- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
PSoC5 has multiple clocks.
I wonder which clock drives the CPU.
Further a analog component can be given a digital clock. Doesn't the noise in the digital clock affect the analog performance.
- Labels:
-
PSoC 5LP
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
The BusClock drives the CPU and is itself fed by the Master Clock which is fed by ... double click in the clock view on a system clock to see all the dependencies!
The ADC's noise (have a look into the datasheet) is a guaranteed value and seems to be not affected from internal clocks as far as you cling to the recommendations of wiring analog signals (see AN58827 and AN58304) you'll get the expected precision.
Bob
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
The CPU is driven by BUS_CLK. You can see this in the clock editor ('configure system clocks'). If you route a clock to an anlog component, this clock will be moved to the 'analog domain' (and it turns orange to show that).
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
The confluence of analog and digital on a chip present many challenges.
Today’s chips use CMOS isolations wells, power and ground routing techniques,
die attach substrate methods to advance into late 70's, 80's analog perform-
ance.
The biggest single problem in our profession, is the lack of understanding on
how to do an end to end analog error budget tally. Examine product claims,
you know intuitively something is amiss. "I have 2,789 bit resolution from 0
degrees K to surface of the sun..."
Aha.
I generally advise folks trying to do 10 bits over T, PSRR, Supply, and Noise, to do a complete
error budget. You quickly find 10 bits is a challenge.
And educating folks on the meaning of resolution, precision, accuracy……And that averaging is not always applicable in your environment, especially when your noise is highly correlated. A number
of ap notes on this topic.
That’s why we pay several $K for good quality analog instruments, its no accident they are worth every penny.
So in short I think that whenever possible you shut off all clocks, UDBs, I/O activity when trying
to do high performance analog measurement. And test, as the saying goes “trust, but verify”.
Regards, Dana.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Trust the engineers at Agilent (HP), Tektronix, Marconi, Phillips, Analog Devices, LTC,
Keithley (Get their book on Resistors),.......for analog advice. Go to their seminars, meet
the guys that design 80 Ghz scopes, always seek those that know 100X more han you do.
We are in a great business.
Regards, Dana.