How sensitive is an RC clock over temp ?

iso9001 · Joined: 02 Dec 2003 Posts: 262

I have a system that is running at 1Mhz using a 100pF and 10k...

It needs to be accurate enough to send out a 50us and 100us pulses every now and then, even at tempurates from 0C to 70C... am I expecting too much ? I have it working at 1Mhz at roomtemp right now.

I want to bump that up to 2Mhz and I know the forumla is T = R * C then F = 1 / T...

But should I try and use a small cap and a larger resistor or a larger cap and smaller resistor or does it matter ?

Which system will give better tempurate stability ?

I couldn't find a chart in the datasheets that had RC freq over temp, does anyone have one ?

(PS: I KNOW I'de be better off with crystal but I dont have enough room on my board for that AND the caps and I dont think I have time to order new parts in)

PCM programmer · Joined: 06 Sep 2003 Posts: 21708

You didn't say what PIC you are using.

I tried to find out by looking at some of your old posts,
but then I find that you are erasing all your posts in
some threads. Why ?
http://www.ccsinfo.com/forum/viewtopic.php?t=20403
http://www.ccsinfo.com/forum/viewtopic.php?t=20456

iso9001 · Joined: 02 Dec 2003 Posts: 262

PIC 16F872 and F876A

(About the posts: My little sh�t brother HAD my password for everything. I was lucky, in my other forums he went on about my girlfriend and my supposed boyfriend (i had no idea about that Confused

)

I was thinking about 10pF + 50k, and/or 100pF + 5k

newguy · Joined: 24 Jun 2004 Posts: 1911

Make sure to use NP0 caps (negative positive zero). They're made of material that has a zero (nearly zero) temperature coefficient. By using them, you can assure that the caps won't be a source of drift.

bdavis · Posted: Tue Sep 21, 2004 10:27 pm

Also, don't use carbon resistors - at least use metal film, and you can get special low noise and low temp coef. resistors, but typically have to special order them.

At least use NPO caps, but I think you can get better ones than that also...

If you know your temp deviations, you may also be able to correct it real time in FW (since temp changes slowly)...

iso9001 · Joined: 02 Dec 2003 Posts: 262

I have NPO (5% I think) and think and thin metal film resistors here, I'll use those.

Any one have a chart of what I can expect for variations ? I suspect this unit will be at 120 degree F during most on its ON time. I cant find a chart for RC tollerance anywhere, Confused

bdavis · Posted: Wed Sep 22, 2004 12:31 pm

I would recommend making your own chart - measure at ambient temp (25 degrees C), then at min temp, and finally at max temp.

That would give you a fairly good idea of typical. For maximum, you will need data sheets for the components you are using and calculate it.

bdavis · Posted: Wed Sep 22, 2004 12:35 pm

I would recommend not going below 100pf, because it won't help. The error is more from the type of cap, not the size of the cap. If you go too large for the cap, then leakage can be a source of error, but I would think 100pf should be fine...

drh · Posted: Wed Sep 22, 2004 2:01 pm

The data sheets for both devices state that the RC osc. frequency is a function of supply voltage, the resistance and capacitance values and tolerances, and the operating temperature. Frequency can also vary from unit to unit and different package types also.
If you want accuracy and repeatability of the 50uS and 100uS pulses, I would recommed you use ceramic resonators or xtals for the oscillator.
The RC option is really meant for lowest cost, no precise timing needed applications.
_________________
David

Humberto · Posted: Thu Sep 23, 2004 3:23 pm

I wouldn�t take the risk using an RC oscillator for such a wide temperature range.
The cost difference between R+C compared with a Xtal would be cost sensitive
only if your application is going to be sell by millions units, (I wish that for you !! Very Happy

)
meanwhile I will use a Xtal.

Best wishes,

Humberto