int RC clock good enough for serial comms?

[Don P]

Searching the archives didn't net me any subject lines that looked relevant.
On the 16F628 the internal clock is nice because it frees up board space & I/O pins, and works fine for many apps without strict timing requirements. But I recall having problems doing 9600 baud serial comms when using it. Now as I look at the 628 manual, I can't find any spec that indicates that there might be clock accuracy problems. i.e. they show only a 4MHz "typical" frequency over the standard temp range.
Have others used the internal clock successfully with serial comms?
Regards,
Don

[bluetooth]

From the 628 Data Sheet showing the min - nominal - max for INTRC:

[Humberto]

Unless:

1) your project is very tight in size

2) your sales forecast qty is > 10K units where
the crystal cost will have high incidence in the total cost.

3) your comunication strings are very short and long time spaced.

I would use a resonator/crystal as clock time base.

Humberto

[Don P]

Thanks for the insight. I still can't find the page (out of 144) that shows the clock specs. I wonder if the accuracy is related more to temperature or to individual parts. If it was stable for an individual chip, it seems that you could just specify the clock frequency in your code as something other than 4MHz (assuming you could determine it) and then the baud rate would come out OK.

[bluetooth]

This is where I found it (old data sheet):

DS40300B-page 139


Some of the newer parts have an OSCTUNE capability - these parts have been around for a while - look at some of the newer ones. They have what you're talking about.

[Charlie U]

Take a look at the 628A. They have an improved accuracy internal oscillator. From 2.2 to 5V and 0degC to 85degC the max variation is +/-2%. Which should be good enough for serial comm.

[Guest]

That figures -- and all mine are 628's. Maybe next batch.