Internal verses External Oscillator

[ibwev]

I am counting the revolutions of a fan (as much as 25 times a second) while using the internal oscillator at 8 Mhz on the Pic16F690. It is very important to obtain an accurate revolution count. Is the internal oscillator reliable enough for this task or should an external oscillator be used?

 

[Sceadwian]

25 times a second is nothing for a micro controller. Accuracy and repeatability are different issues. Internal or external a bad oscillator setup will drift a lot with temperature and supply voltage.

How accurate and repeatable over time do you need it to be?

 

[Nigel Goodwin]

Exactly, how accurate does it need to be? - I can't see any great importance to accurate fan speed?.

If great accuracy is needed, then you need a crystal - as simple as that.

 

[ibwev]

Thanks so much for the response.

The temperature can be as high as 110 degrees Fahrenheit to as low as 0 degrees Fahrenheit. Normal operation will occurr between 70 - 80 degree Fahrenheit. The device will be powered by a 24 volt transformer on a 120 volt circuit.

Once the fan speed drops approximately 5% from a calibrated speed, a trigger will be set.

Is an internal oscillator accurate enough for this setup.

 

[Ian Rogers]

I would have thought so.... most pics have an internal OSCTUNE value for trimming the internal OSC anyway... If I use it I read the value first(with pickit2) then just apply that value in the code ( I know this can all be done in software, but I always simulate in ISIS before making the end result )

 

[Nigel Goodwin]

Once the fan speed drops approximately 5% from a calibrated speed, a trigger will be set.

Is an internal oscillator accurate enough for this setup.

I agree with Ian, should be fine for that, it's only low accuracy.

 

[m_seifi_e]

maybe if you enable a low frequency counting timer and just count the interval between two Consecutive pulses sensed by an external interrupt satisfy your needs even with internal oscillator

 

[Sceadwian]

I don't know what chips you guys are using but given the range you really think you'll only have +/- 5% error through a 100 degree temperature range? Depending on the code written the unit could alarm continuously until it reaches 'normal' conditions or never alarm, without knowing the intention saying that this is 'fine' is jumping the gun. If the unit isn't in the air path of the fan it could be permanently stuck at the ambient of 0, this is fine if it's set for it, but if the chips temperature shifts I'm sure you'll get more than 5% error easily over that temperature range.

It might work fine on a cold day the day it was programmed and then when warmed up will alarm continuously, or worse if this does something besides control an alarm and actually alters some other portion of the system it would introduce a massive non-linearity into the feedback loop.

Another case of not enough information of what this will do, or how it will be used.

 

[Nigel Goodwin]

Another case of not enough information of what this will do, or how it will be used.

That wasn't the question, the question was about accuracy of the internal oscillator, which will be perfectly fine.

As we don't have full details of what's going on we can't comment on the project as a whole.