Some powerconsumptions tests

[Koen Kooi]

Hi,

Philip was asking about powerconsumption, so I put a fluke 87 meter in
between the +5V and beagle. The results after some quick tests:

uboot 2008.1orc2:
~290mA at uboot prompt

2.6.27-omap1:
~350mA at X desktop
~400mA using 100% cpu at 600MHz

2.6.27 omap-pm-next:
~275mA at X desktop

Powertop reports with with 2.6.27 omap-pm-next:
C0 4.1ms ( 4.4%)
C1 0.9ms ( 0.8%)
C2 7.7ms ( 3.7%)
C3 13.1ms ( 0.9%)
C4 37.8ms (88.7%)

So we can easily save 75mA by turning on CPUidle and save another 60mA
by turning of the tfp410. Once TI starts putting smartreflex enabled
chips on the beagles we can use that to regulate the CPU voltage.
Another powersaver would be CPUfreq, that's being worked on.

All tests were done on a rev C prototype running at 600MHz with 256MiB
RAM, USB in host mode attached to a powered hub.

regards,

Koen

[Gerald Coley]

The Beagle already has the SmartReflex devices on it. OMAP3530 and TWL4030.

 

Gerald

[Koen Kooi]

On 3 nov, 19:58, "Gerald Coley" <ger...@beagleboard.org> wrote:
> The Beagle already has the SmartReflex devices on it. OMAP3530 and TWL4030.

That is the case, but the E-fuses don't seem to be configured at the
factory, so you can't actually use smartreflex :(

regards,

Koen


> Gerald

[Jason Kridner]

Koen Kooi wrote:
> Hi,
>
> Philip was asking about powerconsumption, so I put a fluke 87 meter in
> between the +5V and beagle. The results after some quick tests:
>
> uboot 2008.1orc2:
> ~290mA at uboot prompt
>
> 2.6.27-omap1:
> ~350mA at X desktop
> ~400mA using 100% cpu at 600MHz
>
> 2.6.27 omap-pm-next:
> ~275mA at X desktop

Via the IRC channel, the decoder ring for this kernel is
http://git.kernel.org/?p=linux/kernel/git/khilman/linux-omap-pm.git;a=summary.

[Koen Kooi]

On 3 nov, 20:29, Jason Kridner <jkrid...@beagleboard.org> wrote:
> Koen Kooi wrote:
> > Hi,
>
> > Philip was asking about powerconsumption, so I put a fluke 87 meter in
> > between the +5V and beagle. The results after some quick tests:
>
> > uboot 2008.1orc2:
> > ~290mA at uboot prompt
>
> > 2.6.27-omap1:
> > ~350mA at X desktop
> > ~400mA using 100% cpu at 600MHz
>
> > 2.6.27 omap-pm-next:
> > ~275mA at X desktop
>

> Via the IRC channel, the decoder ring for this kernel ishttp://git.kernel.org/?p=linux/kernel/git/khilman/linux-omap-pm.git;a....

To be exact: http://git.kernel.org/?p=linux/kernel/git/khilman/linux-omap-pm.git;a=shortlog;h=tmp/pm-next
based on commit f295adb93c2a8c31ca0c58cca5994a983a77f98

Note that this tree is being rebased almost weekly, so commit hashes
_will_ change.

regards,

Koen

[Gerald Coley]

Hmmm, let me look into that. I have a hard time tacking what devices I am getting when it realtes to this feature. I thought they all should be. If nothing else, maybe we can get it enabled on the Rev C boards. I will add that to my hot sheet.

 

Gerald

[wendl...@comcast.net]

Slightly tangential to this...

 

I brought up Angstrom on my beagleboard using a creaky old power supply.  The beagleboard would get as far as "reading uImage" and then reboot.  The cycle would then repeat. 

 

It turned out that the power supply could not supply the peak current required during the booting process.  The voltage must have dropped to the point where the board reset itself.

 

Using a higher current supply fixed the problem.

 

JPW

[Koen Kooi]

On 3 nov, 19:16, Koen Kooi <koen.k...@gmail.com> wrote:
> Hi,
>
> Philip was asking about powerconsumption, so I put a fluke 87 meter in
> between the +5V and beagle. The results after some quick tests:
>
> uboot 2008.1orc2:
> ~290mA at uboot prompt
>
> 2.6.27-omap1:
> ~350mA at X desktop
> ~400mA using 100% cpu at 600MHz
>
> 2.6.27 omap-pm-next:
> ~275mA at X desktop

> So we can easily save 75mA by turning on CPUidle and save another 60mA
> by turning of the tfp410

Using the new DSS and doing

root@beagleboard:/sys/devices/platform/omapfb# echo "dvi e:0" >
displays

current consumption drops by 80mA :)

regards,

Koen

[Sean D'Epagnier]

Can't you save even more power by powering from less than 5v? I heard
talk of 3.8v. If you figure out how to do this I want to know so I
can do it as well.

Sean

[Gerald Coley]

You can run from 3.8V but I would not go below that. The onboard regulator provides power to the DVI encoder, so you need to have enough to keep it in regulation. One note, the monitor ID may not work as it requires 5V to read the ID. If you don't need this, it should be OK. Make sure it is well regulated.

 

Gerald

[Sean D'Epagnier]

Is there a risk of damage below 3.8v? I am tempted to try running a
little lower. Is it best to input the 3.8v at the 5v power jack? Is
there a better place to power the board from to reduce power
consumption.

If I am not using the DVI encoder can I go lower? Is there a way to
disable the DVI encoder to save power other than removing it?

My plan is to use a buck-boost regulator from single cell lithium
batteries and set the output voltage to the lowest possible voltage I
can get away with.

Thanks,
Sean

[Gerald Coley]

The only place you can put the voltage is via the power jack. 3.8V will not cause any damage to the board. Technically you could go lower than that, but the DVI would not work. Doen't go below the specification for the TWL4030/TPS65950. You can disable the DVI device via SW which will result in a reduction of 100mA. If you lower the voltage to a point that the DVI isn't working, I can't guarantee what the current consumption would be. We have powered the whole board with DVI running from a USB Lithium pack for I think a couple of hours. Removing the SD card will also save power, around 50mA.

 

Gerald

[freedom]

We did some quick tests. It seems it's possible to do normal jobs with
3.6 V. The reason we wanna go below 3.8 V is that there are lots
of 3.6 or 3.7 V batteries. We attached a DC power supply to the
power jack and supplied 3.6 V to the beagle board. With that, we were
able to boot the board all the way in to debian+X.

On Nov 12, 8:11 am, "Gerald Coley" <ger...@beagleboard.org> wrote:
> The only place you can put the voltage is via the power jack. 3.8V will not
> cause any damage to the board. Technically you could go lower than that, but
> the DVI would not work. Doen't go below the specification for the
> TWL4030/TPS65950. You can disable the DVI device via SW which will result in
> a reduction of 100mA. If you lower the voltage to a point that the DVI isn't
> working, I can't guarantee what the current consumption would be. We have
> powered the whole board with DVI running from a USB Lithium pack for I think
> a couple of hours. Removing the SD card will also save power, around 50mA.
>
> Gerald
>

> On Tue, Nov 11, 2008 at 3:59 PM, Sean D'Epagnier <geckosena...@gmail.com>wrote:
>
>
>
> > Is there a risk of damage below 3.8v?  I am tempted to try running a
> > little lower.  Is it best to input the 3.8v at the 5v power jack?  Is
> > there a better place to power the board from to reduce power
> > consumption.
>
> > If I am not using the DVI encoder can I go lower?  Is there a way to
> > disable the DVI encoder to save power other than removing it?
>
> > My plan is to use a buck-boost regulator from single cell lithium
> > batteries and set the output voltage to the lowest possible voltage I
> > can get away with.
>
> > Thanks,
> > Sean
>
> > On Tue, Nov 11, 2008 at 2:44 PM, Gerald Coley <ger...@beagleboard.org>
> > wrote:
> > > You can run from 3.8V but I would not go below that. The onboard
> > regulator
> > > provides power to the DVI encoder, so you need to have enough to keep it
> > in
> > > regulation. One note, the monitor ID may not work as it requires 5V to
> > read
> > > the ID. If you don't need this, it should be OK. Make sure it is well
> > > regulated.
>
> > > Gerald
>

> > > On Tue, Nov 11, 2008 at 3:21 PM, Sean D'Epagnier <geckosena...@gmail.com

>
> > > wrote:
>
> > >> Can't you save even more power by powering from less than 5v?  I heard
> > >> talk of 3.8v.  If you figure out how to do this I want to know so I
> > >> can do it as well.
>
> > >> Sean
>

[Gerald Coley]

3.6 is cutting it a little close, but as long as your 3.3V is stable, that should be fine. You might keep an eye on it just t mkae sure it is OK. The PMIC can actiually go much lower, 2.1V, but my concern is the 3.3V rail.

 

Gerald

 

On Thu, Nov 13, 2008 at 8:48 PM, freedom <koans...@gmail.com> wrote:

We did some quick tests. It seems it's possible to do normal jobs with
3.6 V.
The reason we wanna go below 3.8 V is that there are lots of 3.6 or
3.7 V
batteries. We attached a DC power supply to the power jack and
supplied 3.6 V
to the beagle board. With that, we were able to boot the board all the
way in
to debian+X.

On Nov 12, 8:11 am, "Gerald Coley" <ger...@beagleboard.org> wrote:

> The only place you can put the voltage is via the power jack. 3.8V will not
> cause any damage to the board. Technically you could go lower than that, but
> the DVI would not work. Doen't go below the specification for the
> TWL4030/TPS65950. You can disable the DVI device via SW which will result in
> a reduction of 100mA. If you lower the voltage to a point that the DVI isn't
> working, I can't guarantee what the current consumption would be. We have
> powered the whole board with DVI running from a USB Lithium pack for I think
> a couple of hours. Removing the SD card will also save power, around 50mA.
>
> Gerald
>

> On Tue, Nov 11, 2008 at 3:59 PM, Sean D'Epagnier <geckosena...@gmail.com>wrote:

>
>
>
> > Is there a risk of damage below 3.8v?  I am tempted to try running a
> > little lower.  Is it best to input the 3.8v at the 5v power jack?  Is
> > there a better place to power the board from to reduce power
> > consumption.
>
> > If I am not using the DVI encoder can I go lower?  Is there a way to
> > disable the DVI encoder to save power other than removing it?
>
> > My plan is to use a buck-boost regulator from single cell lithium
> > batteries and set the output voltage to the lowest possible voltage I
> > can get away with.
>
> > Thanks,
> > Sean
>

> > On Tue, Nov 11, 2008 at 2:44 PM, Gerald Coley <ger...@beagleboard.org>

> > wrote:
> > > You can run from 3.8V but I would not go below that. The onboard
> > regulator
> > > provides power to the DVI encoder, so you need to have enough to keep it
> > in
> > > regulation. One note, the monitor ID may not work as it requires 5V to
> > read
> > > the ID. If you don't need this, it should be OK. Make sure it is well
> > > regulated.
>
> > > Gerald
>

> > > On Tue, Nov 11, 2008 at 3:21 PM, Sean D'Epagnier <geckosena...@gmail.com

>
> > > wrote:
>
> > >> Can't you save even more power by powering from less than 5v?  I heard
> > >> talk of 3.8v.  If you figure out how to do this I want to know so I
> > >> can do it as well.
>
> > >> Sean
>

[freedom]

Yes, one of our colleagues keeps using 3.6 V to see
if it's really stable. 2.1 V won't boot. however
we can go as low as 2.5 V after we got X running by
3.6 V.

On Nov 14, 10:58 am, "Gerald Coley" <ger...@beagleboard.org> wrote:
> 3.6 is cutting it a little close, but as long as your 3.3V is stable, that
> should be fine. You might keep an eye on it just t mkae sure it is OK. The
> PMIC can actiually go much lower, 2.1V, but my concern is the 3.3V rail.
>
> Gerald
>

[saladino]

I tried running powertop on my beagleboard with debian 2.6.28. It
works except that I don't get the C-state details. What am I missing?
you seems to be getting that level of detail with your powertop.

Thanks.

On Nov 3 2008, 1:16 pm, Koen Kooi <koen.k...@gmail.com> wrote:
> Hi,
>
> Philip was asking about powerconsumption, so I put a fluke 87 meter in
> between the +5V and beagle. The results after some quick tests:
>
> uboot 2008.1orc2:
> ~290mA at uboot prompt
>
> 2.6.27-omap1:
> ~350mA at X desktop
> ~400mA using 100% cpu at 600MHz
>
> 2.6.27 omap-pm-next:
> ~275mA at X desktop
>

> Powertopreports with with 2.6.27 omap-pm-next:

[Philip Balister]

On Tue, Jan 20, 2009 at 3:12 PM, saladino <s_a_...@tokyo.com> wrote:
>
> I tried running powertop on my beagleboard with debian 2.6.28. It
> works except that I don't get the C-state details. What am I missing?
> you seems to be getting that level of detail with your powertop.

You need to build the -pm kernel from Kevin Hilmans git tree.

There is a recipe linux-omap-pm in OE, and also a version of powertop
patched to recognize all the C-states available on the OMAP3.

Philip

[saladino]

I am new to this. In order that I get this right, could you please
send me the step by step commands to get the pm kernel and patches? I
am using debian on my beagleboard. Will this work with the debian
build?

Thanks,
Saladino.

On Jan 20, 6:03 pm, "Philip Balister" <philip.balis...@gmail.com>
wrote:

[Koen Kooi]

Op 22 jan 2009, om 17:01 heeft saladino het volgende geschreven:

>
> I am new to this. In order that I get this right, could you please
> send me the step by step commands to get the pm kernel and patches? I
> am using debian on my beagleboard. Will this work with the debian
> build?

It won't

regards,

Koen

[Philip Balister]

On Thu, Jan 22, 2009 at 9:01 AM, saladino <s_a_...@tokyo.com> wrote:
>
> I am new to this. In order that I get this right, could you please
> send me the step by step commands to get the pm kernel and patches? I
> am using debian on my beagleboard. Will this work with the debian
> build?

If you use the OE build, change the kernel recipe from linux-omap to
linux-omap-pm in the file conf/machine/beagleboard.conf. I'm not sure
how you would change kernels in debian.

Philip