Centerclick NTP390

@stevesommars, I hope to have the NTP390 in a few days.
This device should have a incredibly good clock.

20 PartsPerBilion, not Million like others.

It will replace my NTP220, so you can test it.

I have no clue if it makes any difference, but hey, it should be so precise…will it be better then the 220 or my Garminpuk you tested before.

I wonder…I hope to have it this week.

@dave-ntppool

ntpd reported jitter is about an order of magnatude better than 2xx, but I suspect this is mostly due to cpu/kernel.

20ppb is the free-run for the OCXO, when GPS is present we can discipline it down to <2ppb typical if you keep the operating temp constant (the OC part is good but not 100%). we monitor the XO stability vs PPS via our new timing processor (measures at a resolution of ~5nsec) and tracks the trend vs XO see NTP390-demo7 - Timing Processor Details

I see in the demo you use Baidou and GPS.
I tried to do that in my 220 but it don’t seem to enable Baidou.
Then I only get GPS and SBAS.

Why are you using Baidou in the demo for the 390?

On the 220, if you enable BeiDou, it will disable other GPS satellites; it can only read from two at a time or something like that, but I cannot remember the exact details.

The 390 does allow GPS+BeiDou.
I have just tried it and it works.

As I noticed that Dave’s 390 is a lot more accurate using those 2 instead of the others I used with the 220.

I let it run for a bit with those 2 and see how close I come to Dave’s numbers :crossed_fingers:

My 220 ran with GPS+Glonass+Gallileo, 3 of them.

@stevesommars, can you test this 390? It’s unlimited open for your IP to put the pressure on it.
it should be a lot more stable then the 220 you tested before.

Wow…running GPS+BeiDou is amazing.
Now I know why you @dave-ntppool use it.

I have the same numbers like your demo shows.
And the antenna is indoor.

This is amazing. Going to test the 220 to see if it can do the same.

Update: The NTP220 can do it too. Now testing both of them side by side.

let the OCXO bake for another day or two and it’ll get another order of magnatude stable, you’ve only been running for 24 hours.

Ok, nice to know.

But I’m also comparing the 220 against the 390 both set at GPS+BeiDou as I found that BeiDou seems to have better precision because they are stationary and have a higher SNR then others.

It does take longer to pickup, but once in use, they look to have a better constant PPS signal.

Never knew that.

I wonder how close the 220 will get to the 390.

I have setup a Chrony on my workstation to monitor all my 220’s and the 390, also against my Garmin puck.

All are warming up as we speak:

root@workstation:~# chronyc sourcestats 
Name/IP Address            NP  NR  Span  Frequency  Freq Skew  Offset  Std Dev
==============================================================================
nas Garmin gps/glo local   22  13   339     +0.230      1.051    +16us   139us
ntp220 gps/bei local       54  29   874     -0.220      0.186   +127us   107us
ntp390 gps/bei local       64  35  1035     +0.179      0.170   -151us   116us
ntp220 gps/bei             23  11   31m     -0.423      1.036  +1679us   697us
sprintweb.be vps           13   8   21m     -0.783      1.489   -650us   446us
ntp220 gps/gal/glo         25  14   34m     +0.276      0.385   -193us   278us

I hope they soon show the difference.

Ok tested.

The NTP-220 won’t use the BeiDou. It doesn’t, not even after several hours.
The 390 does.

Not that it matters much for most. :rofl:

Wow - the NTP390 looks impressive with its GPS-disciplined VC-OCXO (20ppb)

Its interesting that all of the models have the same packet rate limit: 10,000 packets/second

Any idea what is limiting the packet rate or if there is a way to increase?

This would be right at the edge of what my FreeBSD SuperMicro stratum 1 is running with 3gbit net speed in North America

The limit is quite simple.
The installed NIC is limited to 100Mbit, not 1Gbit.

So the packet/second is limited to that NIC speed.

I checked the port, it has all 8 pins.
As the software is not OpenSource, there is no way of making it 1Gbit.

@dave-ntppool , maybe Dave can answer this.

And yes, it’s running for some time now, it’s impressive how accurate it is.

A 100 Mbit/s NIC can handle more than 100000 NTP queries per second, so that’s not the limiting factor.

But the limit isn’t 10K…where do you get that?
Reading is hard for some:

afbeelding

It states, it tracks 10K, but serves 1M.

And the 390 tracks 250K but still serves 1M.

So?

It does state 10K/sec, but I doubt that is a hard limit.

It can serve 1M untracked IPs at a rate of 10k packets per second

My question is the 10k packets per second - for NTP packets that limit is less than 10 mbps, the fast ethernet NIC should be able to handle much more.

I can not answer that, but @dave-ntppool can.

Maybe it’s just a limit he states it can do a minimum…no idea.

I don’t reach that limit.

10,000 packets/sec is all we’ve tested to, both could probably do more before hitting a CPU limit

@dave-ntppool , I have a question, somebody here in the forum is about to buy my first 220 I bought from you.
As I replaced it with the 390. Will he be covered by the remaining warranty I’m supposed to have?

I hope you do, as the first 220 I have bought is now on the shelf doing nothing, and I do not expect anything to happen to it.

But some other person wants to have it.

Is this a problem? I hope not.

Hi,

interesting to see those statistics!
What means local? is it connected via PPS OUT?
I ask because there is a difference between the ntp220 local and non local. The latter one seems to have a huge offset.

I don’t know what it is or does.

@dave-ntppool I presume Dave knows, as I can’t find anything about it on this website.