Crystal Oscillator exceeding spec?

Thread Starter

jjmalove

Joined Jul 6, 2016
51
Hello,

I have a specific question about a problem I am trying to debug. I have a PCB using the ATMEGA640-16-CU that I suspect is messing up, but I don't have access to the firmware on it. Here is the datasheet:
http://ww1.microchip.com/downloads/...r-ATmega640-1280-1281-2560-2561_datasheet.pdf

Pg 358 shows the Maximum Frequency vs. Vcc. It is using a 14.8Mhz crystal and Vcc is 3.3V. It is currently unknown what speed the firmware is running at (trying to get that answer).

The graph clearly shows 14.8Mhz to be outside of the safe operating area. What I don't know is if the firmware is running slower than 14.8Mhz, would it be fine? Is this chart strictly hardware speaking I cannot use higher than around 9-10Mhz at 3.3V, or does it depend at all on the firmware?

Thanks!

Edit: FYI I am fairly inexperienced with microcontroller firmware so if my word choice or idea is completely wrong sorry! Much more experienced with electronics hardware.
 

jpanhalt

Joined Jan 18, 2008
11,087
There is variation between parts. The graphs represent more or less guaranteed performance. Some parts can do better. I am not familiar with Atmel parts specifically, but that holds true for Microchip's PIC line. I would stick to what is guaranteed, unless you want to get into the practice of buying dozens of parts to find those that work. By analogy, think abut 5% resistors. For any batch, you will find some, maybe most well within the 5% tolerance limit.

John
 

Thread Starter

jjmalove

Joined Jul 6, 2016
51
There is variation between parts. The graphs represent more or less guaranteed performance. Some parts can do better. I am not familiar with Atmel parts specifically, but that holds true for Microchip's PIC line. I would stick to what is guaranteed, unless you want to get into the practice of buying dozens of parts to find those that work. By analogy, think abut 5% resistors. For any batch, you will find some, maybe most well within the 5% tolerance limit.

John
Hey John,

Actually I am debugging something I didn't design. Just trying to figure out why its failing and one direction its pointing is the atmega processor. My question is whether that graph I am reading is talking strictly about what sort of frequency input I am providing to XTAL_1, or if its talking about what max speed in firmware the processor can be programmed to run at, or a combination of the two.
 

Papabravo

Joined Feb 24, 2006
21,159
There is variation between parts. The graphs represent more or less guaranteed performance. Some parts can do better. I am not familiar with Atmel parts specifically, but that holds true for Microchip's PIC line. I would stick to what is guaranteed, unless you want to get into the practice of buying dozens of parts to find those that work. By analogy, think abut 5% resistors. For any batch, you will find some, maybe most well within the 5% tolerance limit.

John
The firmware is running off of the main clock frequency. For the Vcc you have chosen you cannot guarantee correct operation. Here is what I would do if you can manage it. Raise the Vcc to 5V (4.5V) and see if the operation changes. It is certain that the part is guaranteed to operate with a Vcc of +5V and an oscillator at 14.8 MHz. Alternatively, lower the frequency to a value inside the Safe Operating area those two experiments should give you some insight.
 

Thread Starter

jjmalove

Joined Jul 6, 2016
51
The firmware is running off of the main clock frequency. For the Vcc you have chosen you cannot guarantee correct operation. Here is what I would do if you can manage it. Raise the Vcc to 5V (4.5V) and see if the operation changes. It is certain that the part is guaranteed to operate with a Vcc of +5V and an oscillator at 14.8 MHz. Alternatively, lower the frequency to a value inside the Safe Operating area those two experiments should give you some insight.
Awesome, thank you very much!
 
Top