[time-nuts] Answers to regulator choices comments

[MailLists]

At higher (load) currents, thermal, and parasitic resistance effects are 
getting more prevalent, to the limits of monolithic IC technology.
Why don't consider a more classical approach with external pass 
transistor (a much larger selection available), if a linear regulator is 
really necessary? Also look up the old Unitrode UCx83y family, for 
example...

A mixed bag of IC regulator technologies  were mentioned... the LDOs 
(usually with series p-device for positive voltage ones) are much more 
picky about load conditions, and tougher to stabilize, let alone working 
in parallel configurations.
To equally share the load current, the paralleled regulators should 
tightly track each other over load, and ambient conditions... don't 
consider it to be guaranteed over all operating conditions.

The 3 terminal regulators were designed for convenience, not highest 
performance.

note: LM338 is not a LDO like LT1084 (as TI proudly classifies it)

On 6/30/2013 7:19 AM, Perry Sandeen wrote:
>
>
> Wrote: If you are in the US (maybe elsewhere) you can
> request two free samples.
>
> That might work for projects on and two, but what
> about projects three and onward?
>
>
> Wrote: I suppose that the same philosophy
> [paralleled regulators] would apply to getting more power with a 7805 farm.
>
> It does however with the general output rating of
> 1 to 1.5 amps for each regulator it wouldn’t be very practical.  It’s much cheaper and simpler to use ones in
> the 3 amp plus range.
>
> Wrote: A regulator needs to be specifically
> designed for parallel operation. If it's not designed that way you will have a
> very hard time with it.
>
> I’ve never heard of any three terminal regulator
> designed for parallel operation.  I believe
> that all three terminal regulators use a pass transistor.  When one uses them in parallel they need a
> slight resistance added to each pass transistor to prevent current hogging just
> as one had to do when paralleling power transistors in other high amperage
> circuits.
>
> Wrote: It may be a bit more complicated than
> that.  You need some way to share the
> load and you also need to make sure things are stable.
>
> The TI/National data sheet doesn't show anything
> about paralleling regulators.  The AD
> data sheet shows 2 ft of #18 wire between each regulator and the load.
>
> I'm not enough of an analog guru to reverse
> engineer that setup and figure out the stability constraints and transfer them
> to 78xx type devices.
>
>
> I stand corrected about the LM 1084 showing
> paralleled regulators.  However the data
> sheet says it is pin compatible with the LM 317.  So we get to the paralleled regulator
> circuits by a bit of a circuitous route. If we go to National Semiconductor Linear
> Brief 51 March 1981 titled “Add Kelvin Sensing and
> Parallel Capability To 3-Terminal Regulators” it
> shows how to parallel two or more three terminal LM 338 regulators.
>
> The stability problem is solved for us by the
> Nation Semiconductor engineers. The 2 ft. of #18 wire for each regulator
> provides the load balancing resistance needed.  One could use an ordinary resistor instead if it had the value of 30
> mili-ohms.
>
> The operation of all 3 terminal regulators are the
> same.  The internal circuitry looks at
> the relationship between the output voltage and the *ground* terminal.  As the data sheets show, if we change that
> relationship with resistor combinations we can manipulate the output voltage to
> our needs.
>
> For most low voltage applications one can usually
> find a three terminal regulator that will fit the current needs,
>
> My original point was that the LM 1084 [$14]
> IMNSHO is very expensive for what it does. By paralleling two far cheaper of
> the LM 338 family one gets a larger ampacity of 10 amps instead of 7.5 amps for
> $3 to $5 instead depending on one’s scrounging abilities.
>
> In the end you pays your money and you make your
> choices.
>
> Regards,
>
> Perrier
>
>
>
> ________________________________
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.