Weishi No 1900 Multifunction Timegrapher - teardown

[Thomas]

Just got this from China through eBay. It is for analyzing the performance of mechanical watches. It does this by listening to the sound of the movement and then interpreting the signal.
Paid $177 + $15 shipping for it. Quite impressed with the design and build quality on this thing. Even the firmware seems good
Anyway, here it is:

The watch is clamped to the microphone assembly. It can be rotated in two axes for testing the watch in different positions. 5 or 6 positions is normal.
This watch runs 18.2 seconds per day fast


Unexciting back of the unit. The USB calibration port is for factory use. If anyone knows anything about this, please let us know


Inside the unit, slightly more interesting. Notice the threaded metal inserts for the screws, also for the PCB screws. Not bad.


Microphone assembly with preamplifier. The red wire is ground, black is +8V Yellow is signal.


A closer look at the microphone assembly. Seems adequate.


Front of the PCB. The display is colour LCD, 480x272 pixels. Nice and bright.


Back of the PCB. Modern design compared to other watch analyzers I have seen. Good quality as far as I can see.


Closeup of the microphone preamplifier.


Closeup of the microphone itself. Seems to be a piezo element, fixed to the metal thing that sticks out of the microphone assembly. Curled wire for mechanical isolation I suppose.

[glarsson]

But the kerning on the front panel!   

MUL T I FUNCT I ON T I MEGRAPHER

[Thomas]

Yeah, monospace font FTW

[ruffy91]

And the name Weishi, just to remember you that the original is Witschi.
I did my apprenticeship at them.

The microphone head is not copied 1:1 mechanically but does look absolutely the same.

[Thomas]

I know about Witschi, their products look amazing. But they are just not accessible to the general consumer.
Your apprenticeship must have been interesting.
And I agree, the Weishi microphone does look very similar to the Witschi one. That said, I think it is easy to distinguish the two manufacturers from each other.
The Weishi name is laughable in my opinion.

[s_premkumar999]

Hi Thomas,
Is it possible to get another image of the microphone section in order to get more idea how the piezo is connected to the mechanical structure and also to view the mechanical assembly shape.

[ralphrmartin]

The Weishi name is laughable in my opinion.

It's actually a sensible Chinese name. I once had a postdoc working for me called Weishi.

[amyk]

I bet the CPLD is being used as the LCD controller, along with the RAM beside it.

The main CPU is a 8051.

[Mr. Scram]

I bet the CPLD is being used as the LCD controller, along with the RAM beside it.

The main CPU is a 8051.

Why would you use a CPLD for that?

[Thomas]

Hi Thomas,
Is it possible to get another image of the microphone section in order to get more idea how the piezo is connected to the mechanical structure and also to view the mechanical assembly shape.

Hi, sorry about the late answer. Hopefully this will make things clearer:



The vibrations from the watch will be picked up by the metal fork and transferred to the piezo element on the lower part of the fork. I think it's made from a single piece of metal.
The fork is grounded with a wire soldered in the middle of it, visible on the picture.
Notice the fork is mechanically isolated from the rest of the assembly, with silicone tubing. This will probably increase the sensitivity of the microphone assembly.

[PointyOintment]

I bet the CPLD is being used as the LCD controller, along with the RAM beside it.

The main CPU is a 8051.

Why would you use a CPLD for that?

I don't know, but it's clearly connected to the LCD's FPC. I can't tell which chip the analog circuitry eventually feeds into.

[LarsG]

Thanks Thomas for a great tear down and pictures!

I am planning to buy only that Weishi microphone on Aliexpress and use it connected to the soundcard of a computer via a cable adapter box with a built in 9V battery for powering the microphone amp. I will run Watch-O-Scope and Tg0.5.0 on the computer.

For that reason I would like to know which pin # on the Timegrapher microphone connector is the +8V out, which is ground and which is Signal?
I just want to doublecheck, should the microphone manufacturer not be using red for ground! and black for +8V!, as in the microphone you received

[cdev]

Can you tell if the crystal is a plain crystal or a TCXO? Its too small for me to tell in your pictures. I think because its marked "X0" its likely a SMT quartz crystal which likely has a few ppm error. Your accuracy would likely be highest if the watch was warmed up to its temperature on the wrist.

As far as peizo elements, they are easy to find. Google "peizo disk" - quartz elements are inside them. They are easy to find both encased and raw.

They are often used as tiny speakers in musical greeting cards. I have used them as pickups for home-made musical instruments.

You could also use a frequency counter or a function generator and scope to find out the oscillation frequency and the resultant error at a given temperature.

[Thomas]

Thanks Thomas for a great tear down and pictures!

I am planning to buy only that Weishi microphone on Aliexpress and use it connected to the soundcard of a computer via a cable adapter box with a built in 9V battery for powering the microphone amp. I will run Watch-O-Scope and Tg0.5.0 on the computer.

For that reason I would like to know which pin # on the Timegrapher microphone connector is the +8V out, which is ground and which is Signal?
I just want to doublecheck, should the microphone manufacturer not be using red for ground! and black for +8V!, as in the microphone you received

Hi Lars
The pinout of the connector on the instrument is attached.

I think the male connector on the instrument is this one:
https://www.elfadistrelec.no/no/panelplugg-forniklet-3p-tsay-m001/p/30020640

And the female on the microphone cable is probably this one:
https://www.elfadistrelec.no/no/hunnkabelkontakt-forniklet-3p-tsay-m002/p/30020264

Well, except for the threads, there are none on the Weishi.

Hope that helps!

[LarsG]

Hi Thomas,
that certainly helps!
Many thanks!

[cdev]

For that much money they should have used a DS3231 or similar as their reference instead of a plain quartz crystal.

[Thomas]

Yes they could have used a more accurate reference. But I don't think it's important. If you had a 100% accurate watch analyzer and adjusted your watch to 0.0s/d it would still be inaccurate when worn.

I start by analyzing the rate error of the watch with a smartphone app (WatchCheck, synced to GPS). This is done over a week or two, with the watch in normal use. The result is an in-use rate error including different use-patterns for weekends etc.
Then I put it in the analyzer and wait for the watch to stabilize from the temperature change. When it is stable, I correct the rate error from the app.
I usually end up around +6.0s/d on the analyzer. This is slightly different between watches, and I expect it to vary quite a bit between analyzers.

[cdev]

My current watch (a an old Suunto Vector) does much better than most watches, it gains maybe one or two seconds a month. If that. But these days I almost never wear it because I almost never get to go hiking. Around the house, what's the point of wearing a bulky watch?

The highest 'mountains' around here measure maybe 500 meters. If that. So the altimeter, etc. isn't so useful now either. The barometer is useful for the weather aspect. But, to its credit, even though its at least 20 yrs old now, the Suunto is still an accurate timepiece. It also appears to be temperature compensated. The inaccuracy remains pretty much the same whether its warm. (mostly on my wrist) or off (sitting in a drawer) It is waterproof too so if I want to, I can shower, swim, etc. with it on. It's never had a single problem.

One thing that I have been meaning to do is attempt to sniff the frequency either via inductance or sound. Then I could get a more exact measure of how off it is.

My gut impression however is that its unusually accurate for a 'sport' type watch. I don't think the newer computer bearing watches could be anything near as accurate were they not able to set themselves from the network.

[Thomas]

But that's a quartz watch, right?
The Weishi 1900 is suitable for adjusting mechanical watches only, and totally unsuitable for quartz watches.
I think you should look for a quartz watch analyzer.

[cdev]

Isn't quartz 'mechanical' ? (being a crystal) It is a moving part.

But I could see how the two would be quite different from the point of view of analysis.

The process might be best accomplished via a coil pickup. It depends.

35 khz is not audible to people but it should be audible to a piezo contact mike. If no mike can hear it (perhaps because its plastic, like mine, and the plastic is lossy) a coil should be able to pick it up via induction. Thats what I was planning to do.

A metal watch might conduct vibration better but a metal case would likely prevent reading the frequency of an oscillator via induction. Some quartz watches might well resist both methods.

[Thomas]

A quartz watch is generally not considered "mechanical" even though the quartz has (microscopic) movement. If it were then all watches would be mechanical
The Weishi will reject any frequency above 12Hz (43200 beats per hour).
Mechanical watches typically beat at 5 or 6Hz.

You probably want this:
https://www.hswalsh.com/product/witschi-new-tech-handy-ii-quartz-watch-testing-machine-ht70
But I'm pretty sure it's out of price range.

[Miti]

I just found this post. What am I missing? There are no pictures in this teardown. 

[Miti]

I think I know what happened. the pictures are posted on Flickr and I think our IT at work blocked access to Flickr. I see them on my home computer.