DM42- clock losing time

Post here to share useful tips and tricks, to ask questions about using your DM42 or to report software-related problems
toml_12953
Posts: 637
Joined: Wed May 03, 2017 5:46 pm
Location: Malone, NY USA

Re: DM42- clock losing time

Post by toml_12953 » Wed Jul 29, 2020 3:35 pm

whuyse wrote:
Wed Jul 29, 2020 7:51 am
Regarding the RTC correction: if your DM42 clock is fast (i.e. it shows a later time), should the factor C be positive or negative? It's not clear from the description, so I have a 50/50 chance of being right, meaning I'll be wrong ;-)
Werner
LOL! Even with only two choices, it takes me three tries to get a USB device plugged in right!
Tom L

I bought a moving blanket but it just sits there.

DM10L SN: 059/100
DM41X SN: 00023 (Beta)
DM42 SN: 00025 (Beta)
DM42 SN: 00221 (Shipping)

NoisyBoy
Posts: 5
Joined: Thu Jun 25, 2020 5:49 am

Re: DM42- clock losing time

Post by NoisyBoy » Tue Aug 04, 2020 5:05 am

I wonder if there's an actual conversion factor between the RTC correction value and the intended/actual frequency discrepancy of the crystal? I have a 12 digit counter with a GPSDO reference at my home lab that I use to tune RF equipment and OCXOs to well under 1 ppb error. If I can dial in the proper correction value, it would be interesting to see what the crystal is capable of when it comes to absolute accuracy.

grsbanks
Posts: 932
Joined: Tue Apr 25, 2017 9:23 am
Location: Preston, Lancs, UK
Contact:

Re: DM42- clock losing time

Post by grsbanks » Tue Aug 04, 2020 6:56 am

When all else fails, there's always the manual, which tells you how to go from a ppm drift to a correction factor:

https://www.swissmicros.com/dm42/doc/dm ... correction
There are only 10 kinds of people in the world: those who understand binary and those who do not.

x55
Posts: 9
Joined: Sun Jul 26, 2020 5:32 am

Re: DM42- clock losing time

Post by x55 » Tue Aug 04, 2020 2:30 pm

toml_12953 wrote:
Wed Jul 29, 2020 3:35 pm
whuyse wrote:
Wed Jul 29, 2020 7:51 am
Regarding the RTC correction: if your DM42 clock is fast (i.e. it shows a later time), should the factor C be positive or negative? It's not clear from the description, so I have a 50/50 chance of being right, meaning I'll be wrong ;-)
Werner
LOL! Even with only two choices, it takes me three tries to get a USB device plugged in right!
In Type A, mini and micro connectors the USB symbol should be imprinted on the moulding and face the user when the device is in its normal operating position.

The DM42 gets that right but most Chinese made consumer devices, except Huawei and Xiaomi, get it back to front. I have no idea what the Chinese translation of the spec says and it seems as though the Chinese don't either.

In the case of a Type A receptacle in the front panel of a desktop tower or rack mounted device I think it depends on the casing designer's judgement of the average viewing angle of the average user sitting or standing averagely.

The only Korean (Philippines-made) device I have has a Type C connector, which is of no help at all or a great help altogether, not depending on which way you are looking at it.
--
HP: 55>67>41C>HP-IL+>+ProtoCoder>41CL>+MLDL>41CX>16C>71B>42S>33s**>35s ...
DM: 42 ...
Etc: If a calculator has equals, why buy one unless it's the cheapest?
**: Carly Failurina model

NoisyBoy
Posts: 5
Joined: Thu Jun 25, 2020 5:49 am

Re: DM42- clock losing time

Post by NoisyBoy » Tue Aug 04, 2020 10:35 pm

Just perform a measurement on the crystal on board the DM42, the spec frequency is 32.768 kHz. I tested its output on my counter, the crystal is stable to the 1Hz, but digit below the decimal point is not stable.

I had the counter collect 100 samples at full resolution, the average is 32,767.548954 Hz, with a 0.095852 Hz standard deviation. It translates to a -13.765 ppm error, with an annual time lost of 434.09 second. For those of you who collects Swiss watches, it is within the COSC spec for mechanical watches.

Applying the correction equation, C should be -14.4338 in my case, rounding to the nearest integer, I will dial in a correction factor C of 14 to optimize its time keeping capability.

grsbanks, thank you so much for pointing me to the manual, I should have looked at it first before asking.

It is a datapoint of one unit, but I hope it gives some idea of the type of correction needed in my case. My hobby lab temperature is 74F, so temperature should not be an issue for the frequency deviation, and the crystal is within the 20 ppm spec. But as long as there is a way to compensate for it, we can all improve on its time keeping ability.
Last edited by NoisyBoy on Thu Aug 06, 2020 4:23 am, edited 2 times in total.

whuyse
Posts: 64
Joined: Thu Dec 21, 2017 12:23 pm

Re: DM42- clock losing time

Post by whuyse » Wed Aug 05, 2020 7:07 am

So, your clock is slow, and you apply a negative C. Let me know if that is correct;-)
Werner
42S #3249S01123
DM42 #00345

NoisyBoy
Posts: 5
Joined: Thu Jun 25, 2020 5:49 am

Re: DM42- clock losing time

Post by NoisyBoy » Wed Aug 05, 2020 7:41 am

Good catch, Werner.

I meant C=+14

I set the time with my lab master reference clock, that clock is accurate to <1 ps error. Unfortunately, I had to set it by hand as I don’t know a way to do a programmatic sync. My guess is RTC correction value is intended to compensate crystal error to 0.5 ppm - under ideal theoretical case of course, as there are a multitude of factors that can change its frequency. If that is the case, and it is working as intended, the best case is time deviation of 1.4 sec/month.

As others have pointed out earlier, a crystal that has an ever changing orientation, constant vibration, constant temperature variation, is not going to get anywhere close to the theoretical result.

I will let you know how this work out. I hope resetting the clock in the setting menu is indeed all it takes for the DM42 to pick up the correction value.

whuyse
Posts: 64
Joined: Thu Dec 21, 2017 12:23 pm

Re: DM42- clock losing time

Post by whuyse » Thu Aug 06, 2020 6:50 am

Nono, it was a *question*! I have no idea what the sign of C should be in the RTC correction file.
So let me (us) know whether your correction for a slow clock works.
Cheers, Werner
42S #3249S01123
DM42 #00345

ljubo
Posts: 4
Joined: Sat Jun 20, 2020 11:46 am

Re: DM42- clock losing time

Post by ljubo » Thu Aug 06, 2020 7:04 am

It's more important to give the option the right name - e.g. call it "Clock: Add sec per week" or similar - sign becomes clear then.

NoisyBoy
Posts: 5
Joined: Thu Jun 25, 2020 5:49 am

Re: DM42- clock losing time

Post by NoisyBoy » Thu Aug 06, 2020 7:32 am

Hi Werner,

No, it does not seem to work. The instruction is very vague, I am reversing the sign and will check in a couple days, I am currently gaining around 2 seconds after 24 hours. The following from the manual:

"Note that to apply RTC correction you have to know (measure and calculate) ppm drift of calculator clock first.
Use following steps to apply clock frequency correction.
Calculate correction factor C which best matches required ppm correction P"

I know my crystal is slow, therefore I would define the "drift" on line 1 as a negative number (-13.765 ppm).
On line 3, they refer to "required ppm correction P". I have no idea what this means, should this correction be what it takes to compensate for the "drift" - thus +13.765 ppm? Or are they referring to the "drift" value?

I hope to find out more and see how the sign in P relates to the actual drift by check the clock against my reference clock in a day or two.

Post Reply