Lmst

I AM GETTING A FOUR POINT MOUNT SC CUT 10 MHZ CRYSTAL!

Yes I'm shouting. #timenuts

If you have a Leo Bodnar LBE-1420 series GPS locked oscillator like mine I suggest you update to the latest firmware, even if you're already running the latest firmware. There was a bug fixed in version 1.06 of the firmware and the new version is also 1.06 #timenuts #metrology #electronics

Hardware Hacking / Timenut Nerds: This is but a lightning talk, but I'll *really* be looking forward to it:
https://events.gnuradio.org/event/23/contributions/612/

"The cheapest GPSDO yet"

It's going to be streamed!
See you all at the European GNU Radio Days '24.

#GPS #GNSS #clocknerd #frequency #pps #timing #timenuts #pll #SDR

This is the Allan deviation of the same data. It looks pretty good until about tau=3000 seconds. (I'm not sure what the bump around 1400 seconds is.) Regardless, they look like reasonable oscillators to help stabilize the system clock for a few hundred seconds.

Oh yeah, and the code is hacky so data points aren't quite as clean as they could be. #timenuts #timekeeping

A plot of Allan deviation of the two devices. They have a nicely downward slope from 16s to about 600s. They the get a little wobbly as they continue down. EM0 starts to turn upward around 3600s. EM1 continues down but at a lesser slope until the end at about 9000s.

Unsynchronized PPS Experiment
https://blahg.josefsipek.net/?p=611 #electronics #timenuts #chrony #FreeBSD #PPS #OCXO #timekeeping

Photo of a breadboard with a circuit connected to the back of a server.

@ErikUden The 2038 page has a month name that ends with “ber” when the overflow happens in January 2038. Yeah, I know… I’m nuts… #timenuts

In 2019, I bought 3 cheap "#GPS pucks" on AliExpress. Last night, I cracked one open to see what's in it.

The case isn't intended to be opened, but once inside a single PCB falls out. One side with a patch antenna and a red LED (not visible through the opaque plastic case 🤷), and the other with the receiver, few other components, and unpopulated footprints. The #PPS period measured as 18ns short of 1s (photo from before the frequency counter warmed up). #GNSS #electronics #timenuts

Photo of a black rectangular GNSS receiver "puck" on a window sill. The top of the plastic case says "GPS" in an oval. A black USB cable is coming out of a side.

Photo of the top side of the PCB. Nearly the entire PCB is taken up by the patch antenna. There is a red LED near the USB cable on the few mm of PCB between the antenna and the edge.

Photo of the other side of the PCB. Most of the space is take up by a can with a sticker saying "Stoton", "GNSS Receiver", "GN-801", a barcode, and CE, FCC, and RoHS logos. The other components are a small coin battery, and a handful of passives. There is a 4-pin chip of some sort (a voltage regulator maybe?).

There are silkscreen outlines for a few components - including a 16-pin chip of some sort, and a 6-pin connector.

There are 6 test pads near the edge and 6 labels (GND, VCC, TXD, RXD, PPS, EN).

The USB wires are soldered directly to 4 of 6 test pads.

Photo of Racal-Dana 1992 frequency counter showing 1000.00050e-3 seconds.

@electronic_eel And the results are in! Setting lagghash to l2 - ok. l2,l3 - ok. l2,l3,l4 - latency spikes. Everything else stayed the same. #NTP #chrony #FreeBSD #timenuts

rrdtool image showing the NTP peer time offset for my server. There are 7 lines which are mostly stable once I switched away from l2,l3,l4 hashing shortly after midnight Aug 29th. In early Aug 30 morning, I switched from l2 to l2,l3 and the graph remains essentially the same. There is a spike mid-day on August 29th that corresponds to when I ran a disk I/O intensive command.

I was curious about how fast the ADM3202 RS-232 line drivers were. So, I fed one with 3.3V pulses and looked at the output. Apparently, it takes 100ns before the output moves and then it takes 250ns to go from +6.3V to -6.3V. However, the standard defines anything lower than -3V as a mark, so at that point the serial device should see it. This means only 166ns for the transition or 266ns delay in total. This is significant, but it seems very constant. #electronics #timenuts

Oscilloscope screenshot (50ns/)

Channel 3 (cyan, 1V/) showing a quick rising edge from 0 to about 3.3V in less than 10ns
Channel 2 (magenta, 5V/) showing a slowly falling edge from 6.3V down to -6.3V over 250ns

Last week I got myself a #TinySA Ultra. It has a signal generator mode, so here I am today comparing its oscillator stability against the bench signal generator using a #NanoVNA reflashed as #TinyPFA.

I was having trouble believing the output of my adev script because it showed such high values, so I loaded up the data in TimeLab which drew the same exact plot. 🤷 I also plotted the measured phase delta between samples. I do *not* think that's thermal (air conditioning) related. #timenuts

Photo showing TinySA Ultra in signal generator mode and NanoVNA running the TinyPFA firmware.

TimeLab plot of the logged phase data showing a downward line (good), but with the y-intercept at 0.4s (bad).

A quick plot of sample-to-sample phase difference showing a saw-tooth pattern.

I added blanks for readability and a button that cycles through the display modes I didn’t mention I had :)

I still don’t have a real external source of time, so things initialize at 2000-01-01 00:00:00.

Previously, you may have noticed the two AVRs on a breadboard. One is the “oscillator” that generates an IRIG-A signal (this is where the initial date comes from). The second chip parses the IRIG data and issues the necessary SPI commands to display it. #electronics #IRIG #timenuts #clock

It’s alive! #electronics #clock #timenuts

The digits are a glorious shade of red which unfortunately doesn’t show up in the video at all.

This is just the “demo” mode, it is a fully generic SPI-programmed 10-digit 7-segment module. Eventually, it will be used for a clock but until now it just counts.

Woohoo! Looks like my math is right. Both TimeLab and stable32 generated the same exact curve as my script. #timenuts

Allan deviation image from TimeLab showing the same kind of wobble as my DIY plot in previous post.

Allan deviation image from stable32 showing the same kind of wobble as my DIY plot in previous post.

The problem with trying to write my own script to make a sigma-tau plot is that when I get a weird curve I have no idea if that's because I screwed up my code or because the data I acquired is funky. I suppose I should install stable32 or timelab and see what kind of plot they generate. #timenuts

A sigma-tau plot.

x-axis [0.1,135] seconds
y-axis shows what looks like the beginning of a sine wave

My first go at a raspberry-pi GPS NTP server. 1μs predicted error ain't bad, but I plan on replacing the system clock crystal with a TXCO in the next few days to see if I can improve that.

I used https://austinsnerdythings.com/2021/04/19/microsecond-accurate-ntp-with-a-raspberry-pi-and-pps-gps/ as a rough guide and https://raspberrypi.stackexchange.com/questions/74482/switch-out-the-x1-oscillator-on-a-rpi-2-3/109074#109074 for inspiration on improving the system clock.

More to come!

#timenuts

Good news everyone!

IERS Bulletin C Number 65 has just been published, and (as expected) NO #LeapSecond will be introduced at the end of June 2023.

#NTP #Time #TimeNuts

https://datacenter.iers.org/data/latestVersion/16_BULLETIN_C16.txt

A graph of the deviation between UT1 and UTC on seconds since sometime around 2005. A sawtooth pattern can be seen as 5 leap seconds have been inserted. Over more recent years the delta has remained more steady.

Woah! International Bureau of Weights and Measures has voted to discontinue leap seconds by 2035.
Catch those last few times the clock says 23:59:60 while you can.
#timenuts #leapsecond #timekeeping

Unhappy because I found my $0.5 crystal is off by 4 parts per million. I think I'm on my way to be the #timenuts - every $1 microcontroller should have a hydrogen maser clock!!! :doge: #electronics #amateurradio #hamradio

#TimeNuts

Client Info