Posted at 2022-01-20 by johan_m_o

Are you using an Android device and Gadgetbridge? If so, check the Gadgetbridge settings for "Sync time".

If you're using an iPhone I have no idea.

Posted at 2022-01-21 by @fanoush

Also in App loader webpage there is option to sync time when connecting - the rightmost settigs tab.

Posted at 2022-01-21 by malaire

According to [1] Bangle.js 2 doesn't support getting time from Gadgetbridge.

Also [2] says that "sync time" only syncs when watch gets connected to phone, which is not enough. Bangle.js 2 is quite inaccurate and needs sync several times per day.

[1] https://www.espruino.com/Gadgetbridge#how-it-works-internally
[2] https://codeberg.org/Freeyourgadget/Gadgetbridge/wiki/Configuration

Posted at 2022-01-21 by malaire

Yes but that is not automatic.

Posted at 2022-01-21 by johan_m_o

That page doesn't mention anything about syncing time, as far as I can see. But maybe that's what you're basing your assumption on.

I have not seen that Bangle.js 2 is that inaccurate... But of course, I do a lot of tweaking and often end up disconnecting and connecting many times during a day. Will have to test that at some point.

Posted at 2022-01-21 by malaire

That page doesn't mention anything about syncing time, as far as I can see. ...

Yes, it lists what is supported - time is not listed so it's not supported.

I have not seen that Bangle.js 2 is that inaccurate...

For example I synced my Bangle.js 2 yesterday evening and it had <1s error. Currently error is already over 3s and it's been well under 24 hours. (Acceptable error is 2s at max, but preferably well under 1s.)

Posted at 2022-01-21 by johan_m_o

Ah, but syncing time is not sent as a message and messages is what you see listed on that page. I agree that it might be good to have it mentioned in there somewhere, but I wouldn't say that the page says that syncing time isn't supported.

About having syncing occurring at other times than when connecting, I leave that to @gfwilliams...

Posted at 2022-01-21 by @fanoush

Bangle.js 2 is quite inaccurate and needs sync several times per day.

Currently error is already over 3s and it's been well under 24 hours.

this is something that can be hopefully improved
bluetooth uses 32kHz crystal with 20PPM accuracy https://github.com/espruino/Espruino/blob/master/targets/nrf5x/bluetooth.c#L2401
that means according to https://www.best-microcontroller-projects.com/ppm.html accuracy of 1.73 seconds per day. I think the RTC that can keep time can use this crystal as a clock source too. Not sure if it currently does and where currently the time value comes from. So unless you reboot the watch which spends some unaccounted time in bootloader (where the hardware including RTC may be reset and disabled) it should/could be accurate.

EDIT: here https://github.com/espruino/Espruino/blob/master/targets/nrf5x/jshardware.c#L1006 it looks like the same time source is used so not sure why it would be so inaccurate (unless you rebooted the watch)

Posted at 2022-01-21 by myownself

@numerist has also mentioned problems with accuracy of the clock. I am struggling to find the thread, but the fact that he wanted to use GPS time may have been a red herring/distraction from whatever @malaire is also seeing.

Posted at 2022-01-21 by malaire

I might've reset the watch, I'm not sure. I'll have to recheck once I get my watch to work with Bluetooth again.

But even that 1.73 seconds per day would accumulate to quite large error since Gadgetbridge doesn't even have daily sync.

Posted at 2022-01-21 by @fanoush

But even that 1.73 seconds per day would accumulate to quite large error

Well, it is also temperature dependent but as long as you wear the watch the temperature may not drift a lot. So if temperature is stable the daily drift of your specific crystal can be constant (?) so it could be compensated in software. This of course matters mainly if you cannot sync with phone/internet.

BTW there is also BLE service for time defined so one could have some accurate time source around (e.g. ESP32 connected to internet adverting this over BLE) and then have app on bangle to scan for it and sync the time.

Posted at 2022-01-21 by malaire

BTW there is also BLE service for time defined so one could have some accurate time source around (e.g. ESP32 connected to internet adverting this over BLE) and then have app on bangle to scan for it and sync the time.

Wouldn't I need to manually unpair Bangle.js 2 from my phone, pair to that timesource, then unpair from timesource and pair with phone?

Posted at 2022-01-21 by @fanoush

No if bangle is the active one in this, currently you can have one connection into Bangle and one from Bangle to some other device. They are unrelated.

Also BTW every recent iphone have such service implemented (as per 40.12.4 Available Services )

But this could be also done over advertisements so no connection would be needed, just scan for devices around and get the data directly from the packet. However that would be custom solution but relatively easy to do on both sides. maybe even esp32 board running espruino could do such current time advertisement (while syncing with NTP servers over wifi)

Posted at 2022-01-21 by Numerist

@MyOwnSelf
It's possible that on Bangle.js 2, the time gains too much. There are variables still to test, and it's still possible that the code I'm using for a very specialized clock (that no one else uses) is the problem. However, overnight the well-known Anton Clock seems to have gained a lot also. Temperature is not the factor here.

I'm interested to know whether others have the gaining problem that @malaire and I have noticed.

Gaining 2 seconds a day is okay with me only if I have a way to automatically correct that daily. At present the only way I know is to reload my watch script manually, although just now even that didn't work.

If I could sync to an iPhone, I'd be all set. I'm not aware that's possible.

I'm not a coder and know very little of javascript or about Bangle.js. I still don't know whether on Bangle.js 2 the GPS time, once set outside, still works inside, and whether I can access it continually. That seemed to solve my time problem on Bangle.js 1.

Posted at 2022-01-21 by malaire

hmm, ok, I thought I read somewhere that Bangle.js 2 only supports one Bluetooth connection at a time.

I wouldn't mind a custom solution. I'm a programmer and my Debian 11 desktop computer is synced to NTP and has Bluetooth dongle, but I don't know enough of Bluetooth to try implementing something like that.

Posted at 2022-01-21 by johan_m_o

hmm, ok, I thought I read somewhere that Bangle.js 2 only supports one Bluetooth connection at a time.

That's correct.

Posted at 2022-01-21 by Numerist

1. Does GPS time run on Bangle.js 2 indoors after being connected outdoors?

2. Is it possible to sync with an iPhone's time? Automatically once or twice a day, if possible.

Posted at 2022-01-21 by Mark_M

Yeah, I got same idea. Even tried to write such time announcement service in Linux. But low experience with BLE and shortage of time holds me...

Posted at 2022-01-21 by malaire

I can't resist trying to learn something new, so I managed to implement quick&dirty BLE time broadcast for Debian 11.

Broadcast side (bash, bluez), one broadcast per second for 60 seconds:

sudo hciconfig hci0 leadv 3
for (( n=1; n<60; n++ ))
do
    time=$(date +%s)
    hex=$(printf '%010X' $time)
    echo $hex
    sudo hcitool -i hci0 cmd 0x08 0x0008 1E 02 01 1A 1A FF FF FF \
        ${hex:0:2} ${hex:2:2} ${hex:4:2} ${hex:6:2} ${hex:8:2} \
        00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
    sleep 1
done
sudo hciconfig hci0 noleadv

Receiver side, received_time is Unix timestamp in seconds:

packets=100;
NRF.setScan(function(d) {
  packets--;
  if (packets == 0) {
    NRF.setScan();
  } else if (d.manufacturer == 65535) {
    let received_time = 0;
    for (let i = 0; i < 5; i++) {
      received_time <<= 8;
      received_time += d.manufacturerData[i];
    }
  }
});

Posted at 2022-01-21 by myownself

Nice quick work.
I'm curious though - does doing something like this actually have the kind of accuracy you are after, or is it just adding a different offset? (maybe it is better, I have no idea whatsoever)

Posted at 2022-01-21 by malaire

Currently accuracy is quite bad - that sudo hcitool sets up repeated broadcast which I change every second, so accuracy is about 1 second here.

To get better accuracy I'd need to learn how to send broadcast at once without delay. BLE seems to have inherent random delay of around ~10 ms so e.g. 1 ms accuracy is not possible, but I think that accuracy of <100ms should be possible.

Posted at 2022-01-21 by malaire

btw, there is a Python BlueZ example here and Bangle.js 2 does see those messages, so that would probably be better base (no need for bash or sudo), but I don't really know python so I didn't try modifying that code.

Posted at 2022-01-21 by diego

Hi, the code below should work to send the time to the device: it opens the connection, sends the command to update date, and then closes the connection.

import asyncio
from time import time
from bleak import BleakClient


UUID_NORDIC_TX = "6e400002-b5a3-f393-e0a9-e50e24dcca9e"
UUID_NORDIC_RX = "6e400003-b5a3-f393-e0a9-e50e24dcca9e"


async def main(address):
    buffer_size = 20

    print(f"Connecting to {address}...")

    async def send_command(command):
        c = command
        while len(c) > 0:
            await client.write_gatt_char(UUID_NORDIC_TX, bytearray(c[0:buffer_size]), True)
            c = c[buffer_size:]

    async with BleakClient(address) as client:
        print("Connected")
        print(f"Time: {time()}")
        await send_command(bytes(f"setTime({time()});\n", "utf-8"))

asyncio.run(main("d8:ee:7c:31:a9:c0"))

I'm using something similar to this to periodically sync weather data and my personal agenda with the device

Posted at 2022-01-21 by malaire

Does that require that Bangle.js 2 is not connected to phone or computer, as it can handle only one Bluetooth connection at a time?

Posted at 2022-01-21 by diego

Yes, I don't use the gadgetbridge, so it works for me.

Posted at 2022-01-21 by Mark_M

yes, we do not want to connect, just to scan for broadcasted ads with a time in data. May be in future there should be a certificate or at least a shared public key for security. So some hacker do not set all Bangles around to a wrong time. :)

Posted at 2022-01-21 by myownself

Thanks @diego - I think I might end up using this, I think connecting will result in a much more accurate time than relying on the right advertisement being picked up at the right time.

I had a look at Bleak recently (it is in the Espruino tutorials somewhere), your example will be useful to me for another project I think.

Posted at 2022-01-22 by @fanoush

1 ms accuracy is not possible

I think it is but not with such simple code. Great quick example. Few random ideas -
you can use setScan with right filter, this needs to be combined with some unique data in advertisement - e.g. manufacturer or serviceData. I think for serviceData id you could use id of specific characteristics in ble current time service or any other defined to contain time and then the data could match the format. It is not exactly to the standard but is close enough to not conflict with anything. see also https://devzone.nordicsemi.com/f/nordic-q-a/4048/what-service-data-can-be-added-to-the-advertising-package

manufacturer is ok too but if you have more espruino devices you'll get more matches, with serviceData it would be unique

https://www.bluetooth.com/specifications/assigned-numbers/
https://www.bluetooth.com/specifications/specs/current-time-service-1-1/

the 16bit assigned numbers for different entities are unique and not overlapping between types so that's why I think you can put it safely to serviceData even if it is not id for service

as for accuracy - ntp prococol can do it quite well over internet, bluetoth timing is strict, advertising packets are timed/spaced in deterministic way. the bash+hci command is probably the worst part here. I tried latest espruino with esp32 yesterday, looks like it can do ntp over wifi and can do Bluetooth advertising. I hope we could possibly generate advertising packet (or scan response packet) on the fly there. But anyway even second accuracy is good enough too.

EDIT: for esp32 to be usable for that we'd probably need something like this https://github.com/gmag11/ESPNtpClient to have the time in esp32 accurate to 1ms

Posted at 2022-01-22 by @fanoush

I think connecting will result in a much more accurate time than relying on the right advertisement being picked up at the right time.

I guess both ways could work. The callback in setScan gives you the packet as it is received, not sure whether with connection you can get it faster. Connection/GATT stuff is another layer on top received packets that may slow things down or make it less deterministic. Advertisement packets are much simpler/lower level. Try with nrfconnect on android, there is history of advertisement packets received - click 'MORE' on device list to see how they arrived to the phone - it is not random at all.

Posted at 2022-01-22 by malaire

@MyOwnSelf

I think connecting will result in a much more accurate time than relying on the right advertisement being picked up at the right time.

I thought so too but in my tests that connection takes several seconds while advertisement-solution can be improved down to e.g. 50ms delay between updating advertisement, so using connection is a lot worse for some reason.

@fanoush

bluetoth timing is strict, advertising packets are timed/spaced in deterministic way

No it's not deterministic - Bluetooth protocol has builtin random delay for timing which is around 10 ms or so, so accuracy better than that is not possible using normal API. (But yeah it's good enough.)

Posted at 2022-01-22 by @fanoush

No it's not deterministic - Bluetooth protocol has builtin random delay for timing which is around 10 ms or so

Oh, you are right for the advertising, there is delay indeed. Decribed e.g. here https://www.argenox.com/library/bluetooth-low-energy/ble-advertising-primer/ didn't know this. Makes sense. However it is not like that for the rest, the rest has strict timing so receiver device can wake at the precise moment sender is supposed to send the packet, that is part of the 'low energy'.

But anyway even with advertising the timing for scan response packet may be better. This is with active scanning, it is second packet for advertising data.

EDIT:
Scan response is more deterministic, see https://devzone.nordicsemi.com/f/nordic-q-a/29719/beacon-scan-response-request-and-packet-timing

Posted at 2022-01-22 by malaire

Scan response timing seems deterministic only in relation to preceding advertising packet. For that to be useful I'd need to
a) know the exact time when preceding advertising packet was sent
b) set response packet content after preceding advertising packet has been sent, so it's uptodate

I don't think either of those are possible with normal APIs since advertising isn't meant to be used for sending time-sensitive data like this. (Sure both would be possible in theory with custom Bluetooth protocol stack.)

Posted at 2022-01-22 by Numerist

@diego
Unfortunately I don't know what the recent responses mean. I'm not a programmer. (Someone else built my clock.) Is it possible to insert one of the scripts above (yours?) and get good time? What do they do? Where do they get the time from? (Where's the BLE source?) How often? Is the connection "automatic"?

Thanks.

Posted at 2022-01-22 by @fanoush

For that to be useful I'd need to
set response packet content after preceding advertising packet has been sent, so it's uptodate

Yes. This may actually work with nordic softdevice 6.1.x , see espruino/Espruino#2000 since 6.1.0 you need to supply static buffer that needs to be allocated for whole advertising time. So most likely current data will be sent as is from that buffer, so updating timestamp in it periodically or when some event happens (like when someone asks for scan response) may work. Documentation says it should not be modified but if one does not change the structure and just update timestamp in specific location it may work.

Anyway, opening connection and getting time from characteristics (possibly few times) may be easier after all if millisecond accuracy is preferred.

Posted at 2022-01-22 by diego

This is one approach: it connect to bangle then sends the computer time (hopefully a good time)
within the command to update bangle. Then you have only the delay of the communication with the device, which can be measured and adjusted.

On my device, this is the average for 100 samples: 0.05636221885681152s, so I could adjust to add 50ms on the time sent to the device to compensate.

Posted at 2022-01-22 by diego

Sorry, @numerist, I think I was still a bit programmy on my response.
In more broad terms, the script above when executed on a computer with bluetooth sets the clock on the bangle to the time of the computer.

As the communication of the command between pc and bangle is kind of fast, there's little need for compensating (50ms in my tests, but you can antecipate and correct for that if you wanted)

Posted at 2022-01-22 by diego

Also, I've just made a pex (executable) file from the sample script I've sent earlier, it's attached, if you want to try it out.

The pex is just a glorified zip file, so you can open it and check out inside, but you can execute with: ./send_time.pex 'aa:bb:cc:dd:ee:ff'

The usage:

usage: send_time.py [-h] [--correction CORRECTION] address

Bangle send time

positional arguments:
  address               bluetooth address to connect

optional arguments:
  -h, --help            show this help message and exit
  --correction CORRECTION
                        milliseconds to add to time to compensate for connection time

Attachments:

• send_time.pex

Posted at 2022-01-22 by Numerist

@diego
No problem; I'm the odd one here with no programming knowledge. I don't know how to run scripts like yours on my computer but may be able to find out…but what I really want is something to put on the watch that will get the time from my phone, preferably automatically at a specified interval (e.g. once or twice a day). Or if that's not possible, from my computer.

Even better would be something like what worked for Bangle.js 1: set the GPS time once, and it controls the watch indefinitely. I still don't know if that's possible with Bangle.js 2. People say the GPS setup on the new watch is different, but what does that mean for this question?

Posted at 2022-01-22 by malaire

My BLE time advertising is starting to work quite well. (I switched to Rust for broadcast code instead of bash as I know Rust quite well.) Current version doesn't actually change watch time, just applies the delta to displayed time.

https://www.youtube.com/watch?v=UikY25KiXDU

Posted at 2022-01-22 by @fanoush

Oh nice :-)

BTW one possibility for Service Data is to advertise "Date Time" 0x2A08 as per https://forum.micropython.org/viewtopic.php?t=9577&p=53698
it is 7 bytes format

             0 1  2     3   4    5      6
Description  Year Month Day Hour Minute Second
Minimum      1582 1     1   0    0      0
Maximum      9999 12    31  23   59     59

Then one could pass options {filters: [{serviceData:{"2a08":{}}}]} and it should work. I tried it with ESP32 running espruino and it worked for me
On ESP32 I've run

NRF.setAdvertising({
  0x2a08 : [0xe6,0x07,1,22,17,19,30]
})

and then on another device this worked

>packets=2
=2
>NRF.setScan(function(d) {
:  packets--;
:  if (packets<=0)
:    NRF.setScan(); // stop scanning
:  else
:    console.log(d); // print packet info
:},{filters: [{serviceData:{"2a08":{}}}]});
=undefined
BluetoothDevice: {
  "id": "b4:e6:2d:89:01:27 public",
  "rssi": -51,
  "data": new Uint8Array([2, 1, 6, 10, 22, 8, 42, 230, 7, 1, 22, 17, 19, 30, 14, 9, 69, 83, 80, 51, 50, 46, 106, 115, 32, 48, 49, 50, 53, 0, 0]).buffer,
  "name": "ESP32.js 0125",
  "serviceData": {
    "2a08": new Uint8Array([230, 7, 1, 22, 17, 19, 30]).buffer
   }
 }
> 

Posted at 2022-01-22 by malaire

I'm now including milliseconds also for more precision. Bangle.js 2 doesn't support setting time at millisecond precision, but this works quite nicely (wait until next exact second and then set time):

      let delta = 1000 - (received_time % 1000);
      setTimeout(function() {
        setTime((received_time + delta) / 1000);
      }, delta);

Posted at 2022-01-22 by malaire

0x2A2B "Current Time" seems better. It has those same fields and then 3 more single-byte fields, most importantly fractions256 for 4 ms accuracy.

1. dayOfWeek: 1=Monday, 2=Tuesday, ..., 7=Sunday
2. fractions256: 0-255 ; 1/256th:s of a second
3. adjustReason: 0 for this application

Posted at 2022-01-22 by @fanoush

or just the 0x2a0c "Exact Time 256" which is without adjustReason

They are nested structures, see
https://infocenter.nordicsemi.com/index.jsp?topic=%2Fcom.nordic.infocenter.sdk5.v15.3.0%2Fstructexact__time__256__t.html or https://community.silabs.com/s/question/0D51M00007xeHRw/implementation-of-folded-ble-characteristics-in-gatt?language=en_US

my current ESP32 code is

// enable NTP
require("Wifi").setSNTP("pool.ntp.org","+01")
// advertise time
setInterval(function(){
  var d=Date(),y=d.getFullYear(),day=d.getDay();
  NRF.setAdvertising({
//    0x2a08 : [y&255,y)>>8,d.getMonth()+1,d.getDate(),d.getHours(),d.getMinutes(),d.getSeconds()]
    0x2a0c : [y&255,y>>8,d.getMonth()+1,d.getDate(),d.getHours(),d.getMinutes(),d.getSeconds(),day?day:7,d.getMilliseconds()>>2]
});
},250);

Posted at 2022-01-22 by malaire

It seems that the 0-10ms random delay is not added to first advertisement, only to subsequent advertisements.

I'm currently testing code like this which seems to give quite steady rate of one advertisement every second (I hope to publish full code soon, maybe tomorrow):

• start advertising
• wait 100 ms
• stop advertising
• wait 900 ms
• loop from beginning

For example here are differences in milliseconds of Bangle.js 2 time and my computer time for 5 minutes, one update every 10 seconds:

63 64 71 65 65 69
66 69 70 70 70 69
71 73 75 73 73 72
76 77 77 77 79 78
80 82 81 84 88 81

You can even see the clock drift here. With 1.73s/day you'd expect clock to drift 6ms in 5 minutes - at least for this short test it's clearly more than that.

Posted at 2022-01-23 by pelrun

The 10ms jitter is a deliberate collision avoidance mechanism in BLE. Do you really need sub-10ms accuracy on a watch?

Posted at 2022-01-23 by malaire

No not really, I would've been happy with 200ms accuracy, but it's fun to see how accurate I can get.

I could add my own 10ms jitter to the 900ms sleep. The main thing is that there is no random delay between getting time from computer and sending advertising with that time.

Posted at 2022-01-23 by Mark_M

additionally a delay can be figured out if there are 2 BLE devices on a computer.
One Advertiser, another Scanner. The Scanner, when it receives time can compare it with the computer time.
I may be try it with laptop internal BLE and USB dongle.

Posted at 2022-01-23 by malaire

Advertiser+Scanner could be used to advertise really accurate times:

• Advertiser sends time advertisement with ID
• Scanner receives that advertisement, compares included time with actual time, reports error to Advertiser
• Advertiser sends next time advertisement with next ID and error of previous advertisement
• ...

Then Bangle.js would need to receive two subsequent time advertisements to know exact time first advertisement was sent.

ps. I've now added my initial code to GitHub.

Posted at 2022-01-23 by @fanoush

I have also modified your bash example a bit, it runs on raspberry pi and uses the 0x2a0c "Exact Time 256"

advpad="00 00 00 00 00 00 00 00 00 00 00 00 00 00 00"

ble_advertise_datetime(){
  values="$(( $1&255 )) $(( $2/256 )) $3 $4 $5 $6 $7 $8 $(( $9/4000000 ))"
  sudo hcitool -i hci0 cmd 0x08 0x0008 10  02 01 06  0c 16 0c 2a $(printf '%02X %02X %02X %02X %02X %02X %02X %02X %02X' $values) $advpad
}

sudo hciconfig hci0 leadv 3
i=60
while [ $i -gt 0 ]  ; do
  echo $i
  ble_advertise_datetime  $(date "+%-Y %-Y %-m %-d %-H %-M %-S %-u %-N")
  sleep 0.5
  i=$((i-1))
done
sudo hciconfig hci0 noleadv

Posted at 2022-01-24 by malaire

That's certainly simpler than cross-compiling my Rust code to Raspberry Pi. (I just got that working and I'm now reading on systemd services to make this a service. I can provide executable for latest Raspberry Pi OS if anyone wants to test this.)

Posted at 2022-01-24 by BartS23

It may be possible to make it much simpler.

The GPS chip already provides the time about 1-2 seconds after switching on (even without fix).

I wrote an app, which turns on GPS every 4 hours, sets the clock if necessary and then turns off the GPS again.

If no time is received within 10 seconds, the GPS is also switched off.

A log file is written to measure the deviations and possibly make adjustments in the rhythm (this must be deleted manually).

Unfortunately I am not that experienced in writing js.
I would be happy if someone could review my PR (espruino/BangleApps#1342).

Posted at 2022-01-24 by Numerist

Does your script work indoors? I've been asking here how to access the GPS, including after an outdoor fix, and whether it continues to work indoors, but no one has answered. Bangle.js 2 is clearly different from Bangle.js 1 in this, but again no one has said how that affects getting an accurate time.

Posted at 2022-01-24 by myownself

What sort of drift have you noticed so far between the two? Aren't the crystals in the GPS and Bluetooth at the same frequency (32.768 kHz)?

Posted at 2022-01-24 by Mark_M

GPS time can be good and precise. But yes, indoor reception can be very poor. Also it is UTC. You need to keep timezone and time savings adjustments somewhere.

Posted at 2022-01-24 by Numerist

With Bangle.js 1 I had the GPS run the clock after setting the GPS outside. It worked. Indoor reception, which is nil, didn't matter. Time zone etc. are no problem: they're in settings.

Again, what's the difference in how GPS time works with Bangle.js 2?

Posted at 2022-01-25 by Mark_M

As I understand there is no much difference. GPS should receive a time from a sat and then it can keep it in its RTC. My uderstanding, the main battery is connected to AT6558 backup power supply. When you turn off GPS power AT6558 will keep time in RTC and ephemerides in memory.
But the quartz is external, not in the chip, so if it is not good enough it can drift similar way as a "main" watch quartz. In normal operation GPS always receives time from sats. So RTC is just a backup, when no reception. Because it is not that important, they may put a cheap quartz there.

Posted at 2022-01-25 by Numerist

I've read that there are a few significant differences in the GPS operation. I myself do not know.

Today I did a complete reset and uploaded only one clock (Anton). In 8 hours it gained a little less than 2 seconds. Does anyone else have such an observation? Does your Bangle.js 2 gain throughout the day?

(Then I uploaded my clock. It reset Anton (and set itself) to the correct time.)

Posted at 2022-01-25 by BartS23

Does your script work indoors?

In my apartment and at work it also works indoor.

What sort of drift have you noticed so far between the two?
Approximately every 12 hours the difference increases to 2 seconds.

I reset my GPS chip to factory settings with

$PCAS10,3*1F

and still got a time within a second.
I think as long as there is a satelite in view, it will also find the time.

Posted at 2022-01-25 by malaire

... In 8 hours it gained a little less than 2 seconds. Does anyone else have such an observation? Does your Bangle.js 2 gain throughout the day?

Yes, my Bangle.js 2 is also gaining several seconds per day (I havn't measured it exactly yet.)

(Then I uploaded my clock. It seemed to reset Anton (and set itself) to the correct time. Weird.)

Every time you connect to https://banglejs.com/apps/ the watch time is set by that website.

Posted at 2022-01-25 by myownself

@gfwilliams I noticed espruino/Espruino#2125 last night. Is this related, or am I conflating two completely different things just because they relate to timers?

Posted at 2022-01-25 by @gfwilliams

@MyOwnSelf they're different - the high res timer is very specifically for sending timed pulses (and software PWM).

The time on Bangle.js comes from the RTC hardware on the chip, which is driven directly from a low speed oscillator. About the only thing that can change the time is if you do a hard reset or run DFU - during that time the RTC won't be running until Espruino boots.

A second or two of drift in smartwatches/PCs/phones is pretty normal I'm afraid - the oscillators are usually rated at around 30 parts per million. Since there are 86400 seconds in a day, 86400 * 30 / 1000000 = 2.5 seconds/day

Usually you don't notice because phones/PCs/etc keep time updated from the internet all the time

Posted at 2022-01-25 by @gfwilliams

Just a quick note on the BLE advertising code above - I didn't spot it at first.

Looks like NRF.setScan(function(d) { is being used. I'd strongly recommend just using NRF.requestDevice as this basically implements the 'stop after 1 packet received' internally, and will behave much better in areas where you'be got loads of BLE devices advertising

Posted at 2022-01-25 by @fanoush

It uses filter for serviceData, also maybe with setScan you get the packet with time sooner just when it arrives so that timestamp inside is more fresh? Was thinking to suggest requestDevice at first too but then was thinking setScan with filter may be better.

Also if you get more packets (from one or more devices) then you have more data to adjust the time.

Posted at 2022-01-25 by Numerist

I'm noticing about 5 seconds' advance per day on Bangle.js 2. You helped me to get GPS time into Bangle.js 1 and read often enough. Is the same possible for Bangle.js 2? Any specifics?

Alternatively, is it possible for Bangle.js 2 to access an iPhone's time via a script for the watch?

Or is there some other way to avoid the large gain per day that others have also noticed?

Posted at 2022-01-25 by Numerist

Sorry, I had forgotten that uploading from the computer resets the time.

Let us know how much your watch gains per day.

Posted at 2022-01-25 by malaire

I just did a 3-hour-test and Bangle.js 2 gained 0.68 seconds in 3 hours, which is about 5.4 seconds per day, so about same what you got.

Posted at 2022-01-26 by @gfwilliams

@fanoush requestDevice works with the exact same filter. It's basically doing exactly what you do with setScan, but inside Espruino - so if anything it should be marginally faster.

Posted at 2022-01-26 by @fanoush

so if anything it should be marginally faster.

the total time may be faster but with setScan you get the packet data sooner? As I understand it the requestDevice sets up a callback to collect data, starts it, collects data, stops it and resolve promise and call your code. setScan first calls your code with the data as the packet arrives and then you do the rest. And with setScan you can get more packets in one scan to do some average on the timestamps and stop when you are happy with the result.

Of course it doesn't matter with precision to seconds but @malaire tried to do more

Posted at 2022-01-26 by malaire

... requestDevice vs setScan ...
Of course it doesn't matter with precision to seconds but @malaire tried to do more

I don't think that matters as there is even worse inaccuracy. I've now realized other 0-20 ms inaccuracy in BLE advertisements I overlooked before: Each advertisement is send on 3 channels with up to 10 ms between, and receiver doesn't know which one they received.

I've been testing to make Bangle.js 2 determine its clock drift and accuracy from advertisements just doesn't seem to be enough to do that (quickly enough).

So next I'm going to test connecting watch to current time device and sending current time via that connection with 1 ms accuracy (custom format).

Posted at 2022-01-26 by @fanoush

I've been testing to make Bangle.js 2 determine its clock drift and accuracy from advertisements just doesn't seem to be enough to do that (quickly enough).

There is still scan response. I will try the hack with SoftDevice >=6.1 with updating timestamp in scan response static buffer on the fly (there is even BLE_GAP_EVT_SCAN_REQ_REPORT event you get when someone requests scan response, see e.g. this) and see whether it breaks or softdevice will pick up the changes.

next I'm going to test connecting watch to current time device and sending current time via that connection with 1 ms accuracy (custom format).

That may be easier and more predictable after all.

Posted at 2022-01-26 by Mark_M

Each advertisement is send on 3 channels with up to 10 ms between, and receiver doesn't know which one they received.

theoretically receiver should know which one they received. Because this reception will happen on a particular channel. I am not sure if that low-level info can be available from the device.
Also I read that a BLE device can be configured to advertise on particular channels only. For example, you can advertise on 37 if (second % 3) == 0, on 38 it is 1 and on 39 if it is 2. So it will cover all channels and have no inter-channel delay. I hope. The price is - overall latency of receiving time can be up to 3 sec. But this time will be more precise.

I will try the hack with SoftDevice >=6.1 with updating timestamp in scan response static buffer on the fly

this is promising approach. I think a checksum needs to be sent as well, because there can be a risk of sending partially updated buffer.

Posted at 2022-01-26 by malaire

BLE device can be configured to advertise on particular channels only

Yes, I read that also, but the Rust library I'm using doesn't seem to support that. (BlueR - official BlueZ library)

Posted at 2022-01-26 by @fanoush

Yes, I read that also, but the Rust library I'm using doesn't seem to support that. (BlueR - official BlueZ library

with the HCI way there is 0x0006 LE Set Advertising Parameters Command with Advertising_Channel_Map parameter with 3 bit array for channels 37,38,39

Posted at 2022-01-26 by malaire

next I'm going to test connecting watch to current time device and sending current time via that connection with 1 ms accuracy (custom format).

That may be easier and more predictable after all.

Strangely this seems to be much worse. Connecting to get few bytes can take several seconds and BluetoothRemoteGATTCharacteristic.readValue() part of that takes at least 90 ms but sometimes up to 300 ms - much worse than using advertisements.

Posted at 2022-01-26 by HilmarSt

Every time you connect to https://banglejs.com/apps/ the watch time is set by that website.

... as long as "Always update time when we connect" is not switched off, like on my PC (I did not do this).

I reset my GPS chip to factory settings with
$PCAS10,3*1F
and still got a time within a second.
I think as long as there is a satelite in view, it will also find the time.

... or the AT6558R has a RTC inside that's not affected by a reset to factory settings and only loses time with a flat battery.

Posted at 2022-01-26 by myownself

I may be misunderstanding, but does the time it takes to get a connection really matter? Surely you connect first, then check and send the accurate time. The time to check and send is the only bit that matters, isn't it?

Posted at 2022-01-26 by malaire

I don't think it matters (except for maybe power usage). But the problem was that the "check and send the accurate time" takes 90-100 ms usually and occasionally up to 300 ms, i.e. it's not accurate at all.

Posted at 2022-01-26 by @fanoush

it may be related to negotiated connection interval, also in BLE this can be done via subscribing to notification, not reading. when you subscribe the other end sends data when it makes sense (temperature changes, in this case seconds value changes) and you just receive it without request

Posted at 2022-01-28 by malaire

If anyone else is using Raspberry Pi as time source, note that by default new Debian 11 -based image uses awful systemd-timesyncd which can make Raspberry Pi have 10x worse clock drift than Bangle.js 2.

You can install chrony for far better time syncing.

Posted at 2022-01-28 by malaire

Now that that is fixed I'm finally starting to get stable clock drift measurements. Current code can measure Bangle.js 2 clock drift in just 5 minutes somewhat accurately, usually giving 61-69 ppm for me, with occasional worse values.

Basic idea:

• Raspberry Pi broadcasts time once per second with microsecond resolution.
• Bangle.js 2 scans for time broadcasts in bursts, getting 4-5 time-deltas in one burst.
  • Largest time-delta is discarded and average of the rest is taken.
• Clock drift is calculated from two time-delta-averages separated by 5 minutes.

Posted at 2022-02-01 by Numerist

[from another thread]
The question is now how to get the watch to use Apple's Time Service to adopt the iOS time at specified intervals. The information I see is some years old or requires a developer's knowledge.

Needs a UUID? NRF input? A little more code?

Can someone here help?

Posted at 2022-02-02 by @gfwilliams

It may be possible to use the JS ANCS code from @jeffmer as a base.

https://github.com/espruino/BangleApps/blob/3e0d263ceaa7f3dd3bdec29fd95822e52f913aba/apps/widancs/ancs.js#L106

Now ANCS is implemented in Espruino, most of the work is done internally and it's just the bit I highlighed you need. It'll need modifying to use Apple's time service UUID, as well as a few other changes.

It's going to be a decent size job though.

Posted at 2022-02-02 by jeffmer

Actually, I have a bit of code that sets the Bangle time using the iPhone Time service. However, I did it as a quick hack from the original ANCS app and it is a bit clumsy as you must first be paired with the iPhone and the routine then disconnects and reconnects so that it can create the needed GATT service. The code is here if you would like to try it.
@gfwilliams - There must be an easier way of creating the GATT service since all the information must already be held internally? I am sure I am missing something obvious!

Posted at 2022-02-02 by @gfwilliams

@jeffmer I notice you've got a getGattforCentralServer in there. Did you ever implement that in your own server?

Right now as I know the JS implementation of Bluetooth Central can only deal with one connection, which is handled with a global m_central_connection variable. As a quick hack I guess we could set that to the same as m_peripheral_connection with a function, but the best solution would be to store the handle in the BluetoothGattServer structure and then we could interface with >1 central device at the same time

Posted at 2022-02-02 by jeffmer

@gfwilliams Yes, I have the function implemented in my Bangle firmware, however, getting time does work if it is not defined. It is a bit frustrating as actually its not dealing with >1 central device, as you are connecting back to to the iPhone you are already connected to for ANCS. If you could get the address of the iPhone you are connected to then something like getGattforCentralServer would let you get the time without disconnecting and reconnecting.

Posted at 2022-02-02 by DDDanny

I can only confirm that the Android/Gadgetbridge TimeSync after connect works like a charm.
Personally, for me, it doesn't matter if the time difference is e.g. 5s per day.

1. I don't use a clock with seconds active
2. The clock reconnects 10 times during day and resync time (because the mobile lays somewhere on desk and i leave the BT radius -> resynced time after within raduis again)
3. If I would need to know the exact time I took a look at the mobile or some other ntp time-synced device

Posted at 2022-02-03 by Numerist

@jeffmer
Thanks for that help. I'm not a programmer, so don't understand much of your discussion with Gordon. I've placed your code before the code for my watchface, loaded the lot onto my watch, then cut that connection and paired with my iPhone. If something different or more needs to be done, just let me know.

The left side of the IDE delivers these apparently non-fatal messages on loading your code with mine:
WARNING: SD ERR 0x11 (BUSY) (:1803)
WARNING: SD ERR 0x11 (BUSY) (:650)

Anyhow, how often (at what interval) does your code make the watch's time the same as the iPhone's time?

(I'll see how the experiment's gone in 12 hours.)

Posted at 2022-02-03 by jeffmer

The code synchronises time once when you invoke
setTimefromPhone. I have not seen those errors before but they look like they are coming from the Soft Device which is the Nordic Bluetooth code.

Posted at 2022-02-03 by @gfwilliams

Just a note that on the cutting edge firmwares I added a field to NRF.getSecurityStatus which tells you the currently connected device's address, which should make things easier @jeffmer :)

Posted at 2022-02-03 by Numerist

@jeffmer
As expected, what I did makes no difference. I assume that I have to call setTimefromPhone in my code and make sure nothing interferes with that. I have my own font size, colour, etc., and I'll have to see whether my determination of the time interferes or may preferably be overridden by yours. (Someone else wrote my code but has left the project.)

If your code synchronizes only once, is there an easy way to get it to do that 4 or 24 times a day, for example? Just call the function at the desired times?

I'd be grateful for any further advice, including whether Gordon's update changes what I may or need to do.

Posted at 2022-02-03 by Mark_M

An easy and universal way to make regular updates is to create a widget that will sort of a backgroung job. It will get data from a time server and set the watch RTC. Then any watchface will get the correct time.
Probably you do not have to even to create it. As DDDanny wrote
that the Android/Gadgetbridge TimeSync after connect works like a charm.
Some people here is working on a connectionless (ads based) way of time synchronizing with jitter < 10ms. Is it what you need?

Posted at 2022-02-03 by malaire

... the Android/Gadgetbridge TimeSync after connect works like a charm.

no, that is awful: clock error can be over 3 seconds after Gadgetbridge sync

Posted at 2022-02-03 by Numerist

@Mark_M
Like some other people here, I'm interested in whatever will work with an iPhone. (Android and Gadgetbridge are not possibilities.)

Some people here is working on a connectionless (ads based) way of time synchronizing with jitter < 10ms.

Which devices will provide the synchronization?

Posted at 2022-02-03 by malaire

Which devices will provide the synchronization?

Using Bluetooth LE ads should in theory work on any device which can send those ads. However I only have example code for Linux (tested in Debian 11 and Raspberry Pi OS) as I don't know how to use Bluetooth LE on other devices.

Posted at 2022-02-03 by jeffmer

Thanks Gordon, I will see what happens if I use that instead of disconnecting and reconnecting.

@numerist Better to wait to see how this works out as it would make the operation much more lightweight.

Posted at 2022-02-03 by malaire

This is a fun project, now I'm trying to implement temperature-corrected clock drift. :)

This graph shows clock drift on my Bangle.js 2 in PPM in different temperatures (Celcius).

Attachments:

• clock-drift-in-ppm-by-temperature.png

Posted at 2022-02-03 by Mark_M

Cool!
This function can be applied if there is no time service available, no GPS, no BT, no GadgetBridge.

Also does it show the watch RTC oscillator is not T compensated (not TCXO)?

Posted at 2022-02-03 by myownself

This is really interesting! Guess it will vary somewhat from watch to watch?

On a tangentially related note, I was looking at the two temperature sensors earlier (barometer and internal temperature sensor), and I was surprised that they read pretty much the same. I don't know why I have in my head that another component might have a temperature sensor built in but not exposed... I might be misremembering something though.

Posted at 2022-02-03 by malaire

@Mark_M

Also does it show the watch RTC oscillator is not T compensated (not TCXO)?

yep, TCXO would be much more stable over temperature range.

@MyOwnSelf

This is really interesting! Guess it will vary somewhat from watch to watch?

Most likely, and this article says that crystals have aging effect so it also depends on watch age.

... two temperature sensors earlier (barometer and internal temperature sensor) ...

What internal temperature sensor ?

Posted at 2022-02-03 by myownself

https://www.espruino.com/Reference#l_E_getTemperature

This one.

Use the microcontroller's internal thermistor to work out the temperature.

Posted at 2022-02-04 by Poolitzer

Most likely

Want me to test that? I have a raspberry pi 3B+.

Would be much appreciated if you can provide a step by step guide to set the measuring up. I can probably gather it from the thread but it will take quite a bit of time.

Posted at 2022-02-04 by malaire

That would be nice. Is it OK if I just give the compiled binary with installing instructions? (I could also give cross-compile instructions but that requires installing Rust etc. on separate computer and has only been tested in Debian 11.)

Also this may require current (Debian 11 based) Raspberry Pi OS. I'm using the 32-bit OS currently.

Posted at 2022-02-04 by Poolitzer

Yes, current 32 Bit raspi os here as well. I am fine with the compiled binary, nothing on it which can be abused.

Posted at 2022-02-04 by malaire

ok, here are quick instructions:

Get ble-current-time-broadcast.bz2 and ble-current-time-broadcast.service
from release 2022-02-04.

You can just run extracted ble-current-time-broadcast directly in terminal, or if you want you can install it as service:

; much better time-syncing than default systemd-timesyncd
sudo apt-get install chrony

bunzip2 ble-current-time-broadcast.bz2
sudo cp ble-current-time-broadcast /usr/local/bin
sudo chmod +x /usr/local/bin/ble-current-time-broadcast
sudo cp ble-current-time-broadcast.service /lib/systemd/system
sudo systemctl daemon-reload
sudo systemctl enable ble-current-time-broadcast
sudo systemctl start  ble-current-time-broadcast

If you didn't have chrony installed, then you should probably wait some time for it to stabilise.

Then upload ppm_logger.js to Bangle.js 2, it will show one line per 10 minute interval (by default) with average/min/max temperatures, interval length in seconds and measured clock drift in PPM.

Some variance in PPM is expected especially if using default 10 minute interval, so I usually let it run for some time.

Here's what it looks like for me currently (I should probably add logging to file also):

Attachments:

• ppm_logger.png

Posted at 2022-02-04 by Numerist

I assume (as a non-programmer) that the recent (interesting) posts aren't about trying to get the time from an iOS device. Do correct if that's wrong.

Posted at 2022-02-04 by Mark_M

this is for Linux

Posted at 2022-02-04 by Poolitzer

I followed your steps, they worked perfectly. Here are my numbers.

Attachments:

• download.png

Posted at 2022-02-05 by malaire

Yep, that's clearly different. My Bangle.js 2 has about 63 PPM at 19 C and yours is around 71 PPM.

btw, the graph I showed earlier was made with LibreOffice Calc: I measured some temperature/PPM pairs, made a chart of them in LibreOffice Calc and added trend line (polynomial, degree 2).

Posted at 2022-02-07 by Numerist

@jeffmer
Any luck with Gordon's fix for using the iOS time?

Posted at 2022-02-08 by jeffmer

Sadly, you still need to disconnect and reconnect to set things up properly. I think it would need to be done in the firmware to work efficiently.

Posted at 2022-02-08 by @fanoush

Sadly, you still need to disconnect and reconnect to set things up properly.

So you did try without disconnect and it does not work? Iphone refuses connection from Bangle when it is already connected to it in opposite direction? I was thinking ANCS works in same way. Normally Espruino can do connection while being connected to, never tried with same device though. Will test this between two nrf52 espruino devices, both connecting to the other one while having console on serial.

Posted at 2022-02-08 by jeffmer

Yes, I think it is more to do with the internal Gatt structure not getting set up properly. I have not fully debugged as I only tried it on a Bangle - need to try it on a dongle to see all the error messages.

Posted at 2022-02-08 by Numerist

Does that mean disconnecting and reconnecting only once? Or how often?

(A further question: how much of your code is needed, and how should that be used for a given watch script?)

Posted at 2022-02-24 by Numerist

I assume from the above multiple discussions of some weeks ago that the Bangle.js cannot use Apple's iOS time service in an efficient and reliable manner to regulate the watch's timekeeping via an iPhone. If that's incorrect, it would be helpful if someone would elaborate.

Meanwhile, other ways of regulation are being created (not by me).

Posted at 2022-02-24 by jeffmer

I think its fine if you want to synchronise with the iPhone once a day, however, at the moment, as it seems to involve disconnection/reconnection, it would not be advisable to do it at a much greater frequency.