Can you use 15p capacitors for 8MHz ?

We are in a situation where we have to wait for capacitors , and only have 15pF . Our pcb has atmega 328p tqfp32 , and external 8MHz .

1.Could we use the 15p ? how much is it going to hurt accuracy ? (we have i2c bus but expect that we dont care that much about 50ms here or there .

2.If we would like to wake up from sleep the processor every few seconds, and we don't care for 50ms difference, should we use internal oscillator instead ? whats the downsides ?

Thanks .

From the information you've provided there is no way to answer #1. The total capacitance is important. The best path is to use the 15 pF capacitors then measure the processor's clock.

2 has always worked well for me. The downside is that the internal oscillator is ±10% from the factory and that it is sensitive to voltage and temperature. It can be tuned to ±1% over the entire voltage / temperature range.

thanks ,

The device is battery operated , and should work for long time.

Is internal oscillator is less battery drainer ? is it "professional" enough to use internal for mass production?

Setting a lower capacitor, hurt the final frequency that stay const on 8Mhz-xMhz , or is it causing an unstable frequency ?

I would try the 15pF and see. If the oscillator starts then you are not going to notice any difference due to the slight frequency pull.

What capacitance do you want? 22 pF? Two 15 pF in series makes 7.5 pF, add a 15 pF in parallel makes 22.5 pF.

BenStlr: Is internal oscillator is less battery drainer ?

I believe it is. Check the datasheet.

is it "professional" enough to use internal for mass production?

Are you actually asking if the folks at Atmel made the internal oscillator in an unprofessional way?

BenStlr: We are in a situation where we have to wait for capacitors , and only have 15pF . Our pcb has atmega 328p tqfp32 , and external 8MHz .

1.Could we use the 15p ? how much is it going to hurt accuracy ? (we have i2c bus but expect that we dont care that much about 50ms here or there .

2.If we would like to wake up from sleep the processor every few seconds, and we don't care for 50ms difference, should we use internal oscillator instead ? whats the downsides ?

Thanks .

Despite what others may say here, I have found that the crystal load capacitors are:

(1) Not needed at all. (2) Are not critical in case you do use them.

Using or not using the capacitors will affect the crystal frequency only tens of hertz (which is totally insignificant at 8 or 16 MEGAhertz).

Leave the caps off, or if it bothers you, use your 15 pF parts. Either way, it makes no practical difference.

BenStlr:
Is internal oscillator is less battery drainer ? is it “professional” enough to use internal for mass production?

I made some rough measurements with a 328P running a blank sketch at 5V and room temperature:

328p, 5V, room temp, blank sketch loaded

14.5mA - 16MHz xtal, full swing, fast rise (F7 D9 07)
13.4mA - 16MHz xtal, low power,  fast rise (FF D9 07)
11.1mA -  8MHz xtal, full swing, fast rise (F7 D9 07)
 9.7mA -  8MHz xtal, low power,  fast rise (FF D9 07)
 8.6mA -  8MHz internal (lfuse:C2 hfuse:D7 efuse:07)

I’ve never been able to find this information in the datasheet. The supply current for 5V and 8MHz is listed as 5.2mA typical and 9.0mA maximum.

BenStlr:
Setting a lower capacitor, hurt the final frequency that stay const on 8Mhz-xMhz , or is it causing an unstable frequency ?

It will likely work fine but may run a tiny little bit faster than 8Mhz. Or it might turn out that 15pF is the perfect added capacitance for your board.

Krupski: Using or not using the capacitors will affect the crystal frequency only tens of hertz (which is totally insignificant at 8 or 16 MEGAhertz).

The difference due to adding capacitors to an 8MHz crystal can easily be hundreds of hertz. I've measured one 8MHz crystal that was around 200ppm fast without capacitors. That translates into it being off by 1.6kHz.

Whether that's insignificant depends on the application.

jboyton: The difference due to adding capacitors to an 8MHz crystal can easily be hundreds of hertz. I've measured one 8MHz crystal that was around 200ppm fast without capacitors. That translates into it being off by 1.6kHz.

Whether that's insignificant depends on the application.

Not as bad as the vibrating wet noodle they call a "resonator" that's found on almost every Arduino.....