AVR Freaks

With your 10X probe, you can look directly at the oscillator X2 pin. It will shift the frequency very slightly (a small fraction of a percent, maybe). The signal you see WILL NOT be full logic amplitude and it will be a less perfect square wave than the CKO pin. In fact, it may look more like a "sine with corners". But you will not have to change any fuses.

The waveform you see SHOULD be steady in both DC level and amplitude. At that frequency, probe compensation may not show the classic under-peak/over-peak but may, instead, just change the apparent amplitude. Normally, you compensate the probe with a much lower frequency square wave, such as 1KHz. 

By the way, with a 100MHz scope, a 16MHz actual square wave, such as out of CKO, will look pretty square. The rise and fall times are 3.3ns out of 31.25ns high and 31.25ns low. So the rises and the falls only take about 10% of each half cycle. Actually, it will be a bit slower than that, because the logic out of an AVR has rise and fall times of its own around 6ns. So, I guess that, actually, the rises and the falls will take closer to 1/3 of each half cycle. That is a bit less square than I started out suggesting and that is because of the chip's internal rise and fall times, not the scope.

Jim

IIRC, those UNO's mimic an AVRISPMKII or STK500.  You might be able to use Studio and then you can simply go into device programming and a simple checkbox will solve all.

Jim

One could also write a simple program to flash an LED at once / second.

To avoid questions & all the uncertainty about whether a "bad" rate is actually due to any timers settings  (top?,  bottom?,  CTC?,  std?, OCRA?, TCCRA1?)... I prefer the no-doubt way:

loopy:		sbi PORTB, PB1	;2 cycles...(1 on xmega & avr8L)
		nop		;1 cycle
		nop		;1 cycle
		nop		;1 cycle

		cbi PORTB, PB1	;2 cycles...(1 on xmega & avr8L)
		nop		;1 cycle
		rjmp loopy	;2 cycles  ...10 total cycles

This little snip requires no register setup (other than setting port directions, DDRB ) and gets straight to the result.  
Using a 20MHz Xtal, you'd measure a 2MHz pin frequency.   This also verifies the any system clock prescaler software setting is proper & taking hold. Thereafter, you're certain any timer issues are due to their setup.

Right here:

http://ww1.microchip.com/downloads/en/DeviceDoc/Atmel-42735-8-bit-AVR-Microcontroller-ATmega328-328P_Datasheet.pdf

If you want a micro to do "two things at once" put at least one of the processes on an interrupt.

Having said that most AVR timers can happily toggle pins on/off completely independent of any CPU activity once started so I would just set a timer going and have it toggle a pin. Doesn't have to be up in the MHz if you don't want. In fact if you bring it down into Hz range you could stick an LED on it and count flashes!

A 328 has OCOA on pin 12 for example - consider setting up timer 0 to toggle it on/off. It will be "full swing" so should give a nice clean square wave on a scope. Something like:

int main(void) {
    // make the timer output pin (PD6) an output
    DDRD = (1 << 6);
    // set to toggle OCoA pin and run CTC
    TCCR0A = (1 << COMA0) | (1 << WGM01);
    // Set some frequency
    OCR0A = 13;
    // start the timer
    TCCR0B = (1 << CS00);
    // get on with other stuff...
}

Vary the 13 put in OCR0A to vary the frequency.

 I wanted to see both work at the same time to visually compare.

why???  I thought you just wanted to verify your clock rate was proper.

sounds like it's going to be another of those threads that goes on & on wandering from one topic to another ...

So just enable the CLKOUT fuse.

How did you mnage to have some signal **before* 9600 bauds rise?

That's the supreme power of the digital storage scope...it's essentially storing away the waveform(s) all the time, then when desired trigger occurs, you can view what happened after or before the trigger.  Old analog scopes could only give the afterwards, since there was no storage (neglecting analog storage scopes, which were somewhat like a photographic plate).  

edit: changed their to there!!!  hate when that happens

My serial signal is 0x55 being continuously sent as 1 start bit, 8 data bits, and a 1 stop bit, giving me a square wave output.

And yes, as it is a digital scope, I can see signals both before and after the trigger point. Think of it as a sliding window on a set of signals.

@OP

What scope do you have? 100MHz scope's are pretty decent and is beyond the "cheap" stuff.

Something like a Rigol / Hantek / Uni-T ?

"cheap" scopes nowaday's cost USD20 or so ( jyetech dso138 ) and those are toys.

If they would have taken the effort to write bit better software for it then it would have been useful however for signals upto 200kHz or so.

Unfortunately they completely bodged the triggering on that thing.

The AVR's have a crystal oscillator, which runs (if configured correctly) at the same frequency as the crystal.

Cristal should have reasonably sine wave on it's legs. (But the probe capacitance changed the frequency a bit while measuring).

Amplitude is between a few 100mVpp an 4Vpp for the "full swing" configuration of mostly older AVR's.

In the AVR itself this sine is "boosted" / amplified to a square wave, and very likely also phase shifted to generate delay's for internal timing, but those signals can only be found on the silicon itself.

There is a "fuse" in the avr which can divide the clock frequency by 8 (Or is that fuse only for the RC oscillator).

So you AVR runs at the Crystall frequency or at 1/8 of that.

Yow Neil:

your lines are squiggly.

A scope is generally specified by its bandwidth, not sample rate (although digging into the spec should reveal the sample rate).

For example, a Rigol DS1052 50MHz 2 channel scope samples at 500MSa/s when using both channels.

For DSOs, both are key parameters.

BW is traditionally the "headline" number, but the sample rate should be right up there in the key specs