Capacitance of glue gun glue

[kf0rt]

Only semi-joking...

We've had a few power outages here lately and it's a PITA to change the clocks, especially my homebrew LED clocks. Got to looking around and found the Dallas/Maxim DS1302 clock chip. This thing is way cool. Tiny clock chip that interfaces easily to a PIC processor with two parts: a 32.768KHz watch crystal and a supercapacitor. It even charges the supercap. Power dies, the supercap keeps the clock running. Power draw is in the microwatt range, so the supercap should keep things running for days if not weeks. Power comes back on, clock has been running the whole time and it sets itself. I'm using a .22 Farad cap that's smaller than 3 stacked dimes. All 3 parts cost about $5, total.

(It's JUST a hobby!)

So, I added the circuit to my clock PCB and am now playing with the software.

Thing is, the watch crystal is about the size of a grain of rice, so it's pretty fragile. I'd like to glue it in place, but don't want to affect the capacitance which is in the 6pF range. Guess I could bend it over and keep the glue to the case. Maybe a small drop of superglue would do the trick.

...jes' havin' fun. :)

-ØRT

Yeah, if you just glue the case to the PCB, there will not be much change in capacitance.

We make some tuning fork 32.768 kHz crystals, but they are larger than what you say yours is. Those tiny crystals usually exhibit larger crystal resistance and they become more difficult to make to oscillate. Not a problem if you have enough stage gain.

Sounds like a fun project. I have always enjoyed time and frequency projects. I used to work at Austron (Datum). We had clocks accurate to within 4 nS of UTC. Close enough for government work, I guess.

[N1LAF]

Check for the glue conductivity (or inversely resistance). It would have to be conductive to have capacitive effect.

[N3ATS]

Hot glue is not conductive. I've used it as an insulator/strain relief before.

http://www.hardforum.com/archive/index.php/t-713664.html

Here is another discussion on the subject. Heh heh. One guy is worried about the temperatures.

At any rate, the circuit needs to be adjusted after the build is through to take care of all of the stray capacitance and inductance in the circuit. There is little way around this.

[kf0rt]

Neat stuff, Jerry...

I don't guess I'll worry too much about the crystal accuracy. The clock runs off a 1ms interrupt on the PIC (clocked at 20 MHz with a 20ppm oscillator). I originally "hand tuned" the internal dividers to be within a few seconds a week. The new clock chip wants a crystal with a 6pf load capacitance and the closest I could find was 8pf -- seems like most watch crystals are 12.5pf. So, it's bound to be off a little.

With the same 20ppm accuracy, the app notes say the watch stuff should be good to about a minute a month. Yet, I know most digital watches are much more accurate than that.

Anyway, got it mostly working today and it's really weird to plug the clock in and have it fire up with the right time. I'm going to play with it and see which is more accurate, the old way or the new chip. If the new chip is more accurate, I'll sync the clock to that on a regular basis and have a more accurate clock. If the old way works better, I'll reprogram the clock chip on a regular basis so it doesn't get too far off.

Hafta see if I can get a picture of the new board...

[kf0rt]

Here's some pics:

This is the crystal (X-Acto blade for size reference):
http://tatanka02.home.comcast.net/~tatanka02/crystal.jpg

The corner of the board with the DS1302, crystal and cap. Pretty sure that's a dog hair above the board. One screwup (but not too bad). I wasn't paying attention when I ordered the chips, so what I got was DIP-8 packages. These wouldn't work at all, so I reordered what I thought were SOIC-8's. Oops again -- got 8-SO's. They just barely fit. Never even heard of that package before.
http://tatanka02.home.comcast.net/~tatanka02/board.jpg

It's impossible to get a good picture of the clock itself. It looks much better in person:
http://tatanka02.home.comcast.net/~tatanka02/clock.jpg

I've got two of these -- this is the original prototype. They're 12" x 12" so us old geezers can see 'em easily from across the room with our glasses off. The other one looks a lot better with blue and white LED's.

Nice job on the PCB.

Real nice. Very professional. In high school, I knew a kid who let his novice lapse. WN0ORT, and one guy in Parsons got his initials W0RT. He is an attorney.

I have some chip caps if you want me to send some to you.

[kf0rt]

Thanks! I love the PCB work -- been using ExpressPCB and their tools / service. There's a lot of incentive to get it right on the first try to avoid wasting $60. I spend a lot of time looking at the mechanical drawings on parts, and still usually miss something.

No worries on the chip caps, but thanks -- other than the supercap, there are only 2 caps on the board (.01 bypass on the processor), and those aren't really even needed.

[WV6Z]

BEAUTIFUL! Now I have to try to figure out why all of my PCB projects look like I was on crack AND used a 20lb. sledge hammer to put them together in comparison. :-?

[HUGH]

I had to check, professionally, types of hot-melt glue for another purpose and they seem to consist mainly of polyolefines, wax, polyethylene and, with some high-temperature ones, nylon.
Apart from the heat aspect, I shouldn't think a low-frequency crystal would be affected.

In the event, we decided to use a variation of superglue for the particular job, it ends up as an acrylic blob. Other than that I've used tiny pieces of a quality double-sided sticky tape. With the tape you can prise the component off.

[N3ATS]

BEAUTIFUL! Now I have to try to figure out why all of my PCB projects look like I was on crack AND used a 20lb. sledge hammer to put them together in comparison. :-?

You gotta stop using a railroad spike for a soldering iron! :mrgreen: