Dragon32 Power-Supply

From: Dwight K. Elvey <dwight.elvey_at_amd.com>
Date: Wed Nov 24 17:01:59 2004

>From: "John Honniball" <coredump_at_gifford.co.uk>
>
>Dwight K. Elvey wrote:
>> Although, not particularly good practice, as long as the
>> power dissipation is kept down, this is a design within specifications
>> of the part. They have built in current limit and are suppose
>> to be able to handle continuous loads in current limit.
>> This means that they can be paralleled.
>
>Hmmm... OK, but they still won't be sharing the current equally.
>The one that has the lower current-limit point will be at
>its full capacity, while the other one will have a smaller
>share of the load. At least, that's how I understand it.

 It is true that they won't share. One will run until it
reaches its max and then the other will pick up the rest.
Remember, the biggest issue is the power ( how hot it gets ).
That is related to voltage drop and current, not just current
by it self.

>
>> I once talked with some application engineers at National
>> about this type of operation.
>
>I'm sure I've seen a note somewhere to the effect "do not
>connect in parallel", but of course I've no idea where I
>read that!

 Some regulators do not have a current shut down. They you
shouldn't parallel. Others are intended to run with the
current limit. These can often be paralleled. The ones that
only do thermal shut down shouldn't be paralleled.
 If you know the current load, one can add a
parallel resistor. The current load has to be relatively
constant.
 If the regulators are close, adding some series resistance
can make them share. In some cases, even some additional
circuit board traces may be enough. I've seen memory banks
where there were regulators on either side of the array
and both sides shared relatively well. This is just from
the natural drop of the traces.
Dwight

>
>--
>John Honniball
>coredump_at_gifford.co.uk
>
Received on Wed Nov 24 2004 - 17:01:59 GMT