Searching \ for '12c509 clock frequency vs voltage' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/timers.htm?key=clock
Search entire site for: '12c509 clock frequency vs voltage'.

Truncated match.
PICList Thread
'12c509 clock frequency vs voltage'
1998\02\05@151313 by PIC development

flavicon
picon face
Someone here asked about how the internal RC clock
varied with supply voltage.
Well I did a little work on this a while back and I
have eventually got around to publishing it.
See www.beowulf.demon.co.uk/pic_12c509_clock_freq.html
for the results, but for the impatient, the conclusion
I reached was that the frequency varies about 3% over
the range 2.5 to 5.0 volts.
This was done with a sample size of 1 device. YMMV

Pete

..............................................................................
. Never trust a man who, when left alone .......... Pete Lynch               .
. in a room with a tea cosy, doesn't try .......... Marlow, England          .
. it on - Billy Connolley ......................... spam_OUTpicTakeThisOuTspambeowulf.demon.co.uk ..

1998\02\05@163238 by Mauro, Chuck

flavicon
face
Pete,

Thanks for your good work.  Nice graph, BTW.  There, is however, one
item we should note.  In fact, you briefly and indirectly touch upon it.
You stated that your testing was done with a sample size of 1.

Now, for all practical purposes, that's great when you build just 1
widget.  You show Fosc as a function of Vdd.  OK for when you just want
to be concerned about Fosc at a single temp as Vdd is varied.  We'll
assume for the moment that your measurements were done at a fixed T of
25C.

Fosc, however, is really a function of 3 things:

1) Vdd
2) Temp
3) Process variations (w/ normal distribution)

(OK, OK - 4 things if you count OSCCAL adjustability...)

In order to be a better test, you may want to sample several units (the
more the better - hundreds would be nice but impractical for most of
us).  What you would wind up finding is Microchip's PROCESS CORNERS in
your testing.  Then your graph would show a BAND of tolerance around a
center frequency.  This, in fact, is what is published in the data
sheets by Microchip.  I worked very closely with their test engineer to
help fully characterize the true range of the oscillator.  And, in fact,
yes, you can count on the osc being guaranteed to work within a +/- 3
sigma variation at a given Vdd and T.

If you want to really try something interesting, and help beef up your
web page, take the same test unit (or several if you can to show process
tolerance), and derive Fosc the same way, only vary T over a fixed Vdd.
Do this several times for various Vdd's, and draw a FAMILY of curves
showing Fosc variation over T for the FAMILY of Vdd's.  Don't forget,
each one of those family curves will also have a Process variation
tolerance band.  And, you can do that for the T's too.  Fix a Vdd and
vary T.  Again derive a family of curves and show a tolerance band
around the individual family members, according to the process corners
you discovered.

As you can see, this can be quite an extensive set of tests, but,
fortunately for us - Microchip has already done a lot of this work for
us, and used thousands of units over many test wafers.  The result?
Their original prediction for Fosc (in the early, Prelim data sheet) was
pretty darn close!  The curves in the graphs at the back of the 12C5xx
data sheets tell it like it is!

And, if you still don't like the Fosc you get with the original OSCCAL,
you can play with the value!

Bottom line (and this was really important to me at the time I did the
analysis work back in '96-'97): you can use this part for a RS232
bit-bang routine if you want, it will work!  (My research showed about
-3% - +2.5% variation over process, temp and voltage.)

Again, nice web page.  Keep up the good work!

Cheers,

Chuck Mauro

PS:  what's YMMV?

> {Original Message removed}

1998\02\06@111249 by John Shreffler

flavicon
face
What is a "tea cosy" please?

-----Original Message-----
From:   PIC development [SMTP:.....picKILLspamspam@spam@BEOWULF.DEMON.CO.UK]
Sent:   Thursday, February 05, 1998 3:13 PM
To:     PICLISTspamKILLspamMITVMA.MIT.EDU
Subject:        12c509 clock frequency vs voltage

Someone here asked about how the internal RC clock
varied with supply voltage.

[snip]

This was done with a sample size of 1 device. YMMV

Pete

.............................................................................
. Never trust a man who, when left alone .......... Pete Lynch               .
. in a room with a tea cosy, doesn't try .......... Marlow, England          .
. it on - Billy Connolley ......................... .....picKILLspamspam.....beowulf.demon.co.uk ..

Attachment converted: wonderland:WINMAIL.DAT (????/----) (00012271)

More... (looser matching)
- Last day of these posts
- In 1998 , 1999 only
- Today
- New search...