulling on, the gear on that one rotates the winding wheel – job done. Effectively 3 simple parts resulting in a bi-directional highly effective winding system – that’s pure genius. Now I’m not sure if they actually invented that. IWC uses a “pellaton” winding system quite similar to this, except that it has a heart-shaped cam in the centre instead of the eccentric pin. I don’t know who came up with it first, but I dare say Seiko’s is better – simpler, less surfaces to lubricate, less wear, less to go wrong. IWC’s should be a bit more efficient, irrelevantly.

Below the auto bridge there is a bit of predictable dirt. It either was severely overoiled at some point, or the lubricants from the auto bridge have seeped down, you can see an oil stain surrounding the central jewel. The mainspring barrel pivot jewel is ruby, but it looks black because it is so dirty. Interestingly, despite jewels being generously strewn over the movement, the reduction wheel pivot is metal. Wonder what that’s all about.

Central plate off. Still looking good.

The central second and reduction wheel out. The gold coloured bit is the wonderfully simple hacking lever. When you pull out the crown to set the time, the stem barrel drops down onto the gear below so you can move the hands around, at the same time the lever is pulled along, contacts and stops the balance. Watch hacked.

Not much of interest beyond that. So just the final shot: the broken watch.

Of course, there’s also the most important bit – the balance assembly. I have identified the reason for the watch stopping in the hairspring being dirty. There was a big drop of liquid between the regulator pins, obviously on the hairspring as well. Before I disassembled it I had it for quite some time on the timing machine, trying to provoke it to stop again, different positions, shaking, little shocks. When I finally managed it, there was a faint ringing noise as the amplitude fell rapidly to a standstill, the indication that the hairspring is touching or catching on something. It was stuck on itself, two of the coils on the extremely tightly wound spring were held together by a tiny droplet of liquid. Usually I’d prefer not to separate the balance from the balance cock, it is incredibly fiddly and touchy work, screw that up and you’re in a world of hurt. However it was needed. I rinsed the balance several times in some benzene or naphtha (lighter fuel), then in alcohol. I’m not sure I managed to clean it as thoroughly as I should, after I rebuilt it it stopped once again. It was suggested to me that the hairspring may also be magnetised, so that is the working theory for now. Before I attack it with more aggressive solvents I’ll give that a shot. At the moment however it is working, so the problem might have gone away.

Putting it back together was quite uneventful. Straightforward thorough cleaning, inspection, oiling, more checks… I seem to be getting better at that, I’m glad to say, I managed to clean and oil the shock protection in one go, without dropping or losing anything once. I seem to be getting sloppier as well, I put a bit more oil in the pivots than I intended, but it should not be too much.

This is getting a bit longer than I expected, so I’ll split it up now. I’ll continue in the followup to this, stay tuned if you’re still awake.

Next installment. So I had the wiggly and rotatey bits all together and working, next up was the dial and hands. This may not please those who pursue utter originality, but I had intended from the start to make it up to date as necessary. Had the lume been perfect I might have even let it stay, but it was decaying and fouling up everything. It was also the older stuff, not as bright as the lume used today. Since I’m not the custodian of a museum and I intend to wear this watch just like any other, I relumed it.

Here is the hour hand with the black slime, and the dial with the lume partly removed. Everywhere underneath the black parts there was a bit of pitting, luckily not too much to polish out. I had to learn as i was going along. I used a blade to remove the lume from the hands, and I fashioned a kind of lancet out of brass wire hammered flat, assuming I would be able to scratch the lume out without damaging the metal. Which was almost right. The lume was harder and the metal softer than I imagined. I slipped a couple of times, and while I did not scratch anything off, I managed to leave a bit of a mark. I decided I need to go more slowly and carefully, which did not really work. Just made it terribly slow, and I found I was still marking the indices. So I had to try something else, and found out that if I dipped a cocktail stick cut to chisel form into acetone I could carefully work enough of it onto the lume to soften it up without getting acetone on the dial and damaging the paint on it. That worked far better and quicker. I did not want to rush anything, by as I was doing the last one I scratched all the lume out as a single piece, and it took me just a couple of minutes.

The dial was original, here you can see the back of it. Interesting construction. The shape of the indices is stamped into sheet brass, they are then ground flat and chromed. The date window is an extra piece glued into place. The two hollows are ground out bits used to widen the feet of the “Seiko” on the front and hold them in place. The Seiko itself is stamped out of thin metal, very delicate. Someone had worked on the dial previously, wiped the centre of it with some paper presumably, so there was a shiny rectangle in the centre, and the Seiko was slightly bent out of shape. I had corrected that as well, extremely carefully.

So here are the end results. Looks fairly unspectacular, but that is the work of several days. Especially the hands are incredibly difficult to do. You are supposed to scratch stuff off and polish pitting and scratches out, but they’re so delicate that you hardly dare breathe on them. Especially the bevels on the side are a complete ba*tard, you can hardly hold the hands in the correct angle to polish them at all. Can’t use tweezers since that would bend them immediately. Can’t use pressure or you’ll bend them. I can understand why people usually just chuck them away and get some new ones. But I had set my mind on using the original ones, and I’m stubborn. Came out ok in the end.

Here is the dial cleaned and prepared to take the lume. You will note that I painted the surface to be lumed white, which what is usually done. I was surprised to find that Seiko had not done this. The index pans were painted black just like the rest of the dial, with the lume directly on that.

First layer of lume in indices 5-8. You will note I went over a bit. I was not bothered since I knew I can remove the excess later, the important thing is to get it as evenly and regularly as possible. Did not exactly manage that as well, but this is a good dial to practice on. Lots of space, you can put a lot in, and you can correct later.

The problem is of course that I can’t lume. I have done it before, but not with superluminova – I had to learn that as I was going along as well. When you mix it as recommended it is just lovely to work with, but it dries up fast. It forms a blob of dry lume on the oiler, it dries in the mixing pan as well (from the edges first), you get lumps everywhere, hard and runny lume mixed together… It just needs a bit of practice of course. The trick is to go rather quickly (which you can) whilst constantly managing the prepared lume, pulling it from the edges to the centre and mixing a bit as well. Also important is to accept that at some point it is too hard to work with. Most problems I had were due to applying useless lume in order to prevent waste, the stuff is bloody expensive. Toward the end though I quite got the hang of it. You apply all you can reasonably do, then mix some fresh lume on top of the rest, add some thinner to get the right consistency. I would waste less and do a better job if I were to do it now, but of course it’s too late for that now :).
<img src="
The hands. The trick with them is to take enough lume to fill them in one go. If you have too little, it wil just collapse and ooze out of there. Did that once. Instead of luming the watch hand, I lumed my hand. At the next go I got my largest screwdriver and picked up what I thought was a ridiculously big amount of lume, but it was just enough. Hands take a lot of it.

Some premature lume-shots (blurry, I know – my camera is rather crap at this, and I was busy as I was taking them). I quite like the second one, it sums it up nicely. From top to bottom you have the lume powder container, the mix bowl with traces of picking up the lume, the tiny dot is the tip of the oiler used to apply the lume with a little bit still on it, and the dial itself. The last one is the initial comparison with my spork. 6105 dial is brighter. Mind you, that’s just the first layer of lume on. And the spork itself is a veritable torch – wipes the floor with my mate’s 007 without a problem – he had to break out the sumo to beat that one. Just.

Full set of indices done, still only the first layer though. Lights on / ligths off shots, I took them with a view of making animated gifs, haven’t come around to doing it though.

To illustrate that this was not as smooth going as it perhaps seems; here I have screwed something up and needed a bit of fresh lume to smooth it over, presto. Wiped the oiler, wiped the bowl out, wiped the stirring rod – and snapped the end off in the hurry. Doing this kind of thing makes you figure out stuff you never imagined. I needed a stirring rod, so I broke out the bunsen burner and melted the end of the larger bit to a round shape. Never thought I’d want to melt and shape glass. Works better than the original, though :D.

So, now we’re done here. As I was finally assembling it I was too impatient to make any photos, I just wanted to wear it! Not much to see anyway, just putting the bits together. Now, I wanted some shots to absolutely awe everybody and make you ooooh and aaaaah about my fantastic looking 6105. Well… I spent yesterday evening trying to make such a shot. You see, the problem is that the scratched crystal is still on. Doesn’t look that bad really, but when you take pictures of it, you get almost nothing but the scratches. What I ended up was a 6105 I spent about 2 weeks restoring and it still looks like shit :). I will be changing the crystal once I get hold of a correct one, but for now, I can only let you see my wonderful crap looking watch. This is the best I came up with. Enjoy 🙂

To conclude, finally (groan), it has been a success so far. The lume is not original, but I have not broken anything, it works quite well (about -5s a day). In fact, since I deliberately put an absolutely ludicrously ridiculous amount of lume on it, it is the brightest glowing thing I have ever seen. It outshines my mate’s Sumo now without breaking a sweat. Yes, I know that it’s not correct for this vintage of watch, but it makes it even more special. Still a bit to do on it, the crystal of course, and I need to work up some courage to pry off the bezel (rotates nicely but won’t let go), but no hurry there, I’m quite happy with it like it is.


Now my 6105 is finally complete. One of the responses to this thread was by Derek (LuvWatch), kindly offering me his unused 6105 replacement crystal for free if I wanted it. He declined my offer to at least cover his cost or even pay for shipping – a gesture so magnanimous that I feel quite ashamed by it, knowing that it’s something I myself would not think of doing. Thanks again, mate.

So today it arrived, with a nice note stuck onto it:

The last shot of my 6105 complete with the scratched crystal, and the replacement next to it. No denying it was needed:

On Derek’s Seiko blog there is an article on how to replace the lost click ball and spring on a 6309 (…your-6309-7040) – along with a useful tip how not to lose it in the first place – pop the bezel off inside a plastic freezer bag, so that is what I did:

I have tried to get the bezel off for cleaning before and could not get it off as easily as I expected, so I let it be until the time I’d be changing the crystal. To allow it to come off more easily I sprayed it liberally with WD40. Unsurprisingly, there’s a lot of dirt in there. The click spring was 10% spring and 90% dirt:

I had expected the crystal to be press-fitted into the case, instead it has a rubber gasket and a crystal retainer bezel. Even more dirt in here – and something worse: rust. Uh-oh.

Oh well. That’s no big problem as long it’s only here, this won’t affect the fit of the crystal gasket. Unless if it is in the gasket seat as well!. Damn.

Now the package from Derek was not the only thing that arrived today. I have also taken delivery of an USB microscope, so immediatelly I put it to good use:

That clearly would not do. I had hoped for a quick replacement, but now I had to clean that mess out and ensure a seal again, and also to make sure the corrosion won’t eat any further into the metal. Most of the discolouration was just rust particles washed under the seal (which did not seal much thanks to that), it was easily rubbed off, but there was also a fair bit of pitting. I busied myself with a glas fibre brush and got it almost entirely down to blank metal:

Now that looks much better than before, but I don’t trust it to seal. So I helped with a tiny bit of silicone bathroom sealant under the gasket, just to fill the voids and protect it from rusting again. There is a special compound used to glue and seal plastic crystals in place which I’d have preferred, but I did not have it, so bathroom sealant it is. The crystal fit nicely into place, the retainer bezel also. I was a bit busy doing it so I forgot to take pictures in between:

I don’t know if you can see it on the pictures, but it made a hell of a lot of difference, so after seeing this I was shaking with anticipation, I wanted to see how it looked once complete. But I still had to lume the bezel pip – I don’t want to take the bezel off any more often than absolutely necessary. I had vaguely planned to just get a new standard bezel pip and glue it in place, however one look showed that there is hardly place enough for that, I’d have to lume it. That was lousy planning, I should have known I’d get impatient to see the final result – had I done it first it would have time to dry as well. As it is, I just waited for it to dry enough not to run out, put some backing paint on it, applied a liberal quantity of silicone grease and slapped it on.

By the way, I decided against using SuperLuminova luminous powder here. There is enough room to accommodate larger crystals (which glow better) so I used luminous powder from and SuperLuminova varnish.

Then I was set. After the quickest movement installation I had ever done… ta-daaaaaa:

Not out of focus, just focussing on the reflection of my Areca in the crystal:

That was quite fun, so I set up one shot with more reflections, you can see the webcam at 6, my head with the loupe on the forehead at 9, the Areca above and a bit of the monitor at 11-ish:

Light on/off to show off the lume:

…and a couple more with lume and reflections (using the bag-o-lume as a sandbag):

Now I have to qualify something: at the time I wrote the initial post, my service of the movement was not actually successful as I thought. I wrote it right after I finished the lume work and assembled everything, at that point the movement was complete and running in the movement holder for about a week. I had no reason to assume it would stop again (as it used to when I got it), but about two days thereafter I noticed that the watch had suddenly lost about 10 seconds. I reset the time hoping that was something out of the ordinary, it was fine again for a couple of days and then again lost a bit more. No mistake, it was stopping again.

The first time I was servicing it there was oil on the hairspring. I cleaned that off but was not sure I had done it thoroughly enough. I got some benzene and some isopropanol, took the balance out and rinsed it several times in both, then pushed the coils around to check if they will stick the least bit upon contact – everything was fine. That’s it then, it’s fixed, thought I and put it back together. It was fine for a while, then lost about 15 minutes overnight.

I took it apart again, inspected all the wheels again, cleaned everything, lubricated once more. It was fine a day or so, then it stopped again. I decided I needed to improve the amplitude, which at about 200 was slightly below that of my SRP043 and that of the 6138. I spent a long time checking everything up again, and finally found that the mainspring barrel is a bit worn and at times can touch the second wheel. I have no replacement and I have no skill or tools to tighten it, so I decided to rather keep the second wheel (which has a quite large endshake) from slipping low enough to touch the mainspring barrel. It took me several hours to make a brass washer just as big as the jewel diametre and thin enough to leave a minimum of endshake. I then took everything apart again, cleaned, inspected, lubricated… The amplitude rose to well over 220°, peaking at about 250°. That was quite reassuring, I was pretty confident that it won’t stop any more, that’s it, sorted. That same night, it stopped again.

I was beginning to despair. The maddening thing about this was that I could not observe what the hell was the matter with it. The watch did not just stop, it would invariably start running all by itself after a couple of seconds or minutes (record: 20 minutes). I had it on the timing machine for hours, trying to find a position in which it would stop, it would cheerfully just keep on running without even dropping in amplitude. There was no pattern either. It did not stop at some specific time of day, indicating something wrong with the hour or minute wheel or the date change, or anything from the ancilliaries, it was all entirely random. It would run for days on end and then stop twice within an hour. Only hope was to keep wearing it and catch it when it stops.

Last week I managed it. I had gone to bed a bit late, just as I was about to fall asleep I checked the time and noticed the second hand was not moving. Held it to my ear, tilted it this way and that, and in certain positions it would make a rubbing noise and stop. Jumped out of bed, took the movement out and half an hour later I had it: the impulse jewel carrier was contacting the anchor. The balance endshake was too large, either because the shock protection was not sitting properly in the socket, or the whole socked was pushed a bit out of the plate. It had to wait till tomorrow, so I went to bed. Of course could not fall asleep thinking about how to fix it. And why the devil this happened only now and not every time, or most of the times the watch would be at this angle.

The next day, after finding that it was the shock protection socket knocked out of place and after unsuccessfully trying to make a tool to press it into place I decided to try gently knocking it to where it belonged. Overdid it at first, then spent hours knocking it back by perhaps a thousanth – every time checking by doing one of the scariest operations for me, installing and taking out the balance, until I finally got it: no noticeable endshake with balance moving completely freely. Of course, it has been running since, at consistent -5 sec a day. I have no doubt that it is fixed now.

Here also the saga ends about the most elusive, expensive, valuable, troublesome, annoying, magnificent watch I own – thanks for reading, and above all for your support 🙂

1 comment

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: