Timegrapher setup for PC running Linux (EndeveourOS/Arch)

[praezis]

On 3/16/2023 at 10:20 PM, JohnR725 said:

I wonder if you can change the averaging time of the software? If you look at timing specifications for watches they usually specify 20 seconds averaging time-based are you having a witschi timing machine. So maybe the software's averaging over a much shorter time span and mechanical watches unfortunately very so you're going to see that variation.

John was on the right track!

With PCTM there comes a file named ReadMe. If you do what it asks you, you will find detailed advice how to calibrate with inhibition-disturbed quartz watches.

Frank

[xyzzy]

21 hours ago, JohnR725 said:

Okay as I have the older version where I get the new version for the PC?

It was in the first paragraph of my long post just a few posts earlier in this thread, "I suggest getting my version of tg from github, https://github.com/xyzzy42/tg " and so on.  As I said then, I do not have a system for making Windows installer packages, so Windows users will have to build it themselves unless someone else can do it.

21 hours ago, JohnR725 said:

Normally the zoom of 2 looks like what I would think it should look on well in a timing machine. Then I went to 8 and then 16 and trace still looks pretty good.

This is interesting, but what is zoom "8"?  8 what?  Perhaps zoom 1 means the width of the strip is one beat?  In that case, the default tg zoom is 10.  My version can smoothly zoom, not just in jumps.  I haven't seen anyone else able to do that.  Maybe it is not a good idea if no one else does it?  At least, I like it.

21 hours ago, JohnR725 said:

I did run a time plot I'm not sure if we can really see every  15 seconds but you'll see there's the minor issues. But in real life stuff like this will average out over time. They just look interesting on the time plot. Then the numbers are all really close or not seeing dramatic differences.

The last graph?  Yes, it looks like a cycle of 45 seconds.

I noticed this single glitch in the Witschi plot,

So I wonder if there is something there?  Perhaps the similar glitches in tg are caused by this too?  I wish I could go back and look in detail at the sound and see why it thought this was the correct location.

Can you post a recording of the sound, like 90 seconds long, I could determine the levels and see what is causing the individual glitches.

I suspect what you see is the result of a few things:

• The gain might be too high, resulting in clipping during the loudest part of the lock.
• Timing based on the lock rather than unlock adds extra "jitter" to the paperstrip.  This does not effect s/d and amplitude measurements, but the paperstrip has more noise.  You do need to zoom in to even see this noise, but then we get to the next cause.
• Tg is by default more zoomed in than a Weishi 1000.  It is probably like zoom 10 on software that does zoom "x" factors.

Edited March 22, 2023 by xyzzy

[eincognito]

FYI, I tried to compile Windows via the instructions on the github...needed to also install the gcc compiler before ./configure completes. However, make yields errors (something about integer). It does output an .exe, but can't run it in the output directory (missing .dlls). I tried copying all files in the output dir to the 0.0.5 windows install dir (which contains all the missing .dlls), but it gets an error and won't start. Haven't tried to progress further after that.