# Are watts per MHz linear?

Started by October 7, 2005
```In looking at various 32-bit MCUs, I've seen watts per MHz mentioned. A
linuxdevices.com news article talks about Powerpc 405: "Power usage is
listed as 0.2mW/MHz".

So lets take the simple PIC. Can I clock it down and watch power usage
go down linearly? If a chip is 10mA at 10MHz, will it be 10uA at 10kHz?
I take it theres always a base current at DC. so the equation should be
(K x MHz) + Base leak at DC.

Is this correct for a fixed temperature and pressure? Can I rely on
this equation for extreme low power designs?

```
```On 7 Oct 2005 16:17:42 -0700, "Ghazan  Haider" <ghazan.haider@gmail.com>
wrote:

>In looking at various 32-bit MCUs, I've seen watts per MHz mentioned. A
>linuxdevices.com news article talks about Powerpc 405: "Power usage is
>listed as 0.2mW/MHz".
>
>So lets take the simple PIC. Can I clock it down and watch power usage
>go down linearly? If a chip is 10mA at 10MHz, will it be 10uA at 10kHz?
>I take it theres always a base current at DC. so the equation should be
>(K x MHz) + Base leak at DC.
>
>Is this correct for a fixed temperature and pressure? Can I rely on
>this equation for extreme low power designs?

I'd only rely on it if it's specified in the data sheet, NOT a press
release. If it's quoted as a "typical" value, I'd give it an order of
magnitude either way. Also, not all designs are amenable to very low
clock speeds. Not familiar enough with the PPC to know if it's good down
very close to a "DC clock."

--
Rich Webb   Norfolk, VA
```
```Microchip is kind enough to provide us with the "Electrical
characteristics" of their chip (usually at the end of a datasheet). I'm
looking at a 18F2550 right now, and it's not linear in respect to
Watt/Fosc.

Have a look at the datasheet for the device of your choice, it's all
there. Good luck!

// Erik N=E4slund

```