Eng-Tips is the largest engineering community on the Internet
Intelligent Work Forums for Engineering Professionals
-
Congratulations waross on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!
IC Power Consumption
- Thread starter swb1
- Start date
Look at the data sheet.
It will specify "output pins not connected", "Oscillator pin held low", "Vcc = 5.0V", 25C, etc.,etc.
Some IC data sheets will specify quiescent current as a function of clock frequency and provide an equation.
Why are you bothering to check an IC? Usually you want to know a product's draw not just an IC. The data sheets are rarely wrong in their statements as this can be easily checked.
Like all things IC, measuring one instead of using the data sheet is not usually a good idea.
Keith Cress
Flamin Systems, Inc.-
It will specify "output pins not connected", "Oscillator pin held low", "Vcc = 5.0V", 25C, etc.,etc.
Some IC data sheets will specify quiescent current as a function of clock frequency and provide an equation.
Why are you bothering to check an IC? Usually you want to know a product's draw not just an IC. The data sheets are rarely wrong in their statements as this can be easily checked.
Like all things IC, measuring one instead of using the data sheet is not usually a good idea.
Keith Cress
Flamin Systems, Inc.-
Not to contradict, itsmoked, but just to add a caveat.
If you add up all the worst case power requirements from all the devices and design a power supply to meet this requirement, more often than not, you will wind up with a significantly oversized power supply (not that there is necessarilly anything wrong with this). The worst case consumption from the data sheets assumes that the device is operating at full capacity, which is not usually the case.
If you add up all the worst case power requirements from all the devices and design a power supply to meet this requirement, more often than not, you will wind up with a significantly oversized power supply (not that there is necessarilly anything wrong with this). The worst case consumption from the data sheets assumes that the device is operating at full capacity, which is not usually the case.
Data sheet accurarcy (datasheet Typ = 50uA but in reality it's closer to 75uA) or usefullness ("Average" current may be calculated on one, non-applicable use case) varies based on the complexity of the IC. For a simple logic gate, the data sheet is usually good enough. However, if you are talking about a multi-core processor, with several different modes of operation (sleep, wake, doze, etc), and hundreds of different use cases, then the data sheet may be less helpfull.
Finally, the validity of the data sheet numbers are directly related to the age of the design. In other words, early versions of IC (pre-production releases) may not always meet the final data sheet numbers.
Finally, the validity of the data sheet numbers are directly related to the age of the design. In other words, early versions of IC (pre-production releases) may not always meet the final data sheet numbers.
MacGyverS2000
Electrical
In cases such as multi-core processors and the like are being used, one must always go for the datasheets worst-case scenario, specifically because you don't know how it may be used. If you know your application is only going to have two lines high at any one time over a large number of chips, pulling the worst case is over-engineering.
The best advice we can give is "know your system". The data sheets are based upon laboratory settings where the chip manufacturers can push the chips to the limits and create an envelope of failure rates. If your chip ever goes outside of that envelope, you're beyond the 'X'-sigma it was designed for... the chances of that happening are extremely small, and if you absolutely must have a higher sigma, you'll pay for it.
For example, I know the typical PIC can source 20mA per line, but the entire chip with five 8-line ports may have a limit of 200mA. Since I know my lines are almost always used for sub-mA signaling, I never worry about that worst-case scenario, and therefore I don't design for it. I know my system. I don't bother checking the manufacturer's numbers, because if I was ever knocking at the door of the limits I know the datasheet values are going to be correct in 99.999% of the chips that roll across my desk. If one unit fails out of 10k, that number is acceptable in most cases... if it's not, I underdesigned the system.
Dan - Owner
The best advice we can give is "know your system". The data sheets are based upon laboratory settings where the chip manufacturers can push the chips to the limits and create an envelope of failure rates. If your chip ever goes outside of that envelope, you're beyond the 'X'-sigma it was designed for... the chances of that happening are extremely small, and if you absolutely must have a higher sigma, you'll pay for it.
For example, I know the typical PIC can source 20mA per line, but the entire chip with five 8-line ports may have a limit of 200mA. Since I know my lines are almost always used for sub-mA signaling, I never worry about that worst-case scenario, and therefore I don't design for it. I know my system. I don't bother checking the manufacturer's numbers, because if I was ever knocking at the door of the limits I know the datasheet values are going to be correct in 99.999% of the chips that roll across my desk. If one unit fails out of 10k, that number is acceptable in most cases... if it's not, I underdesigned the system.
Dan - Owner
Most datasheets show tables of power consumption for increasing loads (idle, WDT,ADC, etc.)
With MCUs, take into account which peripherals are you using. Most IDE packages allow simulating the program, even watching thread duration in machine cycles, so you may calculate current x thread time for each peripheral, thus getting average power estimates for your specific firmware. As said before, it will be likely lower than that shown in the datasheet.
If you want nWatt accuracy, keep in mind that IC 'thinking' drains power too. We know CMOS devices spend more power when switching between states, so any firmware with a tight loop ("loop: goto loop;") may give you a close figure of the highest power consumption in idle mode.
Good Luck!
With MCUs, take into account which peripherals are you using. Most IDE packages allow simulating the program, even watching thread duration in machine cycles, so you may calculate current x thread time for each peripheral, thus getting average power estimates for your specific firmware. As said before, it will be likely lower than that shown in the datasheet.
If you want nWatt accuracy, keep in mind that IC 'thinking' drains power too. We know CMOS devices spend more power when switching between states, so any firmware with a tight loop ("loop: goto loop;") may give you a close figure of the highest power consumption in idle mode.
Good Luck!
- Status
