[ltp] X61 power consumption

[James Knott]

Richard Neill wrote:
> James Knott wrote:
>
>>> Having thought about this a bit more, I reckon the equipment could be
>>> obtained for about $25. 3 cheap digital voltmeters, an op-amp, some
>>> resistors and batteries should be enough. [If someone wants to do the
>>> experiment, I'll write a longer post.]
>>>
>>> That would let one measure the power consumption of a given chip to
>>> within < 10% accuracy, in a calibrated manner (provided that the PCB
>>> traces carrying the current can be acccessed - which isn't necessarily
>>> true for BGA chips).
>>>
>>> Richard
>> Electrical power is the product of voltage and current, plus power
>> factor in AC circuits.  Voltage is very easy to measure, with any
>> multimeter.  On the other hand, current is not.  The traditional method
>> is to open the circuit and insert the meter, but this really isn't
>> practical on circuit boards.  It is possible to measure the magnetic
>> field produced by the current, but then you'd have to do a bit of math
>> to apply it.  Another method is to measure the voltage drop along the
>> lead supplying power to the chip, but that method requires a sensitive
>> voltmeter and you also need to know the resistance of the lead.  Again,
>> it's not always practical with circuit boards.  If you try to use
>> temperature, there are a lot of different factors you need to know, to
>> calculate power dissipation, including cooling etc.  In short, there is
>> no simple way to do what he wants.  Measuring power required by the
>> entire system is easy.  Incidentally, many years ago, I used to repair
>> core memory boards.  On those boards you have to be able to measure the
>> current in the wires that go through the cores.  On those boards, there
>> were wire loops which could be used with an appropriate oscilloscope
>> probe that could measure AC, but not DC currents.  The loops were
>> necessary, as the probe had to close around the wire.
>>
>>
>
>
> Dear James,
>
> I had actually thought about how to overcome this.
>
> i) Use 3 separate DVMs and battery supplies. This keeps all the
> voltages floating, and makes it much harder to generate shorts.
>
> ii)One DVM goes always across the supply leads to the chip, to measure
> the voltage.
>
> iii)The second DVM measures the microvolt drop across one of the
> supply leads. We do this using an instrumentation amplifier
> (essentially an op-amp optimised for very tiny offsets) in order to
> add 1000 x voltage gain to the DVM. Once the probes are clipped (or
> soldered very carefully, using kynar wire) in place, don't move them
> till the end of the test.
>
> iv)With the  laptop off, and using the probes from (iii) still clipped
> in place and unmoved, add two more probes that are slightly further
> apart on the track. This forms a 4-wire probe for resistance
> measurement. Introduce a current(*) of about 10 mA down that track,
> measuring its exact value with the 3rd DVM.  This gives you the
> track's resistance.
>
> (*)For safety, we need a current-limited, voltage-limited supply. Eg
> connect a single AA-cell to a chain of 10 x 15 Ohm resistors. Then,
> take your "current source" from across one of the resistors. Thus,
> 0.1V at 10mA maximum.
>
> Cost: about $5 for each of 3 x DVMs,  + some kynar wire + an
> instrumentation amp (about $5) + a few resistors.
>
> HTH,
>
> Richard
>
I had already mentioned measuring voltage drop along the leads. 
However, as you've shown above, it's not simple, particularly when you
consider there may be multiple power connections to one IC, not to
mention leakage current among the various devices etc.


-- 
Use OpenOffice.org <http://www.openoffice.org>