Model A: How low can you go? (Power consumption)

[M33P]

So I got a batch of Model As in the post. "Low Power", they said. "Suitable for battery power", they said.

Well we'll just see about that.

First some theory. CMOS devices have two main power consumption considerations - static and dynamic. Static is as expected, the power consumption that would occur if the silicon were in a state where no clock cycles were fed to it. This is mainly made up of subthreshold leakage (transistors leaking when off) which absolutely plagued desktop PCs until recently. This value is (roughly) a function of the square of the applied voltage VDD.

Dynamic consumption is exhibited when clock cycles change the state of the transistors on the silicon. CMOS transistor gates are essentially small capacitances that must be charged or discharged to turn them on or off - this gives rise to energy expenditure proportional to the square of the applied voltage and proportional to the clock frequency.

There are three main silicon groups that can be poked/played with on the Pi. The DDR2 SDRAM chip has its own clock speed and voltage regulator. The GPU has its own set of clock speeds and the ARM has its own clock speed. The GPU and ARM share a voltage regulator.

To test the power consumption, I went with the reliable and simple current shunt resistor - I picked four 0.75 ohm power resistors out of my bucket of parts and soldered them in parallel, then precisely measured the resistance by putting 1 amp through them with a bench power supply - 0.185 ohms exactly. This went into the +ve lead of a supply from a small wall wart that then powered the Pi directly via the GPIO. Thus I could measure the current and voltage very precisely.

Results:

For comparison the model B 256M (default clocks):
Idle (console): 0.372A 1.89W
Idle (lightdm running): 0.375A 1.91W
Running scp from SD card: 0.413A 2.09W

Model A (default clocks)
Idle (console): 0.113A 0.587W
cat /dev/urandom | gzip: 0.153A 0.795W

Model A (ARM=350MHz overvolt=0)
Idle (console): 0.111A 0.576W
cat /dev/urandom | gzip: 0.133A 0.691W

Model A (ARM=350MHz overvolt=-8)
Idle (console): 0.106A 0.554W
cat /dev/urandom | gzip: 0.125A 0.652W

At this point I started playing with the SDRAM clocks and voltages. Long story short: this does nothing to improve power consumption and most often the Pi simply wouldn't boot at reduced voltage settings. DDR2 inherently incorporates various power saving features into the silicon - there is nothing to be gained from underclocking it.

Then I turned my attention to the GPU - again a story of instability and in most cases ridiculous slowdown. It seems that the most to be gained from underclocking is about 20% reduction in idle power consumption using the cpufreq scaling.

Model A (ARM_min=350M overvolt_min=-6)
idle (no x): 0.106A 0.551W
yes > yes.txt (SD write): 0.174A 0.902W
cat /dev/mmcblk0 > /dev/null (SD read): 0.154A 0.804W

My gut feeling is that the CPU/GPU isn't sleeping when idle. This is evident in the rather small difference between idle and load power consumption: my Core i5 desktop machine achieves a much higher difference between idle and load for CPU-bound activity.

I also note that the model A's biggest single power user is RG1, the 5V to 3v3 regulator. The percentage of total consumption will vary depending on input voltage, but by calculation the device will account for between 30% and 37% of the model A's total power usage. Replace this with a switching regulator >90% efficient and we could end up with 0.35W idle power consumption on the model A.

tl;dr version: Model A uses <33% (0.8W) of a Model B's power consumption (1.89W) using default settings. Underclocking and undervolting give negligible benefit for a ridiculously slow Pi. The low-hanging fruit for power optimisation is and always will be RG1, the 5V to 3.3V regulator on the Pi which will account for 30-37% of total power consumption regardless of version of Pi..

[pygmy_giant]

can someone summarise that for us simpletons?

[daveake]

can someone summarise that for us simpletons?

Model A uses about 1/3rd the power that the model B uses.

Underclocking makes next to no difference.

To save more power replace the on-board linear 3.3V regulator with a switching one.

[pygmy_giant]

Cool thanks - 1/3 - not bad.

I use a switching regulator with lipos into the 5v pins.

[conundrum]

Hmm, thats worth a try.
Maybe use an NCP3064 as that also has an enable pin and comes in a smaller package.

I did wonder if adding a power off feature to the Pi and "wake up on x" ie when a USB is plugged in would be handy for the XBMC folks.
Also adding something that turns the Pi on if it receives an IR signal would be useful here, this could be done using the 3064 itself and some trivial Rx circuitry
See http://www.onsemi.com/PowerSolutions/pr ... id=NCP3064

-A
#include "HowmanydigitsofPicanyoucalculateonaPiin24hrs.h"

[hojnikb]

Someone should put a switching 5V -> 3V3 on model A and measure idle W's then, so we can see whats the real world difference..

[poing]

I might try if I'd know what to do exactly. I can solder, but I don't know what part to use and how to connect it.

[drgeoff]

I might try if I'd know what to do exactly. I can solder, but I don't know what part to use and how to connect it.

Unlike linear regulators, the switching versions usually need a separate inductor so it isn't a direct chip for chip swap.

[poing]

I might try if I'd know what to do exactly. I can solder, but I don't know what part to use and how to connect it.

Unlike linear regulators, the switching versions usually need a separate inductor so it isn't a direct chip for chip swap.

Thanks for your comments but it's no help at all. I have a spare model A, I can solder (somewhat) and I'm (somewhat) in need of the lowest power consumption possible. Another route is to buy another battery.

Seriously, HOW do I replace the linear 5V to 3.3V regulator with a switching one?

[drgeoff]

I might try if I'd know what to do exactly. I can solder, but I don't know what part to use and how to connect it.

Unlike linear regulators, the switching versions usually need a separate inductor so it isn't a direct chip for chip swap.

Thanks for your comments but it's no help at all. I have a spare model A, I can solder (somewhat) and I'm (somewhat) in need of the lowest power consumption possible. Another route is to buy another battery.

Seriously, HOW do I replace the linear 5V to 3.3V regulator with a switching one?

1a. Obtain a ready made switching regulator module which will deliver 3.3 volts at the required current from 5 volts. Sometimes to be found on ebay eg http://www.ebay.co.uk/itm/LM2596S-Const ... 3f2158069e. Make sure to set output voltage to 3.3 before connecting in step 4 below. Or

1b. After googling for suitable chip and documentation with application note, obtain switching regulator chip and required sundry components (eg coil, oscillator capacitor, voltage setting resistors) and solder to breadboard type PCB or other chassis.

2. Find RG2 on RPi and determine the locations of 5v, 3.3v and ground terminals. A voltmeter is one way of discerning which is which.

3. Remove RG2 from RPi.

4. Connect replacement regulator from step 1 above with its input going to 5v location, its output going to 3.3v location and its ground to ground.

5. While keeping fingers crossed that no magic white smoke escapes from new regulator or RPi, reapply power to RPi.

[ait]

plugwash wrote

Afaict the model A* really only uses the 3.3V rail to supply an IO voltage and a few other minor bits and peices so don't be surprised if the gains from using a switched mode 3.3V supply on a model A are negligable.

http://www.raspberrypi.org/phpBB3/viewt ... 25#p280025

[M33P]

I think I might have derped somewhat in my analysis of the power usage of the regulators.

The 5V input goes to the SMPS on the BCM2835 (VDDBAT) as well as the chained linear regulators.

5V is then stepped down to 3v3 for use by VDDIO or stepped down via 2v5 or 1v8 regulators. From the schematic, 1V8 will probably have the most load on it for the model A (SDRAM). Using an accurate temperature probe (my finger) the hottest componets are IC1/2 followed by RG2 (3v3) followed by RG1 (1v8).

This makes some kind of sense if 1v8 is used for the SDRAM Vdd. This suggests that the biggest power savings may be going from 5V to 1V8 by using a switcher in place of RG1.

[drgeoff]

This suggests that the biggest power savings may be going from 5V to 1V8 by using a switcher in place of RG1.

Sometimes equipment can be sensitive (*) to the precise sequence in which power rails come up. Changing the RPi's 1.8 volt rail to come off the 5 volts might change that.

* best case = RPi does not boot properly.
worst case = I/O stages in some chips are permanently damaged.

[M33P]

This suggests that the biggest power savings may be going from 5V to 1V8 by using a switcher in place of RG1.

Sometimes equipment can be sensitive (*) to the precise sequence in which power rails come up. Changing the RPi's 1.8 volt rail to come off the 5 volts might change that.

* best case = RPi does not boot properly.
worst case = I/O stages in some chips are permanently damaged.

Well since you asked so nicely:

Here's the powerup waveforms for 5v0/3v3/2v5/1v8. Red in each case is TP1-TP2 (5V powerup from external PSU) and blue is the respective supply rail.

Marked on the cursors is the time from 5v0 starting to rise to the blue rail reaching 0.9x its final value (rise time). 5V0 is the overall slowest to reach the final value which is as expected; the linear regulators downstream will reach their operating value as their input reaches tolerable levels before the 5V0 stabilises.

[poing]

1a. Obtain a ready made switching regulator module which will deliver 3.3 volts at the required current from 5 volts. Sometimes to be found on ebay eg http://www.ebay.co.uk/itm/LM2596S-Const ... 3f2158069e. Make sure to set output voltage to 3.3 before connecting in step 4 below. Or

1b. After googling for suitable chip and documentation with application note, obtain switching regulator chip and required sundry components (eg coil, oscillator capacitor, voltage setting resistors) and solder to breadboard type PCB or other chassis.

2. Find RG2 on RPi and determine the locations of 5v, 3.3v and ground terminals. A voltmeter is one way of discerning which is which.

3. Remove RG2 from RPi.

4. Connect replacement regulator from step 1 above with its input going to 5v location, its output going to 3.3v location and its ground to ground.

5. While keeping fingers crossed that no magic white smoke escapes from new regulator or RPi, reapply power to RPi.

OK, thanks, that was helpful. I ordered said regulator which will take about three weeks to arrive. When it does I'll report back, hope you guys will hold my hand

[Lob0426]

I have been looking at these from Adafruit.
http://www.adafruit.com/products/1066
to replace RG1. A bit expensive though.

[drgeoff]

I have been looking at these from Adafruit.
http://www.adafruit.com/products/1066
to replace RG1. A bit expensive though.

Probably a typo but RG1 provides the 1.8 volt supply and the product you link to is designed to give 3.3 volts.

[Lob0426]

Probably a typo but RG1 provides the 1.8 volt supply and the product you link to is designed to give 3.3 volts.

Not a typo, just a brain ****. That would replace RG2. The easiest course I can find to reduce loss at the regulator. That regulator is a low-dropout part. it could directly replace RG2 as long as you are at 4.75V at the test points.

To create a RasPi that would work well for robotics you would need to have a 5V part as well as the 3v3 unit. This would allow you to input 5v to 32v to the RasPi. Of course both of those regulators cost almost as much as a B model.

[poing]

1a. Obtain a ready made switching regulator module which will deliver 3.3 volts at the required current from 5 volts. Sometimes to be found on ebay eg http://www.ebay.co.uk/itm/LM2596S-Const ... 3f2158069e. Make sure to set output voltage to 3.3 before connecting in step 4 below.

Check

2. Find RG2 on RPi and determine the locations of 5v, 3.3v and ground terminals. A voltmeter is one way of discerning which is which.

Check

3. Remove RG2 from RPi.

Check (although I just cut the 5V and ground end and bend them slightly upwards so that I can re solder them easily)

4. Connect replacement regulator from step 1 above with its input going to 5v location, its output going to 3.3v location and its ground to ground.

Check

5. While keeping fingers crossed that no magic white smoke escapes from new regulator or RPi, reapply power to RPi.

Check

With the original regulator in place I measured 129.5 mA, with the alternative regulator I measured 131.2 mA

[poing]

And re soldered, works like normal again. Phew

[M33P]

1a. Obtain a ready made switching regulator module which will deliver 3.3 volts at the required current from 5 volts. Sometimes to be found on ebay eg http://www.ebay.co.uk/itm/LM2596S-Const ... 3f2158069e. Make sure to set output voltage to 3.3 before connecting in step 4 below.

5. While keeping fingers crossed that no magic white smoke escapes from new regulator or RPi, reapply power to RPi.

Check

With the original regulator in place I measured 129.5 mA, with the alternative regulator I measured 131.2 mA

The problem with that regulator is the lack of synchronous rectification. From the efficiency curves in the datasheet, the efficiency approaches that of a linear regulator the lower the output voltage as it has to contend with the forward voltage drop of the diode. It _is_ possible to get highly efficient switchers but they must be optimised for the application. In this case the max current we are dealing with is 150mA - using a 3A switcher will lead to excessive standing losses due to the larger FET being switched.

[MozisII]

Hi,
I make some test on raspberry to quantify influence of underclock and overclock on Raspberry consumption. For ones interested in result you can take a look on http://moze.free.fr/blog/index.php?post ... -Overclock.
Part of conclusion is that only core frequency have a big impact on overclocked raspberry in load without graphics and there is no lot thinks to do on the bcm to gain in idle.
(I moved this post from the new topics I had created to this better place)

[M33P]

@mozisll
So your conclusions are similar to mine - and you did more charting of data which is nice. It's clear that there's not much gain from underclock at the expense of a very slow Pi.

About regulators - does anyone feel like giving this a spin:
http://uk.farnell.com/texas-instruments ... dp/2066267

Costs nearly as much as a Pi... How long will it take for it to save enough power?

Pi @ 24/7 usage, idle 90% of the time, loaded 10% of the time, no GPU usage.
0.9 x 0.587 + 0.1 * 0.795 = 0.6078W
Typical efficiency of wall wart rated Energy Star EPS2.0 class V - 66% at that load
0.6078 / 0.66 = 0.921W (active power)
In a year with a typical UK energy tariff @ £0.15 per kwh:
0.921 Wh/hr * 24*365.25 = 8.073kWh/year
8.073 * 0.15 = £1.21 plus pocket lint per year.

Let's say replacing RG1 with a switcher reduces power consumption by 33%. Difference is 40p a year - which will take approximately 40 years to recoup the cost of the switcher

That aside... aside, I'm going to buy one of these because I have 4 model As to experiment on and only need 3 to automate my house.

[MozisII]

@M33P
My own concern is autonomous powered systems as solar or battery equipment then energy tariff is not the sames. There is 4 voltage levels and we have no informations on current consumption on each levels. Sure the head regulator efficiency impact all levels but final result is unknown.
We need know internal way to reduce idle consumptions to have a real result.

[davips]

Hello,
I have ordered a version A RasPi that will be powered by batteries.
I read about the failed attempt to change the voltage regulator,
but as a not-too-expert electronic, I have a question:
I am almost sure, but, could the regulator be replaced by a series (maybe 3) of common rectifier diodes?
Would the diodes have the same heat dissipation effect as the linear voltage regulators?

thanks