> Well, the iPhone runs on battery power, so yeah, the power supply is 
> probably a little unstable.

Not just that, but from a single voltage (the battery) several very stable 
voltages must be generated.  This involves lots of analogue components (eg 
how do you efficiently make accurate 3.3V, 5V and 8V lines from a LiPo 
battery that can be between 3 and 4.5V?).

> I'm thinking more about desktop PC motherboards. Why do *they* need so 
> many million capacitors?

Noise reduction, basically they allow everything to run correctly at much 
higher frequencies than would be possible without them.  It works two ways, 
the capacitors around each IC help to keep a smooth voltage level for that 
IC, ie shielding it from any noise on the power supply lines.  Also the 
capacitor then prevents a lot of noise being put back onto the supply line 
from the ICs.

There are also a lot of resistors on the signal lines, these in conjunction 
with the tiny parasitic capacitances cause a low-pass filter effect on the 
signals.  This also helps to reduce noise emission from the lines.

Post a reply to this message

On 25/10/2010 01:02 PM, scott wrote:
>> Well, the iPhone runs on battery power, so yeah, the power supply is
>> probably a little unstable.
>
> Not just that, but from a single voltage (the battery) several very
> stable voltages must be generated. This involves lots of analogue
> components (eg how do you efficiently make accurate 3.3V, 5V and 8V
> lines from a LiPo battery that can be between 3 and 4.5V?).

1. Why do you need several different voltages?

2. I'm not aware of any way to change the voltage of a DC circuit. You 
can use a resistor to limit current, but AFAIK there's no way to 
actually change voltages.

>> I'm thinking more about desktop PC motherboards. Why do *they* need so
>> many million capacitors?
>
> Noise reduction, basically they allow everything to run correctly at
> much higher frequencies than would be possible without them. It works
> two ways, the capacitors around each IC help to keep a smooth voltage
> level for that IC, ie shielding it from any noise on the power supply
> lines. Also the capacitor then prevents a lot of noise being put back
> onto the supply line from the ICs.
>
> There are also a lot of resistors on the signal lines, these in
> conjunction with the tiny parasitic capacitances cause a low-pass filter
> effect on the signals. This also helps to reduce noise emission from the
> lines.

Mmm, interesting. (And ridiculously complicated.) I wonder how much of 
this stuff is to do with electromagnetic radiation? (You don't want EM 
interference stopping your motherboard working, and you don't want the 
board to radiate EM either...)

Post a reply to this message

> 1. Why do you need several different voltages?

Different circuits need different voltages.  It's why the connector you plug 
from the PSU to your motherboard has 20-odd wires and not just 2 big fat 
ones.  In an iPhone I imagine you'd at least need different voltages for the 
digital circuits, the audio circuits, the display and the RF ciruits.

> 2. I'm not aware of any way to change the voltage of a DC circuit.

This is often used for battery powered devices as it's very efficient:

http://en.wikipedia.org/wiki/Charge_pump

> Mmm, interesting. (And ridiculously complicated.) I wonder how much of 
> this stuff is to do with electromagnetic radiation? (You don't want EM 
> interference stopping your motherboard working, and you don't want the 
> board to radiate EM either...)

I assume they must pass certain legal test standards for EM compatibility, 
but there are products that require stricter controls, especially ones that 
are used near other RF equipment where interference is unwanted (eg medical 
equipment, circuits in radio receivers, on planes, near to any 
safety-critical device etc).

Post a reply to this message

> 2. I'm not aware of any way to change the voltage of a DC circuit. 

I made an example using that Java circuit simulation applet:

http://tinyurl.com/2cplrbn

Post a reply to this message

On 25/10/2010 02:46 PM, scott wrote:
>> 2. I'm not aware of any way to change the voltage of a DC circuit.
>
> I made an example using that Java circuit simulation applet:
>
> http://tinyurl.com/2cplrbn

Hmm, interesting.

1. It's possible to edit the circuits. I never knew that.

2. It's possible to encode the circuit description in the URL. (I'm 
pretty sure you'll quickly hit a limit on the maximum size for a URL!)

3. No matter what values I change, the circuit always stabilises to 
exactly 8V.

Post a reply to this message

scott wrote:
> http://en.wikipedia.org/wiki/Charge_pump

That's freaky.

> I assume they must pass certain legal test standards for EM 
> compatibility, but there are products that require stricter controls, 

Like, a radio transmitter that's designed to run for hours or days on a 
small battery? :-)

-- 
Darren New, San Diego CA, USA (PST)
   Serving Suggestion:
     "Don't serve this any more. It's awful."

Post a reply to this message

On 10/25/2010 7:17 AM, Invisible wrote:

> 1. Why do you need several different voltages?

Different components require different voltages. Sometimes you get lucky 
and everything will run all one a single voltage. You may have high 
intensity LED's that have a voltage forward of more than 3.3 volts, so 5 
volts is a logical choice to power the backlight circuitry allowing the 
LED's to function, but pumping 5 volts into a chip requiring 3.3 volts, 
would possibly overheat at best or fry at worse, etc...

>
> 2. I'm not aware of any way to change the voltage of a DC circuit. You
> can use a resistor to limit current, but AFAIK there's no way to
> actually change voltages.
>

You're kidding, right?

You can affect voltage by either using a regulator (true, it does this 
by altering current) or if you need more voltage than is supplied by the 
source you can use a charge pump, which will increase voltage by 
alternately charging and discharging a bank of capacitors configured in 
a very specific way (lots of caps for that one)

>
> Mmm, interesting. (And ridiculously complicated.) I wonder how much of
> this stuff is to do with electromagnetic radiation? (You don't want EM
> interference stopping your motherboard working, and you don't want the
> board to radiate EM either...)

Filtering probably helps that. But at those timings every little thing 
in the design becomes critically important. You can literally have a 
parasitic capacitance completely filter out the signal.

-- 
~Mike

Post a reply to this message

On 10/25/2010 8:46 AM, scott wrote:
>> 2. I'm not aware of any way to change the voltage of a DC circuit.
>
> I made an example using that Java circuit simulation applet:
>
> http://tinyurl.com/2cplrbn
>

another simple charge pump based on a 555 timer.

http://tinyurl.com/2a8wf2x

-- 
~Mike

Post a reply to this message

>> I made an example using that Java circuit simulation applet:
>>
>> http://tinyurl.com/2cplrbn
>>
>
> another simple charge pump based on a 555 timer.
>
> http://tinyurl.com/2a8wf2x

Cool!  I know also that you can use a circuit with inductors rather than 
capacitors to make it more efficient, but the downside is the inductors need 
more physical space than the capacitors on the board (only a problem if 
space is very limited).

Post a reply to this message

On 10/26/2010 3:01 AM, scott wrote:
>>> I made an example using that Java circuit simulation applet:
>>>
>>> http://tinyurl.com/2cplrbn
>>>
>>
>> another simple charge pump based on a 555 timer.
>>
>> http://tinyurl.com/2a8wf2x
>
> Cool! I know also that you can use a circuit with inductors rather than
> capacitors to make it more efficient, but the downside is the inductors
> need more physical space than the capacitors on the board (only a
> problem if space is very limited).
>
>
>

I have yet to attempt a boost converter using an inductor. I should 
attempt such a thing one of these days. ;) From what I understand, you 
can get some pretty nasty high voltages with such a circuit. Probably 
one of the reasons I haven't messed with it.

There's a regulator circuit geared toward LED use that, iirc used a very 
small capacity that used the same theory to supply enough voltage to 
overcome the high output LED's VF from a very small battery, even when 
the battery is nearly spent and has a voltage drop. Something called a 
joule thief. Essentially a boost converter using a transformer.

-- 
~Mike

Post a reply to this message