locked Re: Missing decodes #AudioIssues

Tim Dawson

Never used it anyhow, myself . . . if I need to step, I use ntpdate (part of the ntpd package) to step before ntpd starts . . . . (More than one way to skin that cat). For that matter, if ntpd is working correctly and calculating drift factors, unless the system is down for an extended period, there really should be no need for a step change anyhow.

On October 8, 2022 11:55:05 AM CDT, "Michael Black via groups.io" <mdblack98@...> wrote:
One thing I noticed about Meinberg is that they are missing the -G option at startup.  The -G allows an initial step correction where the -g just allows ntpd to keep running when it's off by too much.
It's in this registry keyComputer\HKEY_LOCAL_MACHINE\SYSTEM\ControlSet001\Services\NTP

Should be this"C:\Program Files (x86)\NTP\bin\ntpd.exe" -U 3 -M -G -g -c "C:\Program Files (x86)\NTP\etc\ntp.conf"

I've reported this bug to Meinberg.
Mike W9MDB

On Saturday, October 8, 2022 at 10:45:19 AM CDT, Tim Dawson <tadawson@...> wrote:

One other thing to note (and the original message has been lost) was someone had mentioned an ntp client taking CPU resources away from WSJTX . . . . Not sure where the idea came from that time service is CPU intensive, but generally it is not . . . looking at my primary box (full multi-server full ntpd deployment), uptime is 487 days, and in that period, ntpd has consumed 6 minutes and 49 seconds of CPU time (unless I gaffed my quick math, that's .87 seconds per day, or .096 seconds per hour), . Between that trivial useage, and recent systems typically being multicore, I would not think it possible for time service to have an impact on an app, unless the hardware is decades old and single core, and then itmis still highly unlikely.

On October 7, 2022 11:04:01 PM CDT, Tim Dawson <tadawson@...> wrote:
ntpd (which Meinberg is also a compile of) is designed to compensate for propagation delay. It queries the time server, listens for the reply, and can deduce the transit time pretty well, so that should not be a problem *IF* the Mac is using a proper ntp distribution, and not some cut down sntp kludge.

On October 7, 2022 9:40:16 PM CDT, "Robert Rose via groups.io" <robert.rose@...> wrote:
Just for completeness, I note that I was using using Meinberg on my Mac, and the time differential was 2.1 to 2.3 seconds. I switched to an Apple server in Cupertino (I am in San Diego) and the difference dropped to 0.1 seconds.

I believe the WSJT-X User Guide recommends that you use the closest time server. It takes 0.13 seconds for light to circumnavigate the earth. Add in time for hops and switches, and this sounds like good advice.

On Oct 7, 2022, at 2:46 PM, Phil Cooper via groups.io <pcooper@...> wrote:

Hi to Alan, Brian and Kari,

That is very interesting! I changed the time update period earlier, and that did seem to make a difference. I think I went from 1 minute to 5 minutes, and noticed it didn't drop as much. I will change that to 30 minutes over the weekend, and see whether that improves things.

As for Meinburg vs Dimension4, well, that seems to be one of those things. I know many folk that run D4 happily, and hate Meinburg, and also vice versa, so I will stick with D4 for now.

Thanks for the tips, all much appreciated. I will keep an eye on things, and see what happens.

Very 73


-----Original Message-----
From: main@WSJTX.groups.io [mailto:main@WSJTX.groups.io] On Behalf Of Alan G4ZFQ
Sent: 07 October 2022 17:17
To: main@WSJTX.groups.io
Subject: Re: [WSJTX] Missing decodes #AudioIssues

On 07/10/2022 15:59, Phil Cooper via groups.io wrote:
  is updated every minute. Maybe that is too often?

Use far less than that, maybe several hours or more unless your PC clock
is broken.

In fact the proper NTP of Meinberg is recommended.

73 Alan G4ZFQ

Sent from my Android device with K-9 Mail. Please excuse my brevity.

Sent from my Android device with K-9 Mail. Please excuse my brevity.

Sent from my Android device with K-9 Mail. Please excuse my brevity.

Join main@WSJTX.groups.io to automatically receive all group messages.