--

If you are a developer type there are performance counter functions that can give you the actual clock frequency. The dithering approach still applies.

{^_^}

On 20220512 07:47:26, Brian Morrison wrote:

On Wed, 11 May 2022 22:15:50 -0700 "Radioman ZeroZero" <radioman@...> wrote:

A pc clock cannot maintain time accurately.  It is all very well setting a very accurate time but after half an hour it will have drifted.

If you're running something like Fedora Linux then you can use chronyd to supervise the system clocks, it's far better than ntpd on its own. There are also packages like oscillatord which can discipline PC clocks from a 1pps GPSDO signal.

I'm not sure what amateur radio part of the hobby requires a super accurate time.

Accuracy is addictive, once you have it you can find things to do with it ;-)

In this case synchronizing a fast frequency hopping ECCM radio to a network.

{^_-}

On 12/05/2022 12:27, jdow wrote:

Technical extreme quibble:

In order to get approximately 1 kHz clocks a timer, not atypically running off
a US TV receiver color subcarrier crystal, is divided by 3580. This obviously
gives a frequency about 545 ppm off to which the clock's accuracy of maybe as
bad as 100 ppm is added. The best you can do is dither this divider so that on
the average you get 1 kHz with as many 0s after the decimal point as you wish.
If the oscillator is perfectly on 5e6/88*63 Hz you can work out the dither
pattern to give arbitrary accuracy from 3.57954545454545... MHz. In a simple
case with an imprecise computer oscillator, if the clock alternates between
3580 and 3579 for the setting you move the accuracy down to about 46 ppm from
the basic process and whatever the clock might be. If you compare with an
external precision timer you can work out a dither pattern for the divisor that
leads to greater long term accuracy. Unfortunately this sort of bassakwards
approach is required because the timer clock has no software adjustment
potential so you cannot slide the oscillator right in with a D/A converter
feeding a VCXO input.


{^_-}  Did anybody manage to stay awake through that nonsense?

Completely agreed, but I do wonder when the NTSC colour subcarrier frequency
was last used as a master clock in PCs! I have heard of folk replacing the
master oscillator with a TCXO, but that's a little extreme.

David GM8ARV
--
SatSignal Software - Quality software for you
Web: https://www.satsignal.eu
Email: david-taylor@...
Twitter: @gm8arv

On Wed, 11 May 2022 22:15:50 -0700
"Radioman ZeroZero" <radioman@...> wrote:

A pc clock cannot maintain time accurately.  It is all very well
setting a very accurate time but after half an hour it will have
drifted.

If you're running something like Fedora Linux then you can use chronyd
to supervise the system clocks, it's far better than ntpd on its own.

There are also packages like oscillatord which can discipline PC clocks
from a 1pps GPSDO signal.

I'm not sure what amateur radio part of the hobby requires a
super accurate time.

Accuracy is addictive, once you have it you can find things to do with it ;-)

--

Brian G8SEZ

On 12/05/2022 06:15, Radioman ZeroZero wrote:

A pc clock cannot maintain time accurately.  It is all very well setting a very
accurate time but after half an hour it will have drifted.  I'm not sure what
amateur radio part of the hobby requires a super accurate time.
--
Thanks Steve M0ZEH

Steve,

Reference NTP software (but not all so-called "atomic clock" programs) does not
only set the time initially, but it also measures the drift rate and adjusts
the clock rate so that accurate time is maintained even with a drifting
undisciplined clock.

There's more here:  https://www.satsignal.eu/ntp/setup.html#why

73,
David GM8ARV
--
SatSignal Software - Quality software for you
Web: https://www.satsignal.eu
Email: david-taylor@...
Twitter: @gm8arv

Technical extreme quibble:

In order to get approximately 1 kHz clocks a timer, not atypically running off a US TV receiver color subcarrier crystal, is divided by 3580. This obviously gives a frequency about 545 ppm off to which the clock's accuracy of maybe as bad as 100 ppm is added. The best you can do is dither this divider so that on the average you get 1 kHz with as many 0s after the decimal point as you wish. If the oscillator is perfectly on 5e6/88*63 Hz you can work out the dither pattern to give arbitrary accuracy from 3.57954545454545... MHz. In a simple case with an imprecise computer oscillator, if the clock alternates between 3580 and 3579 for the setting you move the accuracy down to about 46 ppm from the basic process and whatever the clock might be. If you compare with an external precision timer you can work out a dither pattern for the divisor that leads to greater long term accuracy. Unfortunately this sort of bassakwards approach is required because the timer clock has no software adjustment potential so you cannot slide the oscillator right in with a D/A converter feeding a VCXO input.

{^_-}  Did anybody manage to stay awake through that nonsense?

On 12/05/2022 06:15, Radioman ZeroZero wrote:

A pc clock cannot maintain time accurately.  It is all very well setting a very
accurate time but after half an hour it will have drifted.  I'm not sure what
amateur radio part of the hobby requires a super accurate time.
--
Thanks Steve M0ZEH

Steve,

Reference NTP software (but not all so-called "atomic clock" programs) does not
only set the time initially, but it also measures the drift rate and adjusts
the clock rate so that accurate time is maintained even with a drifting
undisciplined clock.

There's more here: https://www.satsignal.eu/ntp/setup.html#why

73,
David GM8ARV
--
SatSignal Software - Quality software for you
Web: https://www.satsignal.eu
Email: david-taylor@...
Twitter: @gm8arv

On 20220511 23:56:07, Alan G4ZFQ wrote:

On 12/05/2022 05:15, Radioman ZeroZero wrote:

not sure what amateur radio part of the hobby requires a super accurate time.

Steve,

Super accurate, not many. http://www.dxatlas.com/faros/ is one that measures propagation delay, it actually contains it's own NTP clock.
The Kiwi radio network uses GPS PPS to use TOA station location.

Meinberg as mentioned is one that sets PC time and maintains it  to within miliseconds.

73 Alan G4ZFQ

I've run across a reference of an ntp server that maintains time by fiddling with the counter settings. If you have a 3.57954545454 MHz timer oscillator (5e6/88*63) then it fiddles the time keeping divider up and down from 3579 to account for the fractional part and apply further correction so that long term timing is in the microsecond range if you can supply it with an accurate time tick. The pro radio astronomers use an even fancier setup to make precision time using motherboards that reference time to 100 MHz or higher oscillators in the PCs. (I wonder how many replace the timer oscillator with a TCVCXO and lock the loop. I bet the VLB array uses this sort of technique.)

PCs have a lot more potential than most people realize, especially for a hardware hacker.

{^_^}

As long as the temperature is fairly stable a PC can do remarkably with a normal NTP installation. The Linux PC used as a router here maintains our ntp install and maintains time within +/- 30 ms across multiple time sources. At the moment it has chosen a GPS locked source and is within +/- 6 ms jitter on teh link and an offset of about 0.34 ms. It is +/- 8 ms across the 5 sites it monitors. Modulo your definition of accurately that's surprisingly good considering it does not vary all that much. It applies corrections to the PC's timers so that they tick off accurate 1ms ticks with jitter in the sub-microsecond range that is intentionally used to keep the clock accurate.

On my Windows PC I am running an old ntp tool that does not touch the timers and instead jam sets 1ms plus, zero, or minus corrections as needed once a minute. (This PC's clock is pretty pathetic, too.) I should probably get a proper ntp for the machine. I believe NIST has one.

{^_^}

On 12/05/2022 05:15, Radioman ZeroZero wrote:

not sure what amateur radio part of the hobby requires a super accurate time.

Steve,

Super accurate, not many. http://www.dxatlas.com/faros/ is one that measures propagation delay, it actually contains it's own NTP clock.
The Kiwi radio network uses GPS PPS to use TOA station location.

Meinberg as mentioned is one that sets PC time and maintains it to within miliseconds.

73 Alan G4ZFQ

A pc clock cannot maintain time accurately.  It is all very well setting a very accurate time but after half an hour it will have drifted.  I'm not sure what amateur radio part of the hobby requires a super accurate time.
--
Thanks Steve M0ZEH

Sorry, being busy all the day long.

It is a Idiom Press RotorCard, DCU protocol.

What is the rotator controller you’re talking to?

 

Simon Brown

https://www.g4eli.com

https://www.sdr-radio.com

 

From: main@SDR-Radio.groups.io <main@SDR-Radio.groups.io> On Behalf Of Max Mucci, N5NHJ via groups.io
Sent: 11 May 2022 01:33
To: main@SDR-Radio.groups.io; SDR-Radio@groups.io
Subject: Re: [SDR-Radio] World Map - Rotator COM Ports

 

Logs attached.

 

On 2022-05-10 19:20, Max Mucci, N5NHJ via groups.io wrote:

I'll check it out tonight. I'll also update the VSPE installation (first..) just in case.

Thanks!

73, Max

On 2022-05-10 19:15, Simon Brown wrote:

So,

 

For those who had a problem, please try: https://1drv.ms/u/s!AovWaZDu7HrdgYBDVujIHq2ooWHKKg?e=7AMYA4

 

Simon Brown

https://www.g4eli.com

https://www.sdr-radio.com

 

--

- + - + -

Please use https://forum.sdr-radio.com:4499/ when posting questions or problems.

-- 73 de Max, N5NHJ (I8NHJ) one of EL2PP, 1A0KM, JT1Y, E30GA, T77V, TT8MM, T32R, K5K, XR0X, S05X, 4O3T, BS7H, K4M, PJ7E, PJ7DX, 6Y1V, C6ANA, VP5/I8NHJ, ZF2NH, KP4/N5NHJ and...

-- 73 de Max, N5NHJ (I8NHJ) one of EL2PP, 1A0KM, JT1Y, E30GA, T77V, TT8MM, T32R, K5K, XR0X, S05X, 4O3T, BS7H, K4M, PJ7E, PJ7DX, 6Y1V, C6ANA, VP5/I8NHJ, ZF2NH, KP4/N5NHJ and...

-- 73 de Max, N5NHJ (I8NHJ) one of EL2PP, 1A0KM, JT1Y, E30GA, T77V, TT8MM, T32R, K5K, XR0X, S05X, 4O3T, BS7H, K4M, PJ7E, PJ7DX, 6Y1V, C6ANA, VP5/I8NHJ, ZF2NH, KP4/N5NHJ and...

--

- + - + -

Please use https://forum.sdr-radio.com:4499/ when posting questions or problems.

-- 73 de Max, N5NHJ (I8NHJ) one of EL2PP, 1A0KM, JT1Y, E30GA, T77V, TT8MM, T32R, K5K, XR0X, S05X, 4O3T, BS7H, K4M, PJ7E, PJ7DX, 6Y1V, C6ANA, VP5/I8NHJ, ZF2NH, KP4/N5NHJ and...

Hi

as info and save a few lines of code for Simon.... and free time for dog walking.... ;-) ;-)

 

I always use with USB GPS on the /P location and laptop PC without internet the BktTimeSync PC APP by IZ2BKT.

It works perfectly and directly adjusts the clock and the whole PC with the GPS data on the COM.

It goes either with GPS with internet or/and COM or USB socket.

 

Never had a TimeSync problem except that since my USB GPS is ten years old, it is not as sensitive if inside home, compared to my cell phone which receives all satellites even in the basement!

 

Perhaps it would be more interesting to be able to use cell phone GPS in some way, to be paired with the PC, but I wouldn't know how...

73

Modulo required accuracy this may be good or bad. Cogitate a little while on the effects of multipath on GPS accuracy, especially when you only get the signal via a bounce path. (If you are interested in nanoseconds.)

{o.o}

Well, for most uses add about 3 or 4 ms to the time mark data read and average out a 7 ms peak to peak jitter and it should be modestly accurate maybe down to usual ntp accuracy.

{^_^}