You can skip this whole background story by clicking here.

I have always thought AGS is a scam. The latest iteration of that is Solar. Whether the generation system is by kinetic generator, or by photoelectric cell, they both rely on a rechargeable battery that gradually loses its ability to keep charge, and eventually the watch stops running.

Moreover, the owner needs to keep the battery topped up all the time, otherwise the battery overall lifespan will be shortened. And while the watch is kept running, whether the owner wears it or not, its gear will continue to wear.

A simple quartz watch using ordinary battery is way more compact without the charging mechanism and easier to maintain as I can simply remove the battery and keep it in a drawer for 10 years. Wherewas an AGS or Solar would run completely flat after 10 years, requiring a battery change.

When an AGS eventually runs flat, the owner would throw it out to the second-hand market as non-running, while all it takes is a battery change, maybe a service. They were made to be serviceable, too.

But they are very well made. Rightfully so—these are expensive watches at their times so the design would have to follow with a more premium feel. Other than a few funky ones, most look pretty good. Therefore when I had a chance to fiddle with one I couldn't resist.

Like this 5M23-6B60 I got from a job lot.

Of course it was not running. The exterior simply need a clean. The bezel can be rotated. It even came with the original dedicated lug shape strap. After overcoming the caseback with some extra traction, a clean AGS 5M23 movement was found inside.

I replaced the original 30235MY battery inside with a modern equivalent MT920. The movement showed no sign of life.

My multimeter indicated a healthy stepper motor coil within the 2-4kΩ range. But the charging coil has such a low resistance (300-ish Ω) meaning at 1kΩ scale the needle on my multimeter wouldn't even move.

OK, the generator may be defective. At least, with a healthy coil the watch should have been running. I gave this 5M23 a service, cleaning the wheel and oiling the pivots, yet still, no sign of life.

I put the watch aside, waiting for a donor watch to come up.

I bought an 5M22-6000 for cheap. If you wait long enough a good deal does show up. This one is fully functional when I dropped an MT920 battery in. 5M22 and 5M23 share the same electrical components.

This time I realized the coil on the original 5M23 is actually good. That leaves the only possible defective part the circuit block. Once I replaced the 5M23's circuit block with 5M22's, the watch was alive again.

According to the Internet, MT920 has a much longer battery life than 30235MY per charge. It has been 3 month since I wore the watch, it's still running ok, with battery indicator advancing 20 seconds when activated via the button at 2 o'clock.

See, I bought the 5M22-6000 simply to donate parts to the 5M23-6B60. But once I have the 5M22-6000 on my hand (and wrist), I didn't mind getting yet another donor to make the 5M22-6000 work again.

Hence began another hunt for a reasonably priced donor.

Remember I said they have a more premium feel to their design and most look pretty good?

The 5M23-7A70 I bought as a donor, turned out looking better than the 5M22-6000!

What do I do?

After carefully weighting my options, I decided to restore the 5M23-7A70 first, as its case showed less sign of use.

This watch came in ticking twice every 2 seconds—a sign of dying battery but otherwise a working movement. I quickly sourced another MT920 and in it went following a complete service of the movement. It was ticking happily and everything looked normal.

I was only annoyed by the high pitch noise the intermediate wheel for generating rotor made. The triangular spring on the bottom side of the wheel interacts with 2 protrusions on the mainplate to make a noise and perhaps to regulate the speed the generator rotor can turn so it doesn't produce a voltage (∝ to speed) too big that may damage the rechargeable battery. 5M23-6B60 also makes a noise, but not as high pitch as this one. Oiling didn't help.

Anyway, I decided live with that know I couldn't fix it without maybe swapping out the intermediate wheel itself. I totally could've, as I have an 5M22-6000 lying around as donor, but nah too much work.

A few days later, I realized this 5M23-7A70 was ticking twice every 2 seconds again, sometimes it would only make 1 second leap, causing the watchto run slow. Fair with a new battery, it would only take 30m 1hr 1.5hr indefinite amount of shaking to charge up the battery…

I couldn't charge it. After a lot of swinging, the battery stayed at 1.4V which is below the half way point of the battery's discharge curve. The charging voltage of this battery is 1.8V to 2.6V. Operating voltage is around 1.5V.

Both the coils are within their designed resistance. Coil block again? I couldn't really tell. The driving circuit seems to work alright with the watch ticking. It's just that the battery wasn't charging. Could it have been the battery? I don't want to buy another one just to test though.

At this rate, this watch will become unusable. Do I give up?

I remember having thought about dropping a regular non-rechargeable battery into the 5M23-6B60 earlier. Would anyone have actually tried it before?

Oh yea they did.

SR920SW happens to have a similar dimension (9.5 × 2.0 mm) and nominal discharge voltage (1.55V) as the MT920 (1.5V). While being non-rechargeable, SR920SW actually holds 35mAh versus MT920 5mAh, 7× more capacity. If MT920 at full charge can keep the watch running for 120 days, I expect a biennial battery change with SR920SW, without having to swing the watch every some weeks.

What a no-brainer.

Oh? That is destroying the purpose of having an AGS? But see, I am a person who have PPAP'ed a pair of 3863. I liked the designs of these AGS but never really quite like the idea of AGS itself.

I couldn't have explained better than people who did this modification earlier in terms of instructions.

However please note, depending on the brand, the contact clip can easily short the SR920SW when not placed spot-on when installed. A MAXWELL or MURATA has a rather short negative terminatl, so the part of the contact clip that bends down may touch the positive terminal, instantaneous render the battery unusable. Its voltage would have dropped to 1.4V or as low as 0.25V in my experience.

Use a multimeter to check the voltage at every step of the process to detect shorted battery early and avoid wasted work.

After shorting all 3 MAXWELL and MURATA I have in stock, I switched to RENATA which seemed to fare better, perhaps its negative termianl is a ted bit taller so when I clumsily misplaced the contact clip, the battery didn't appear shorted as the 3 others I tried earlier.

Yes, I was too lazy to glue/tape the contact clip on.

Oh and I forgot to remove the generating rotor so out with the battery again. Without the generating rotor, and oscillating weight, there won't be a voltage surge that would damage the battery (unless, the watch is placed in a strong magnetic field). And no more noise with the intermediate wheel for generating rotor!

I suspect the high defect rate of 5M2x circuit blocks was due to the unstable charging and discharging voltage and current in the AGS mechanism. Removing the mechanism can perhaps do more good than harm.

I was baffled in the first 24 hours when the watch was ticking twice per 2 seconds, and advancing only 1 second sometimes, running slow overall. This seemed to stabalize after either I reinstalled the battery 3 times or simply letting it run for around 8 hours.

After 24 hours, the watch seemed to keeping time well. The same behavior was observed by the forerunners in this ordeal and obviously, nobody knows why.

Anyway, an AGS without AGS is a better AGS.

One day, I may do the same with the 5M23-6B60.

A side note: 5M22-6000 and 5M23-7A70 dial wasn't supposed to be compatible as the first digit of the case code (6/7) indicates a different dial size, but when I placed the 5M23 with its original chapter ring into the 5M22 case, everything seemed to fit. The differentiating part is actually the dial and chapter ring, while the case and bezel are compatible.

And you see 5M22-6000 has a weathered but better decorated case, I may one day swap the movement and dial assembly of 5M23-6B60 into the 5M22-6000 case, PPAP'ing the thing again.