mitxela.com forum
Welcome. Please log in or register.

Precision clock showing occasional corruption
laney Posted: 3 Sep 2020, 12:09 PM
Avatar


Member
Posts: 3
Joined: 3-September 20
I've seen a couple of threads about similar issues but the indication in there was that these were fixed with newer kits, which I assume mine is (received a few weeks ago).

The clock has GPS lock just fine. It hangs out by the window and that seems like a good location GPS-wise. Similar to those threads, sometimes I'll catch it garbling its display. I think I've only seen it do this at night, so perhaps that lines up with some theories about temperature? Eventually it sorts itself out.

Short video of it happening last night

Would be interested in any ideas!

Last edit by laney at 3 Sep 2020, 12:09 PM

-------------
[top]
mit Posted: 3 Sep 2020, 01:09 PM
Avatar
yeah whatever

Admin
Posts: 283
Joined: 4-May 16
Ah, how annoying! I am sorry about that.

This is going to sound silly but it might simply be that the chip was calibrated and tested during the heatwave a few weeks ago. It was well over 30°C in my lab for several days. I didn't think this would significantly affect the calibration, since this graph from the datasheet suggests the slope is fairly gradual:

(User posted image)


The UART should be fine if the oscillator drifts up to almost 10%. From that graph, it would need to swing about 25 degrees to move that much.

It's possible there's some other factor that's making it more sensitive that I hadn't considered, some chips have a digital lowpass filter that might tighten the tolerances a bit. There is also some amount of quantization introduced by the calibration method that might be significant. I will look into it.

In the mean time I'll send you another chip and hope that fixes it.

I concede that the reliance on the internal oscillator of the attiny chip is a design flaw. Maybe a firmware update could make the chip self-calibrate from the GPS signal...

-------------
[top]
laney Posted: 5 Sep 2020, 11:09 AM
Avatar


Member
Posts: 3
Joined: 3-September 20
Cheers. Kind of amusing, in its own way.

I’d be also happy to try to fit a crystal oscillator, but you’ve easily managed to step outside my comfort zone - the brief instructions on the assembly page are not detailed enough for me. So if you’re willing to tutorialise those just a bit more, that could be a go-er.

-------------
[top]
mit Posted: 5 Sep 2020, 12:09 PM
Avatar
yeah whatever

Admin
Posts: 283
Joined: 4-May 16
I'll try and flesh out the instructions for that at some point. I wrote it with the mindset that fitting a crystal would be to improve the accuracy of the clock under poor GPS reception, which would take some software changes to make it worthwhile.

If it's just to fix this data corruption issue, fitting the crystal is quite straight forward. You wouldn't need to fiddle with different capacitors, those are for adjusting on the parts-per-million scale. The only tricky part is telling the chip to make use of the crystal, which does require a programming cable. It would be the same instructions as changing the timezone, except instead of reflashing the chip, you set the configuration fuse bytes to use the external oscillator.

-------------
[top]
laney Posted: 28 Sep 2020, 07:09 AM
Avatar


Member
Posts: 3
Joined: 3-September 20
Just to close this out for the benefit of the internet (who likes unresolved threads?)

mit sent me a new ATtiny. As predicted, it worked fine.

But my curiosity was piqued by the mention of an external crystal and I wanted to see if this would fix the original chip. I read around on various other internet forums about these and how they work. I tried to use an oscilloscope that the local hackspace has in order to help me select the correct load capacitors, but I was too incompetent to be able to use it correctly :-), so I basically ended up using the equation blind. I’m told it doesn’t matter too much anyway.

Reprogramming the chip to set the fuse bits is quite simple, if you have a programmer and a bread board.

We’ve had a few chilly nights now, and I’ve not seen any corruption!

-------------
[top]

Sign in to post a reply.