View previous topic :: View next topic |
Author |
Message |
dluu13
Joined: 28 Sep 2018 Posts: 395 Location: Toronto, ON
|
|
Posted: Wed Sep 02, 2020 2:00 pm |
|
|
I am now using this crystal:
https://www.digikey.ca/en/products/detail/abracon-llc/ABS07AIG-32.768KHZ-D-T/5817610
This is what we used before:
https://www.digikey.ca/en/products/detail/epson/FC-135-32-7680KA-A5/5259895
Now, they both don't work.
Although there are traces, right now they are not doing anything. Even in the future, they are lines that switch states once very few several minutes at the most frequent. Do you mean they can act like antennas and cause issues?
I agree with the north south orientation. I am not sure why the original designer put them east-west but there is a common ground via in the orientation shown now, just not in that picture.
I still don't really understand why it worked before but not anymore... The layout in that area hasn't changed.
EDIT: I've got half a mind to get a clock oscillator module. Can't hurt to tack a couple on my trial parts order, just to see. |
|
|
PCM programmer
Joined: 06 Sep 2003 Posts: 21708
|
|
Posted: Wed Sep 02, 2020 4:37 pm |
|
|
What about trying a different PIC family, just to check that if that's the problem ? Add one to your parts order (with compatible pinout). |
|
|
temtronic
Joined: 01 Jul 2010 Posts: 9097 Location: Greensville,Ontario
|
|
Posted: Wed Sep 02, 2020 5:18 pm |
|
|
hmm... more straws to pull at....
Has the power supply design/parts/layout changed at all ? Maybe bypass caps are marginal ? Wallwart 'slewrate' worse than before...triggers a 'glitch' that puts the PIC into 'weird mode' ?
I'm assuming you've got a good board next to bad board on the bench ,so some 'EMI' issue isn't the problem ?? Grounds are all the same ??
I know kinda basic stuff but..sometimes... it's the thing we've looked at 100 times and never SEEN it was wrong.......like fixating that 102 means a 10K resistor ?
hay, at least the humidity has gone for awhile !!!
Jay |
|
|
dluu13
Joined: 28 Sep 2018 Posts: 395 Location: Toronto, ON
|
|
Posted: Thu Sep 03, 2020 6:39 am |
|
|
temtronic wrote: | hmm... more straws to pull at....
Has the power supply design/parts/layout changed at all ? Maybe bypass caps are marginal ? Wallwart 'slewrate' worse than before...triggers a 'glitch' that puts the PIC into 'weird mode' ?
I'm assuming you've got a good board next to bad board on the bench ,so some 'EMI' issue isn't the problem ?? Grounds are all the same ??
I know kinda basic stuff but..sometimes... it's the thing we've looked at 100 times and never SEEN it was wrong.......like fixating that 102 means a 10K resistor ?
hay, at least the humidity has gone for awhile !!!
Jay |
Power supplies: Only change is I eliminated a boost converter to 24V off of this board.
Caps and rails: 3V3 looks steady on the scope. Besides, LF internal oscillator seems to work ok, as well as the HF internal oscillator. CLKO pin gives a steady 16 MHz.
Wallwarts: Not using them. For now I am powering it off of a 12V B&K Precision bench supply.
EMI issue: I do have a "good" board here and it's working just fine while the "bad" board does not.
I appreciate the thoroughness.
Humidity: I know eh? I had to wear a sweater on my way to work this morning.
On the other hand, I powered the thing up this morning and it's working... on 2 of the 4 boards.
One board with the parallel resistor and one without are working.
One board with the parallel resistor and one without are NOT working. |
|
|
temtronic
Joined: 01 Jul 2010 Posts: 9097 Location: Greensville,Ontario
|
|
Posted: Thu Sep 03, 2020 7:33 am |
|
|
yeah cold AND dark when I got up.....sigh...
though I did see an ultralight fly by 1/2 hr ago....
OK nuther silly idea..... are ALL the PICS from the SAME batch ??
hmm, can you put a 'bad' board in a toaster oven for 1/2 hr, LOW heat to 'bake the board'. Maybe it is a board moisture issue ??
are they double sided PCBs ?
I'm assuming the PIC is soldered and not socketed ?
sadly, the hay bale's getting smaller...you're making me THINK as to WHY this is happening.
Jay |
|
|
dluu13
Joined: 28 Sep 2018 Posts: 395 Location: Toronto, ON
|
|
Posted: Thu Sep 03, 2020 7:47 am |
|
|
temtronic wrote: | yeah cold AND dark when I got up.....sigh...
though I did see an ultralight fly by 1/2 hr ago....
OK nuther silly idea..... are ALL the PICS from the SAME batch ??
hmm, can you put a 'bad' board in a toaster oven for 1/2 hr, LOW heat to 'bake the board'. Maybe it is a board moisture issue ??
are they double sided PCBs ?
I'm assuming the PIC is soldered and not socketed ?
sadly, the hay bale's getting smaller...you're making me THINK as to WHY this is happening.
Jay |
Hey the solstice was over two whole months ago we knew it was going downhill from there :P
PICs on the old and new board are from different batches. I did not bake them...
Just flipped the boards over to my coworker with the oven... we'll see what happens.
Single sided PCBs and the PICs are surface mount (80 pins whew). |
|
|
Ttelmah
Joined: 11 Mar 2010 Posts: 19215
|
|
Posted: Thu Sep 03, 2020 8:40 am |
|
|
How are the boards being cleaned after soldering?.
If the flux is not being stripped, this could take the capacitances up by
an order of magnitude.
My capacitance figures were for rather long tracks and through hole. The pin
capacitances of surface mount parts are way lower. |
|
|
dluu13
Joined: 28 Sep 2018 Posts: 395 Location: Toronto, ON
|
|
Posted: Thu Sep 03, 2020 8:43 am |
|
|
I am cleaning them with an MG Chemicals flux remover. It appears to be acetone based. While I am soldering, I also brush away excess flux residue with alcohol from time to time.
After some consideration, I am thinking I should try using a MEMS oscillator... It appears to eliminate this tuning. Also, the end product will not exactly be outside, but still will be sitting inside shipping containers and operating in the cold Ontario winters... Any thoughts about this? Like I said yesterday, I already ordered some SiTime 32.768 kHz oscillators. I should probably get one for the instruction clock too. |
|
|
dluu13
Joined: 28 Sep 2018 Posts: 395 Location: Toronto, ON
|
|
Posted: Thu Sep 03, 2020 9:23 am |
|
|
My goodness
After baking them all for half an hour at about 60 degC, all of them are oscillating. I'll need to keep an eye on these. I did go over them with a heat gun before in order to reflow all of the pins but I guess the slow heat got all of the last droplets of liquid out.
But after reading about these MEMS oscillators stability in wider temperature ranges, I'm still going to give them a good look considering my application. |
|
|
Ttelmah
Joined: 11 Mar 2010 Posts: 19215
|
|
Posted: Thu Sep 03, 2020 10:05 am |
|
|
Modern boards can suffer massively from moisture.
The switch in PCB manufacturing to using water as the main solvent can
cause innumerable issues. PCB's are quite absorbent, and when they spend
several hours immersed in baths of various sorts, this gets absorbed.
Glad you have a solution, and now need to think of the best answer for
long term use....
Your wide environmental range makes a Mems oscillator potentially a
better solution. Also you can potentially gain a pin, since you only need
the clock input, not the two oscillator connections.
I use a Mems unit on the clock of my current project, and with a bit of
care selecting the unit used, you can significantly exceed the likely accuracy
of a normal crystal unit. |
|
|
dluu13
Joined: 28 Sep 2018 Posts: 395 Location: Toronto, ON
|
|
Posted: Thu Sep 03, 2020 10:09 am |
|
|
I picked a SiTime one. One other big plus is how simple the MEMS unit seems to be to use. It looks like I only need to provide power and ground, and then it will output a clock signal.
I just need to change a couple of fuses now and it should slot right in. Theoretically hahah... |
|
|
Ttelmah
Joined: 11 Mar 2010 Posts: 19215
|
|
Posted: Thu Sep 03, 2020 12:30 pm |
|
|
They are that easy to use.
The SiT1572, is I think the most accurate 32768Hz unit. |
|
|
dluu13
Joined: 28 Sep 2018 Posts: 395 Location: Toronto, ON
|
|
Posted: Thu Sep 03, 2020 12:33 pm |
|
|
I got the SiT1630 because it comes in that nice SOT32-5 package.
And you're right. I literally just put the chip onto a SOT32-5 to SOIC breakout, plugged it into the breadboard, patched some wires around and I'm getting the 32.768 kHz wave into my PIC. All I had to do was change one fuse. |
|
|
Ttelmah
Joined: 11 Mar 2010 Posts: 19215
|
|
Posted: Fri Sep 04, 2020 12:38 am |
|
|
Yes, easier to work with than the package for the other unit.
You'll find that the D variant will give better accuracies, whatever temperature
range you actually use. Because it is specified to remain within tolerance
over the wider temperature range, these tend to have better initial
tolerances as well... |
|
|
dluu13
Joined: 28 Sep 2018 Posts: 395 Location: Toronto, ON
|
|
Posted: Fri Sep 04, 2020 6:53 am |
|
|
You guys have done it again!
This isn't the first time when I have come here with a problem and left with a solution on a totally different tack :P
I'm grateful for how the members on this forum here are helpful, knowledgeable, and patient. |
|
|
|